By: Jenny Ahn, Till Daldrup, Blake Spendley

Editor: Sam Koppelman

Based on Hunterbrook Media’s reporting, Hunterbrook Capital is short $CSG.AS and long a basket of comparable securities at the time of publication. Positions may change at any time. This article is not investment advice or a recommendation to buy, sell, or hold any security. See our website for full disclosures.

• Czechoslovak Group (CSG.AS) has recently announced billions of dollars in contracts with unnamed “Southeast Asian” customers. It appears most of these transactions may actually be with just one counterparty: Indonesia, which CSG used to name explicitly in its filings. CSG’s announced deals tied to this customer since the IPO could amount to up to $2.8 billion, nearly a fifth of the size of CSG’s entire order backlog as of last year, suggesting CSG’s order book may be less diverse than it appears.

• Virtually all of CSG’s business in the country appears to flow through a single intermediary, Republikorp, a private Indonesian arms dealer. CSG has no disclosed manufacturing subsidiaries, offices, or employees in Indonesia. Republikorp’s founder, Norman Joesoef, is the son of Indonesia’s former ambassador to Slovakia and was initially named the CEO of Czechoslovak Group Indonesia in 2018, though references to that entity name have since disappeared.

• CSG’s history in Indonesia has been marred by controversy, including allegations of bribery and price gouging. In 2023, CSG’s Excalibur Army brokered a $792 million deal for the transfer of 12 used Qatari fighter jets to the Indonesian defense ministry that triggered a public outcry. An Indonesian outlet accused the overpriced deal of including a multi-million dollar kickback to then-defense minister and now-president Prabowo Subianto. Prabowo’s campaign team vigorously denied any kickback but ultimately canceled the deal after public scrutiny.

• Hunterbrook also found a paper trail that points to a potential multimillion-euro payment to an intermediary linked to CSG’s business in Indonesia. CSG’s subsidiary filings show a Singapore-based consultant, Irwan Bin Omar — who has operated under the Republikorp brand and described himself as a bridge between EU entrepreneurs and Indonesia — paid 410,000 euros toward a 24% stake in a CSG subsidiary in 2021. In its IPO prospectus, CSG discloses its re-purchase of the subsidiary as an “acquisition” worth 24.4 million euros, requiring a net cash expense of 17.9 million euros. Asked if CSG paid Bin Omar the 17.9 million euros, CSG did not respond to Hunterbrook’s repeated email and phone requests for comment, after previously engaging on Hunterbrook’s prior reporting about CSG’s production rates and IPO disclosures.

It is a weird transaction.

Buried on page 374 of the European defense conglomerate Czechoslovak Group’s 728-page IPO prospectus is a table describing an acquisition of a Slovak subsidiary called Vývoj Martin.

No counterparty is named.

The deal is relatively small — 24.4 million euros — the kind of purchase that’s easy to gloss over when analyzing the prospectus of what was a $30 billion company at the time of the IPO.

But Hunterbrook — whose deep-dive investigation of CSG published earlier in May sparked international coverage and a press tour from the defense conglomerate’s CEO1 — took a closer look.

What we found was odd. CSG already owned most of the subsidiary, Vývoj Martin, and seemed to be just buying back a minority share it sold earlier.2 The party CSG appeared to be buying out — a Singapore-based entity called RDBSG — had no known ties with CSG or any European defense companies, as far as Hunterbrook could tell. According to its Singapore registry filings, RDBSG is described as providing “human resource consultancy services.” It has no website and no known representatives other than its sole proprietor, a Singaporean national named Irwan Bin Omar.

Its address, as listed on a Slovak government register, matches a coworking space in an eastern commercial district in Singapore.

In 2021, RDBSG paid CSG a total of 410,000 euros for its stake in Vývoj Martin, Hunterbrook’s review of statutory filings suggests.

Three years later, CSG appeared to have spent about 18 million euros to regain those shares, the prospectus shows.

It’s unclear from the scant disclosures how exactly the cash flowed — and whether the entire 18 million euros in cash went to the shareholder, RDBSG, which would amount to an over 4,000% return on its 410,000-euro investment. But records reviewed by Hunterbrook suggested no other counterparties obviously involved in the transaction.

And CSG’s accounting treatment of the transaction suggests the 18 million euros is real cash that physically left CSG with the repurchase, according to JP Krahel, chair of Loyola University Maryland’s accounting department, whom Hunterbrook asked for an independent expert opinion on the puzzle.

As Hunterbrook dug deeper, our findings showed this transaction may be linked to billions of dollars in contracts CSG had recently announced with unnamed customers in Southeast Asia, where the Czech arms dealer apparently has no employees.

Hunterbrook found, based on our research of defense publications, social media, and CSG’s own past disclosures, that the unnamed Southeast Asian customers may all tie back to just one country — Indonesia — with the deals channeled through a single intermediary, an Indonesian arms dealer called Republikorp.

Those deals would add up to a number roughly equivalent to a fifth of CSG’s total 15 billion euro backlog as of 2025. But more than half of that backlog is related to ammunition sales, contingent on meaningfully expanding production — which may prove to be more difficult than CSG has described to investors, as Hunterbrook found in our previous investigation.

That means the Indonesia sales could amount to nearly half the non-ammunition-related backlog reported by CSG.

And Republikorp, the single conduit of that massive sales channel, seems to be linked with RDBSG, the mysterious counterparty that may have earned a big payday when CSG bought back Vývoj Martin.

NOTE: CSG did not respond to repeated requests for comment — and its filings are (perhaps intentionally!) very ambiguous. So it’s unclear what, exactly, happened here. There could be another, better explanation. If you have one, email ideas@hntrbrk.com.

CSG’s Decade-Long Ties to a Mystery Customer Behind Billions in Orders

In February, 13 days after the IPO, CSG announced a $300 million armored vehicle deal — “the largest export order” of the platform to date. Two months later, another press release announced a $2.5 billion air-defense mega-package. Together, the deals represent nearly a fifth of the size of the total backlog the company reported for the past year.

In each case, CSG identified the buyers only as customers in “Southeast Asia.”

CSG’s April announcement of the $2.5 billion deal also included a list of past deliveries to the same “region” — worded in a way that might lead a reader to assume CSG has a diversity of customers across Southeast Asia.

But CSG was not always so coy. In late 2022, CSG’s subsidiary Excalibur International explicitly named the buyer for a medium-range air-defense system — very likely the same MRAD delivery referenced in the April announcement3 — as Indonesia. And in 2024, Indonesian press confirmed Republikorp was also cooperating with CSG’s majority-owned subsidiary Fiocchi Munizioni on ammunition production — consistent with the small-caliber ammunition contract “in the same region” that CSG referenced in its April announcement.

The 2022 deal — which also included a ballistic missile system and had a price tag of more than 500 million euros — came on the heels of another key deal between CSG subsidiary Excalibur Army and Republikorp for the purchase of 122 rocket launchers on Tatra chassis, potentially worth tens of millions of dollars.4

A screenshot of an alleged 2023 Indonesian foreign ministry document posted on a Facebook group for Indonesian military enthusiasts summarizes other strategic agreements between Republik and CSG subsidiary Excalibur International, facilitated by the Indonesian government.

No other Southeast Asian country appears to have had a similar level of cooperation with CSG or institutional infrastructure that could serve a similarly scaled deal, according to Hunterbrook’s review. The only other deal in the region Hunterbrook could identify was what appears to be a 2022–2023 subcontract for 10 armored vehicles supplied to the Philippines through Israel’s Elbit Systems.

But that contract doesn’t seem to have been material enough for CSG to break out the Philippines in the revenue section of its annual disclosures. In CSG’s 2023 and 2024 annual reports — before the company switched to the less granular region-based reporting post-IPO — Indonesia appears to have been the only Southeast Asian country it felt compelled to identify specifically.

The recent $2.5 billion mystery deal appears to be a continuation of the same 2022 air defense system deal.

According to CSG’s press release, the recent contract includes the delivery of complete batteries of multi-layer air defense systems and will utilize Tatra chassis. That’s pretty much how the architecture of the 2022 missile defense systems was described as: Turkish-made missile launchers mounted on Tatra chassis.

Deliveries of a Turkish system mounted on a Tatra chassis began in 2025; a regional defense outlet reported that military enthusiasts spotted a tactical ballistic missile system at an Indonesian army base in Borneo, mounted atop 8×8 Tatra trucks.

At a minimum, those who follow Indonesian military news appear convinced CSG’s mystery customer is Indonesia. In response to a recent article on the $2.5 billion CSG deal published by Indonesian defense news outlet Airspace Review — which did not mention Indonesia as a potential candidate for the unnamed Southeast Asian customer — 300 commenters largely expressed their belief that the mystery customer is Indonesia.

CSG’s Indonesia Deals Marred by Local Controversy

The Indonesian government — along with Republik and CSG — may have its own reason not to publicize the multibillion-dollar deal, given the recent controversy that had previously linked the three names.

In 2023, Excalibur International brokered a $792 million deal to transfer 12 used fighter jets from the Qatari Air Force to the Indonesian defense ministry. Indonesian investigative outlet Ceri reported that Republikorp served as an intermediary for the deal, citing rumors among Indonesian defense ministry officials. The deal, translating to $66 million per jet, triggered an outcry among the Indonesian public, who accused the government of overpaying for old jets that Qatar had once offered to Indonesia for free.

A few months later, a report alleged the deal included a 7% kickback — worth about $55.4 million — to then defense minister Prabowo Subianto, including a $20 million upfront payment.5 Prabowo’s campaign team — he was running for the 2024 presidential election — immediately denied the allegation, describing it as fake news. Indonesia eventually scrapped the deal, citing budget limitations.

The Ambassador’s Son

Across CSG’s deals in Indonesia, virtually everything appears to flow through a single private intermediary: Republikorp.

Republikorp’s founder and chairman is a man named Norman Joesoef.

Norman Joesoef appears to be the son of Harsha Edwana Joesoef, a prominent Indonesian businessman and former diplomat. The elder Joesoef is the founder of the RPX Group, Indonesia’s FedEx licensee — and, from 2009 to at least 2012, he served as Indonesia’s ambassador to Slovakia. That posting placed the Joesoef family squarely in the Czech-Slovak diplomatic orbit.

A 2022 social media post celebrating Norman Joesoef’s daughter cites Harsha Joesoef as her grandfather. A 2019 social media post also suggests Harsha Joesoef is Norman Joesoef’s father. Republikorp’s address on its website is the same as the Jakarta headquarters of the Joesoef family’s logistics company RPX in Jakarta.

Norman Joesoef founded Republikorp in 2013 after returning from his studies in Prague, where he was allegedly inspired by a college colleague working in a European defense conglomerate. Joesoef’s school years in Prague likely overlapped with CSG CEO Michal Strnad’s budding career at his father’s company starting in 2010, at around age 17.

Hunterbrook couldn’t confirm if the two actually knew each other in Prague, but what is clear is that the two men have since been doing business together at an enormous scale — and the corporate paper trail connecting Republik to CSG is unambiguous.

One prominent example: In 2018, Strnad pledged to invest $100 million in a new industrial park in Indonesia, to be headed by Norman Joesoef as the CEO of Czechoslovak Group Indonesia, according to Indonesian Antara News.

The reference to Czechoslovak Group Indonesia appears to have disappeared since then, however, with Republikorp — helmed by the same Joesoef — seemingly emerging as CSG’s Indonesian counterpart, creating at least the appearance of an independent entity.

Hunterbrook repeatedly attempted to ask CSG about all of these relationships — also reaching out to Republik, Norman Joesoef, and Bin Omar. Nobody responded to repeated requests for comment.

But one answer may live in that mysterious payment.

A Mystery Figure Tied to the 24.5 Million Euro “Reacquisition” of a Subsidiary

In December 2021, CSG transferred a 24% stake in a Slovak subsidiary, Vývoj Martin, to a Singapore-registered entity called RDBSG, whose sole proprietor is a Singapore national named Irwan Bin Omar. The agreed-upon price was 5.85 million euros, according to a filing by Vývoj Martin’s parent company, MSM Group, with the Slovak commercial registry.

Bin Omar paid 410,000 euros up front. He never paid the rest. The share transfer deal got canceled.

MSM Group’s 2023 annual report says the remaining purchase price “was supposed to be paid, but this did not happen.6” The parties rescinded the deal, with the reverse transfer of shares expected to occur by the end of 2024 and the return of the deposit on the purchase price by no later than mid-2025, MSM Group’s filings show.

That reversal was recorded in CSG’s 2024 annual statement, and on its 2026 IPO prospectus.7

But the way CSG booked this transaction was interesting.

The payment for the reacquisition of those shares was listed as 24.4 million euros — suggesting an over 300% premium on the 5.85 million euros selling price for the shares in 2021.

The magnitude of that valuation increase is also seemingly contradicted by CSG’s own prior disclosures that put the carrying value of its retained majority shares in the subsidiary as having increased far more modestly — at just 16% between 2021 and 2024.

But it gets even weirder. CSG also reported the value of its pre-existing shares in Vývoj Martin as zero — contradicting CSG’s own disclosure, as mentioned above, that its existing shares in Vývoj Martin were worth roughly 18 million euros as of 2023.

The rules of International Financial Reporting Standard 3 (IFRS 3) — the accounting method CSG used for the business combination — are quite clear, as Krahel confirmed to Hunterbrook. “When a company consolidates an entity which it had partially owned, it has to disclose the fair market value of its ownership stake immediately before the acquisition date,” Krahel explained. This disclosure is supposed to go on the “Investment value as of the control acquisition” line of the IFRS table – which, in CSG’s case, was zero.

The implication of this accounting choice — besides being inconsistent with IFRS — could be that the entirety of CSG’s carry value of its existing shares was impaired to nothing, without explanation. Alternatively, CSG may have just wanted to present the acquisition as a fresh one, with the 24.4-million-euro price tag representing the value of the entire entity, and with CSG’s existing interest in the entity buried somewhere in that overall figure.8 That would, potentially, obscure the exact amount of any payment that went to Bin Omar.

Either way, that 24.4 million euro price tag seems to have led to a cash expense from CSG of 17.9 million (24.4 million net of about 6.5 million cash that was held by Vývoj Martin), as recorded in CSG’s prospectus.

The bottom line is, according to the accounting, CSG seems to have spent about 18 million euros as part of a deal to buy back shares it had sold for just 410,000 euros three years prior.

To whom did the 18 million euros go? We don’t know.

But we found no other counterparties aside from Bin Omar related to the reacquisition disclosed by CSG or its subsidiaries.

CSG did not respond to Hunterbrook’s questions about any of it.

But one potential clue may come from the identities of RDBSG and its sole owner, Irwan Bin Omar.

Among Bin Omar’s extensive list of corporate entities he owns is one called Republikorp Project Management Services Pte. Ltd., since renamed the more nondescript Research and Development Innovations.

Hunterbrook could find few details about Bin Omar, but the Republikorp name may be no coincidence. It matches the name of Norman Joesoef’s company, CSG’s counterparty in Indonesia.

On his own archived website for Republikorp, Bin Omar describes his work as providing foreign companies that “seek to sell their advanced products and services to Indonesia” with “solutions to the administration processes” in the Southeast Asian country. He says he is well-versed in the political context of Indonesia and ASEAN. He studied political science on a joint Singapore-Indonesia scholarship at Gadjah Mada University in Indonesia.

One promotional article appearing on a Chinese outlet in 2017 also described a man with the same name as a member of the Indonesian presidential staff, though Hunterbrook could not independently verify the claim.

After CSG repeatedly ignored emails about Bin Omar — and Republikorp — a Hunterbrook reporter called the company’s spokesperson. “Please communicate in written form,” the spokesperson said.

When the reporter pointed out Hunterbrook’s emails had gone unanswered, and asked for clarity, the spokesperson hung up.

Authors

Jenny Ahn joined Hunterbrook after serving many years as a senior analyst in the US government. She is a seasoned geopolitical expert with a particular focus on the Asia-Pacific and has diverse overseas experience. She has an M.A. in International Affairs from Yale and a B.S. in International Relations from Stanford. Jenny is based in Virginia.

Till Daldrup is an investigative journalist who joined Hunterbrook from The Wall Street Journal, where he focused on open-source investigations and content verification. In 2023, he was part of a team of reporters who won a Gerald Loeb Award for an investigation that revealed how Russia is stealing grain from occupied parts of Ukraine. He has an M.A. in Journalism from New York University and a B.S. in Social Sciences from University of Cologne. He’s also an alum of the Cologne School of Journalism (Kölner Journalistenschule).

Blake Spendley joined Hunterbrook from the Center for Naval Analyses (CNA), where he led investigations as a Research Specialist for the Marine Corps and US Navy. He built and owns the leading open-source intelligence (OSINT) account on X/Twitter, called @OSINTTechnical (over 1 million followers), which also distributes Hunterbrook Media reporting. His OSINT research has been published in Bloomberg, the Wall Street Journal, and The Economist, among other top business outlets. He has a B.A. in Political Science from USC.

Editor

Sam Koppelman is a New York Times best-selling author who has written books with former United States Attorney General Eric Holder and former United States Acting Solicitor General Neal Katyal. He helped build Fenway Strategies into one of the preeminent strategic communications firms in the country. Sam has published in the New York Times, Washington Post, Boston Globe, Time Magazine, and other outlets — and occasionally volunteers on a fire speech for a good cause. He has a BA in Government from Harvard, where he was named a John Harvard Scholar and wrote op-eds like “Shut Down Harvard Football,” which he tells us were great for his social life.

Graphic

Dan DeLorenzo is a creative director with 25 years reporting news through visuals. Since first joining a newsroom graphics department in 2001, he has built teams at Bloomberg News, Bridgewater Associates, and the United Nations, and published groundbreaking visual journalism at The Wall Street Journal, Associated Press, The New York Times, and Business Insider. A passion for the craft has landed him at the helm of newsroom teams, on the ground in humanitarian emergencies, and at the epicenter of the world’s largest hedge fund. He runs DGFX Studio, a creative agency serving top organizations in media, finance, and civil society with data visualization, cartography, and strategic visual intelligence. He moonlights as a professional sailor working toward a USCG captain’s license and is a certified Pilates instructor.

Hunterbrook Media publishes investigative and global reporting — with no ads or paywalls. When articles do not include Material Non-Public Information (MNPI), or “insider info,” they may be provided to our affiliate Hunterbrook Capital, an investment firm which may take financial positions based on our reporting. Subscribe here. Learn more here.

Please contact ideas@hntrbrk.com to share ideas, talent@hntrbrk.com for work opportunities, and press@hntrbrk.com for media inquiries.

LEGAL DISCLAIMER

© 2026 by Hunterbrook Media LLC. When using this website, you acknowledge and accept that such usage is solely at your own discretion and risk. Hunterbrook Media LLC, along with any associated entities, shall not be held responsible for any direct or indirect damages resulting from the use of information provided in any Hunterbrook publications. It is crucial for you to conduct your own research and seek advice from qualified financial, legal, and tax professionals before making any investment decisions based on information obtained from Hunterbrook Media LLC. The content provided by Hunterbrook Media LLC does not constitute an offer to sell, nor a solicitation of an offer to purchase any securities. Furthermore, no securities shall be offered or sold in any jurisdiction where such activities would be contrary to the local securities laws.

Hunterbrook Media LLC is not a registered investment advisor in the United States or any other jurisdiction. We strive to ensure the accuracy and reliability of the information provided, drawing on sources believed to be trustworthy. Nevertheless, this information is provided “as is” without any guarantee of accuracy, timeliness, completeness, or usefulness for any particular purpose. Hunterbrook Media LLC does not guarantee the results obtained from the use of this information. All information presented are opinions based on our analyses and are subject to change without notice, and there is no commitment from Hunterbrook Media LLC to revise or update any information or opinions contained in any report or publication contained on this website. The above content, including all information and opinions presented, is intended solely for educational and information purposes only. Hunterbrook Media LLC authorizes the redistribution of these materials, in whole or in part, provided that such redistribution is for non-commercial, informational purposes only. Redistribution must include this notice and must not alter the materials. Any commercial use, alteration, or other forms of misuse of these materials are strictly prohibited without the express written approval of Hunterbrook Media LLC. Unauthorized use, alteration, or misuse of these materials may result in legal action to enforce our rights, including but not limited to seeking injunctive relief, damages, and any other remedies available under the law.

By: Till Daldrup, Evan Comen

Editor: Sam Koppelman

Hunterbrook Media’s investment affiliate, Hunterbrook Capital, does not have any positions related to this article at the time of publication. Positions may change at any time. Full disclosures on our website.

This is the latest investigation in Hunterbrook Media’s investigative series on the poisoning of American communities. For decades, polluters have sickened towns and cities across the U.S. by releasing toxic chemicals. Hunterbrook has built a nationwide database to expose harm to communities and ecosystems, then hold accountable those responsible. After reporting from Illinois, Nebraska, California, Texas, Louisiana, New Mexico, and Indiana, this investigation takes us to Virginia. Hunterbrook Media has also investigated environmental damage and harm to communities due to industrial activity spanning from Namibia to Argentina to Peru. Please email ideas@hntrbrk.com if you suspect pollution or poisoning in your community.

Separately, ICYMI: Hunterbrook Media posted a thread on X recapping our work over the past three months — from an investigation of a European defense giant to a feature on a hidden photonics supplier in Taiwan. Repeatedly, our reporting has been validated, including this month in a big way. Read more. And stay tuned for our next article, which we plan to publish… tomorrow.

When Louisa Alice Dickison found out last year that the air her daughter is breathing may cause cancer, she felt she could not stay silent.

“I had no choice, ethically and morally, as a mom in this community, it felt imperative for me to put myself out there just to tell people what was going on and give them the facts,” Dickison told Hunterbrook Media.

Dickison recently turned her Instagram profile public to start posting videos and stories about the cancer risk that ethylene oxide (EtO) emissions from medical sterilizers and warehouses handling medical equipment could pose to people living in Richmond, Virginia., and its suburbs.

She lives in the vicinity of two medical sterilization facilities: The Central Virginia Health Network/Bon Secours Mercy Health sterilizer in Northwestern Richmond and Sterilization Services of Virginia’s (SSV) facility in Henrico.

Dickison started talking to community members and learned most of them were unaware that these facilities were located near their homes. “Virtually nobody knows,” she said.

In late March, she attended a town hall meeting organized by the environmental nonprofits Southern Environmental Law Center and Virginia Interfaith Power & Light that aimed to raise awareness about ethylene oxide and Biden-era emissions rules for the chemical that the EPA is attempting to roll back.

Two weeks after this meeting, Sterilization Services of Virginia’s facility in Henrico accidentally released more than 500 pounds of ethylene oxide, according to records from the Virginia Department of Environmental Quality (DEQ). Early on April 9, a faulty check valve released 580 pounds of liquid EtO onto the roof of the facility. As temperatures rose to 50 degrees Fahrenheit, the chemical evaporated and “dispersed into the atmosphere,” according to a DEQ air inspection report.

The release at SSV likely exposed more than 260,000 people living nearby to elevated levels of ethylene oxide, a Hunterbrook analysis shows.

A plume model that Hunterbrook generated using the National Oceanic and Atmospheric Administration’s (NOAA) HYSPLIT tool shows that the evaporated ethylene oxide soon reached residential neighborhoods about a mile from the facility. The tool takes into account local weather conditions at the time of the release.

Dickison said she has not been notified by authorities about this release and first learned about it when Hunterbrook asked her about it. Her nine-year-old daughter was with her father that day and would have been right near the sterilization facility.

“I’m horrified,” Dickison said.

The DEQ issued a notice of violation on April 24, stating that the released EtO amount goes beyond what the facility is allowed to emit in an entire year, according to its permit. The notice says SSV needs to show regulators how it aims to “correct the cause of malfunction and to prevent its recurrence.”

A DEQ spokesperson told Hunterbrook that SSV responded to the notice of violation and submitted a corrective action plan, which includes the installation of an EtO release alarm system and new pressure relief check valves, as well as regular check valve testing and replacement.

Sterilization Services of Virginia did not respond to repeated requests for comment.

Data Shows Elevated Cancer Risk Around Sterilizer

SSV uses ethylene oxide as a sterilizing agent for medical devices. The chemical plays a crucial role in our healthcare system, but it’s also a ruthless carcinogen. In 2016, the EPA in a revised toxicity assessment concluded it is 30 times more carcinogenic to adults — and 60 times more so to children — than previously thought.

Ethylene oxide is linked to blood cancers like leukemia, non-Hodgkin lymphoma, and multiple myeloma, as well as breast cancer. In 2024, under the Biden administration, the EPA issued stricter rules that require commercial medical sterilizers and large chemical plants to reduce their ethylene oxide emissions. The EPA now aims to roll those regulations back.

Henrico County — which includes SSV’s facility and its immediate surroundings — has an overall incidence rate across all cancer types that’s just slightly above the state average, according to National Cancer Institute data for 2018 to 2022.

But the picture changes drastically when the data is filtered for cancers that are associated with ethylene oxide exposure and population groups that actually live in the vicinity of SSV’s medical sterilization site. Henrico County snakes around the state capital, Richmond, in the shape of a dumbbell. While the western part is mostly white and Asian, the eastern part — where SSV is located — is primarily Black.

The county has the third-highest leukemia incidence among Black residents in the entire state. It also ranks seventh for non-Hodgkin lymphoma and ninth for early breast cancer cases among Black patients. That’s out of 95 counties in Virginia.

Hunterbrook’s plume model analysis indicates that Black residents are indeed more impacted by emissions from SSV’s facility than members of other racial groups: 71% of people who live in census blocks where modeled ethylene oxide concentrations reached or exceeded 1 microgram per cubic meter for at least 30 minutes during the April 9 release are Black.

“The evidence is undeniable that ethylene oxide is a chemical that absolutely increases the health risks to people who are breathing this in on a consistent basis. The fact that it’s in a predominantly Black community cannot be ignored,” said Monica Hutchinson, president of the Henrico County NAACP. “The [Trump] administration has proven to be extremely hostile to anything to help safeguard or protect the Black community and other vulnerable communities.”

SSV’s facility is not the only polluter in the area. Richmond International Airport is nearby, as well as a large landfill. But cancer data and the county’s population distribution suggest that incidence rates of cancers associated with ethylene oxide exposure are elevated among people who live near the SSV building.

Science would suggest that people in close proximity to a facility that’s emitting thousands of pounds of ethylene oxide per year have an elevated cancer risk, Joe Goffman, a former assistant administrator in the EPA’s Office of Air and Radiation, told Hunterbrook.

“Ethylene oxide has always been known to be a carcinogen that presents a level of risk that the Clean Air Act recognizes and deems it appropriate for regulation,” he said.

Trump Issued Exemptions From Stricter Rules

Sterilization Services of Virginia is one of 40 medical sterilization facilities that in July 2025 received a two-year presidential exemption from Biden-era ethylene oxide rules that would have imposed new emissions-control requirements on commercial sterilizers.

The Virginia DEQ told the Henrico Citizen, a local news organization, that, notwithstanding the exemption, the Sterilization Services of Virginia facility is already meeting the requirements under those new regulations.

“In 2025, SSV installed additional control equipment to further reduce their emissions of EtO,” the DEQ said in a written statement to the Citizen. “They began operating this new equipment in October of 2025. With the addition of this equipment, SSV is now equipped to meet 2028 EtO emission standards [two years early].”

The statement appears to refer to a dry bed reactor, which breaks down trace ethylene oxide, that SSV started operating that month. But just six months prior, when the company requested the presidential exemption, SSV said in reference to the new ethylene oxide rules that “the technology necessary to implement the standards is not available,” according to a copy of an email obtained by Hunterbrook.

Emissions data that SSV submitted to the EPA’s toxics release inventory shows that while the company has reduced ethylene oxide emissions from its stacks, fugitive releases have increased rapidly, and the facility’s total emissions in 2024 — the latest available data — were higher than they were in 2020.

Fugitive emissions are the industry term for leaks that escape through valves, cracked seals, and loose connectors rather than being channeled through controlled smokestacks or vents.

The latest emissions data for 2024 also shows that SSV was the fifth-largest ethylene oxide emitter in the entire U.S., ahead of large industrial facilities like BASF’s plant in Geismar, Louisiana., and Eastman Chemical’s site in Longview, Texas. According to the EPA data, SSV’s facility has the highest ethylene oxide emissions of all medical sterilizers in the country.

The EPA had flagged SSV in 2022 as one of 23 medical sterilizers nationally that emitted potentially dangerous levels of ethylene oxide. At the time, the agency estimated that the lifetime cancer risk in residential areas near the facility was 100 in a million. That means if 1 million people were exposed to the level of EtO in the air near SSV 24 hours a day for 70 years, at least 100 people would be expected to develop cancer from that exposure.

At a town hall meeting in 2023, the EPA sought to alleviate local residents’ worries about their cancer risk. SSV had “recently installed new controls that have dramatically reduced those emissions,” according to a Henrico Citizen report at the time.

When asked about the current fugitive emissions at SSV, a Virginia DEQ spokesperson told Hunterbrook that the higher emission numbers are “due to calculation differences” and are “not reflective of an ‘increase’ in fugitive emissions.”

When Hunterbrook followed up and asked what those calculation differences are, the DEQ said that the increased fugitive emissions are due to a misunderstanding on SSV’s part on how to report these emissions. The company had since “rectified their emissions calculation process,” a DEQ spokesperson said.

As early as 2019, Sierra Club’s Virginia chapter had pointed to SSV as one of the state’s major emitters of hazardous pollutants. Its report, “Air Toxins and Health Risk in Virginia,” showed that the Henrico facility had the highest Risk-Screening Environmental Indicators (“RSEI”) score in the entire state. RSEI is an EPA metric that estimates the relative potential human health risk from toxic chemical releases based on the quantity released, chemical toxicity, and population exposed. According to the Sierra Club report, the potential health risk associated with SSV’s facility was 10,000 times higher than the potential risk of the entire state of Vermont at the time.

The SSV facility opened in 1990 and “occupies 76,000 square feet of processing and warehouse space,” according to the company’s website. Sterilization Services also operates a second facility in Atlanta, Ga. Its third facility, Sterilization Services of Tennessee in South Memphis, closed in 2024 after persistent community activism against its EtO emissions.

Regulatory Rollback

The April roof leak at SSV shows that while emissions controls can reduce ethylene oxide emissions in daily operations at medical sterilizers, the risk of a large accidental release of the hazardous pollutant persists.

And now the EPA aims to roll back several Biden-era safeguards against ethylene oxide emissions that are already in place.

When the agency announced these regulations in 2024, it said they would eliminate over 90% of ethylene oxide emissions from commercial sterilizer facilities and reduce the number of people exposed to cancer risks from ethylene oxide by 92%.

In a March announcement, the agency struck a different tone: “EtO is the only safe and effective sterilization method available for many medical devices,” the agency wrote. “EPA is concerned that the current Biden-era EtO emission standards actively threaten facilities’ ability to sterilize equipment and jeopardize one of America’s only options for a secure domestic supply chain of essential medical equipment.”

Data cited by the U.S. FDA shows that approximately 50% of all sterile medical devices in the U.S. are sterilized with ethylene oxide. Other treatment methods include steaming and radiation.

The EPA’s legal argument for rescinding the new rules is that the Clean Air Act prevents the agency from conducting more than one risk review. Since the EPA had already completed a review for ethylene oxide in 2006, argues the agency, the second risk review the Biden EPA conducted in 2024 should not have been completed.

“What the current administration is arguing is, you don’t have the authority to do that. You get one shot at the apple, and it doesn’t matter how much more information science provides you after you’ve regulated,” said Goffman, who worked for the EPA under former Presidents Barack Obama and Joe Biden. “We can’t let these guys win on the absurd notion that Congress meant for the agency to remain ignorant while the rest of the world is increasing its understanding of human health effects of certain chemicals.”

Among other things, the EPA is planning to change the requirement to use continuous real-time emission monitoring systems. Under the proposed change, the EPA will give sterilizers the choice between installing emissions monitoring systems or annual performance testing and parametric monitoring, where key indicators of system performance are measured instead of actual emissions.

Louisa Dickison already submitted her comment on the proposed new rules.

“It’s just appalling and unfathomable to support anything other than basic community awareness and protection,” she told Hunterbrook. “What is the EPA’s job if not to protect the environment, and the environment includes the people who live in it.”

The EPA did not respond to a request for comment in time for publication.

Authors

Till Daldrup joined Hunterbrook from The Wall Street Journal, where he focused on open-source investigations and content verification. In 2023, he was part of a team of reporters who won a Gerald Loeb Award for an investigation that revealed how Russia is stealing grain from occupied parts of Ukraine. He has an M.A. in Journalism from New York University and a B.S. in Social Sciences from University of Cologne. He’s also an alum of the Cologne School of Journalism (Kölner Journalistenschule). Till is based in New York.

Evan Comen is currently the senior data editor at U.S. News & World Report, where he focuses on government rankings and accountability reporting. He has worked as a data journalist since 2015, covering climate change, urban economics, and public policy. Evan has a B.A. in economics from the University of North Carolina at Chapel Hill and is based in New York.

editor

Sam Koppelman is a New York Times best-selling author who has written books with former United States Attorney General Eric Holder and former United States Acting Solicitor General Neal Katyal. Sam has published in the New York Times, Washington Post, Boston Globe, Time Magazine, and other outlets — and occasionally volunteers on a fire speech for a good cause. He has a B.A. in Government from Harvard, where he was named a John Harvard Scholar and wrote op-eds like “Shut Down Harvard Football,” which he tells us were great for his social life. Sam is based in New York.

Graphic

Dan DeLorenzo is a creative director with 25 years reporting news through visuals. Since first joining a newsroom graphics department in 2001, he has built teams at Bloomberg News, Bridgewater Associates, and the United Nations, and published groundbreaking visual journalism at The Wall Street Journal, Associated Press, The New York Times, and Business Insider. A passion for the craft has landed him at the helm of newsroom teams, on the ground in humanitarian emergencies, and at the epicenter of the world’s largest hedge fund. He runs DGFX Studio, a creative agency serving top organizations in media, finance, and civil society with data visualization, cartography, and strategic visual intelligence. He moonlights as a professional sailor working toward a USCG captain’s license and is a certified Pilates instructor.

Hunterbrook Media publishes investigative and global reporting — with no ads or paywalls. When articles do not include Material Non-Public Information (MNPI), or “insider info,” they may be provided to our affiliate Hunterbrook Capital, an investment firm which may take financial positions based on our reporting. Subscribe here. Learn more here.

Please email ideas@hntrbrk.com to share ideas, talent@hntrbrk.com for work, or press@hntrbrk.com for media inquiries.

LEGAL DISCLAIMER

© 2026 by Hunterbrook Media LLC. When using this website, you acknowledge and accept that such usage is solely at your own discretion and risk. Hunterbrook Media LLC, along with any associated entities, shall not be held responsible for any direct or indirect damages resulting from the use of information provided in any Hunterbrook publications. It is crucial for you to conduct your own research and seek advice from qualified financial, legal, and tax professionals before making any investment decisions based on information obtained from Hunterbrook Media LLC. The content provided by Hunterbrook Media LLC does not constitute an offer to sell, nor a solicitation of an offer to purchase any securities. Furthermore, no securities shall be offered or sold in any jurisdiction where such activities would be contrary to the local securities laws.

Hunterbrook Media LLC is not a registered investment advisor in the United States or any other jurisdiction. We strive to ensure the accuracy and reliability of the information provided, drawing on sources believed to be trustworthy. Nevertheless, this information is provided “as is” without any guarantee of accuracy, timeliness, completeness, or usefulness for any particular purpose. Hunterbrook Media LLC does not guarantee the results obtained from the use of this information. All information presented are opinions based on our analyses and are subject to change without notice, and there is no commitment from Hunterbrook Media LLC to revise or update any information or opinions contained in any report or publication contained on this website. The above content, including all information and opinions presented, is intended solely for educational and information purposes only. Hunterbrook Media LLC authorizes the redistribution of these materials, in whole or in part, provided that such redistribution is for non-commercial, informational purposes only. Redistribution must include this notice and must not alter the materials. Any commercial use, alteration, or other forms of misuse of these materials are strictly prohibited without the express written approval of Hunterbrook Media LLC. Unauthorized use, alteration, or misuse of these materials may result in legal action to enforce our rights, including but not limited to seeking injunctive relief, damages, and any other remedies available under the law.

By: Jenny Ahn, Till Daldrup, Blake Spendley

Editor: Sam Koppelman

Based on Hunterbrook Media’s reporting, Hunterbrook Capital is short $CSG.AS and long a basket of comparable securities at the time of publication. Positions may change at any time. See full disclosures on our website.

• In the largest European military IPO in history, Czechoslovak Group (CSG.AS) pitched investors a chance to buy the next Rheinmetall, riding the continent’s recent rearmament boom. Hunterbrook Media’s investigation — drawing from statutory filings, subsidiary accounts, satellite imagery, and a range of other sources — suggests investors bought something very different: an arms-dealer marking up old ammunition, with a critical subsidiary under NATO suspension, production numbers that don’t add up, and a major minority owner with “extraordinary rights” holding a sword of Damocles over the company’s balance sheet.

• The story began to fall apart almost immediately after the IPO. In January 2026, CSG raised 3.8 billion euros, with BlackRock and Qatar’s sovereign wealth fund anchoring the deal. 33-year-old owner Michal Strnad cashed out 2.55 billion euros. By February, the nonprofit Ján Kuciak Investigative Center reported that the record 58 billion euro Slovak ammunition framework deal — which CSG highlighted as a growth catalyst during the IPO process — had not been formally joined by any of the eight countries Slovakia claimed were interested. A month later, another nonprofit investigative journalism outlet Follow the Money reported that CSG’s Spanish ammunition factory, which produces essential propellant, had been suspended by NATO’s procurement agency over alleged “sanctionable practices.” Then, a month after that: Another outlet, Seznam Zprávy, reported that Petr Kratochvíl — a minority shareholder with “extraordinary” rights over CSG’s most important subsidiary — had exercised a put option days before the IPO and was demanding 1.4 billion euros. None of this was clearly or comprehensively disclosed in the prospectus.

• CSG’s prospectus positions it as a rival to Rheinmetall, but Hunterbrook’s analysis suggests the vast majority of its ammunition revenue comes not from making shells but from reselling them. Ammunition was two-thirds of the company’s revenue last year and the primary driver of its guided margin increase. The prospectus claims a current capacity of roughly 630,000 large-caliber rounds a year and says 80% of CSG’s production is 155 mm ammunition, the NATO-standard artillery most in demand. Read together, those two figures suggest CSG is producing roughly half a million 155 mm rounds a year in-house — putting it in the same ballpark as Rheinmetall. But “production” is not the same as “production capacity,” and CSG may be exploiting the ambiguity. When Hunterbrook asked CSG to clarify whether it really could produce around 500,000 155 mm rounds, the company refused to answer — saying it does not disclose capacity by ammunition type.

• Hunterbrook’s own reconstruction of CSG’s production footprint points to most of its revenue coming from recommissioning. We could identify only one facility, the Dubnica plant, that appears capable of final assembly of 155 mm rounds; its estimated output last year was between 100,000 and 280,000 rounds. The combined revenue of all four ammunition-producing subsidiaries we identified totaled no more than an estimated 524 million euros in 2024 — roughly a fifth of the 2.5 billion euros CSG reported in medium- and large-caliber ammunition sales that year. The rest likely came from re-selling: acquiring ammunition from third parties, refurbishing it, and reselling it at a markup. CSG has since invested in new production lines — but its total factory upgrades of roughly 205 million euros appear to amount to a fraction of what Rheinmetall has invested in capacity expansion so far.

• The reselling model depends on a global stockpile that may be running out. CSG’s founder once described the core trade bluntly: “You buy one warhead for CZK50 and immediately sell it abroad for CZK150 or more.” It’s a good trade — when there is supply. But CSG admits it has “limited visibility” on remaining stock available for recommissioning. A Reuters investigation put the global market of available large-caliber ammunition at roughly two million rounds as of 2024 — a number Ukraine could burn through imminently. And a key pipeline for lucrative ammunition sales — the so-called “Czech ammunition initiative” — may be losing political support and funding. The new pro-Russia Czech prime minister Andrej Babiš, elected last year, is halting the country’s funding for the program, which procures ammunition through intermediaries like CSG. Other key members like Germany, the Netherlands, and Denmark, are also drawing back support. During his campaign, Babiš said “he was uncomfortable with individuals earning excessive profits from the war,” according to the publication Liga. One document obtained by RFE/RL illustrates that point: CSG was selling ammunition to the Czech government at a 28% markup over similar shells a Turkish company was offering. CSG denies profiteering.

• The minority shareholder seeking 1.4 billion euros for his stake has veto rights over the entity where much of CSG’s cash and trading activity actually lives. CSG is disputing Petr Kratochvíl’s asking price, offering just a tenth of that, but Kratochvíl, the former head of Excalibur Army, owns 10% of CSG Land Systems — the holding company above EA — and holds extraordinary blocking rights over key corporate decisions. EA reported 2.6 billion euros in revenue — nearly two-thirds of the group’s total — and held roughly 575 million euros in cash in 2024. He also has an 8.9% stake in MSM Group, the ammunition parent. The prospectus did not name Kratochvíl or disclose his exercise of the put option. Asked why, CSG told Hunterbrook that its outside legal counsel confirmed Kratochvíl did not effectively exercise this right. Hunterbrook could not reach Kratochvíl for comment, but he has publicly claimed he exercised the right both digitally and in-person.

• Kratochvíl appears to be part of a broader pattern of undisclosed or under-disclosed insiders around CSG’s core subsidiaries and business practices. Before the IPO, CSG transferred 20 to 30 subsidiaries to a vehicle personally owned by Michal Strnad. The IPO described the disposal but did not name Strnad as the recipient. 275 million euros in receivables from those deals remain uncollected. Separately, Excalibur Army disclosed a roughly 4.7 million euro receivable principally from Jaroslav Strnad, the founder and CEO’s father, while also writing off more than 13 million euros worth of time-barred receivables. CSG told Hunterbrook the receivables “will be fully settled in cash” and dismissed the write-off at Excalibur Army as an accounting necessity rather than an actual loss.

• Some of the money pooling may be making its way from CSG insiders to political partners: CSG, for instance, appears to pay rent to a former vice-chairman’s real estate company. That former vice-chairman, Miroslav Dorňák, also controls FinGood, a crowdfunding platform that Seznam Zprávy reported paid millions to the law firm of Robert Kaliňák — the Slovak defense minister behind the 58 billion euro ZVS announcement. CSG told Hunterbrook its payments to Dorňák reflect a “standard landlord–tenant relationship.”

• CSG’s history makes these post-IPO findings less surprising. Jaroslav Strnad built the company by buying, refurbishing, and reselling liquidated Soviet-era military assets. Czech media have tied early surplus deals to family connections inside the Czech defense ministry. Strnad was a major donor to former President Miloš Zeman, accounting for two-thirds of Zeman’s 2018 campaign donations, according to nonprofit investigative journalism group OCCRP, and other outlets have flagged CSG’s contracts and other deals with the government during Zeman’s tenure as irregular.

• Before the Ukraine war, CSG was known for its close ties to Russia and Russian influence networks. JD Vance, now U.S. vice president, wrote to the Treasury Department in 2024 urging the U.S. government to block CSG’s acquisition of Vista Outdoor’s ammunition business, calling CSG “Kremlin-linked” and citing alleged ties to Putin’s inner circle. In its email to Hunterbrook, CSG denied Russian financing allegations and said those relationships ended years ago. But for a company whose valuation rests on selling ammunition to NATO and Ukraine, any residual ties may be meaningful.

• CSG’s performance appears to depend on temporary conditions amid little scrutiny: a wartime shortage, a sufficient supply of old ammunition to refurbish, captive procurement channels, and political tolerance for middleman margins. But the old ammunition stock is shrinking. The Czech initiative is facing political and funding pressure. The Slovak framework may never be more than an idea. Key subsidiary FMG is suspended by NATO. CSG barely converted its 2025 profit into operating cash. And the company is surrounded by unnamed shareholders, related-party vehicles, historical patronage networks, and unresolved legal and reputational problems. If the wartime premium fades before CSG becomes the manufacturer it claims already to be, investors may be left not with the next Rheinmetall, but with a levered, opaque defense conglomerate whose core skill was never mass-producing ammunition — it was finding old ammunition, navigating political systems, extracting wartime margins, and keeping the messy parts out of the consolidated story.

• In response to Hunterbrook’s questions, CSG denied wrongdoing across all areas of inquiry. CSG confirmed its in-house capacity at “more than 600,000 rounds” of medium- and large-caliber ammunition per year, but declined to break it down by type or facility, citing standard industry practice, even after its own prospectus disclosed that 80% of production was 155 mm. The company called its FMG NATO suspension “a temporary and procedural measure” resulting from an internal NSPA investigation into one of the agency’s officials, and said it has had “no material impact.” On the Kratochvíl put option, CSG said outside counsel confirmed Kratochvíl “did not effectively exercise his right” before the IPO, and that auditors found no liability. CSG described the pre-IPO transfer of subsidiaries to Strnad’s Ytara SPV as “standard” restructuring and said all 275 million euros in receivables “will be fully settled in cash”; asked why Strnad was not named in the prospectus as Ytara’s owner, the company said the information is publicly available in the Czech Commercial Register. On the 58 billion euro Slovak deal, CSG emphasized it is a framework agreement with “maximum potential value, not committed orders,” and said it “is not dependent on any single EU funding mechanism.” CSG confirmed that Marián Goga, who faces criminal charges for bribery in Slovakia, “currently serves on the supervisory board of MSM Group” — a fact the prospectus omitted — and said he “has played a positive role in the development of the business.” In addition to CSG denying any wrongdoing across all areas of inquiry, CSG also repeatedly suggested that Hunterbrook’s questions were based on “incomplete information” or did “not fully reflect the underlying facts,” said the IPO process “involved some of the world’s leading financial institutions and law firms” and that it is “inconceivable that any relevant information would not have been properly disclosed.”

• If you know more about CSG, including its potential ties to Russia, please reach out at ideas@hntrbrk.com.

Russia’s invasion of Ukraine may have been the best thing that ever happened to the Czechoslovak Group (CSG).

Before February 2022, CSG was a midsized Czech arms conglomerate. Founded in the mid-1990s as a scrap dealer that refurbished liquidated Soviet-era assets, by the early 2020s, the group had built a small arms empire assembled through aggressive acquisitions of Eastern bloc factories. On the eve of the Russian war on Ukraine, CSG stood at about 580 million euros in revenue with around 3,700 employees.

In just four years, CSG’s reported revenue had increased over tenfold to 6.74 billion euros. The employee headcount nearly quadrupled. CSG was one of the few commercial suppliers in Europe capable of overhauling and delivering Soviet-standard tanks, infantry fighting vehicles, and spare parts that Ukrainian crews were already trained to operate.

More importantly, the invasion gave new life to CSG’s artillery ammunition business. Ukraine was burning through artillery ammunition faster than NATO could produce it, and ramping up production became a top priority for alliance countries.

Deliveries to Ukraine hit 42% of sales in 2024, CSG reported. The rest of the sales mainly went to NATO governments scrambling to arm Ukraine and replenish their depleted stockpiles. In 2025, CSG attributed nearly two-thirds of its revenue, or about 4.1 billion euros, to medium- and large-caliber ammunition sales — an extraordinary figure considering Europe’s leading defense supplier, Germany-based Rheinmetall, reported just 3.5 billion euros in revenue for its combined weapons and ammunition segment.

By late 2025, the company was ready to cash in. A syndicate of eight banks — with J.P. Morgan, Morgan Stanley, and Deutsche Bank among the most recognizable names — signed on as coordinators and bookrunners. They produced a 728-page prospectus, approved by the Dutch Authority for the Financial Markets, a financial regulator. BlackRock, Qatar’s sovereign wealth fund, and Artisan Partners each anchored 300 million euros.

Together, these groups shepherded CSG through a 25 billion euro listing on Euronext Amsterdam in January 2026 — the “world’s largest defence IPO ever recorded,” according to the press release. The story resonated: Missed Rheinmetall? Don’t sweat it. There was a new, publicly traded way to get long the militarization of Europe.

Investors piled in — $57 billion in orders, reportedly 14 times oversubscribed. Shares soared 31% on day one. The company raised 3.8 billion euros. Michal Strnad, the 33-year-old who inherited the arms empire from his father, cashed out 2.55 billion euros in the offering, securing his place as the richest man in the Czech Republic.

But within weeks of the listing, a series of revelations began laying bare what the prospectus — and those eight banks — missed.

In February, journalists from Czech outlet the Ján Kuciak Investigative Center revealed that a 58 billion euro ammunition contract with Slovakia — announced weeks before the IPO and cited as a growth catalyst — appeared to be more bark than bite: None of the eight countries Slovakia claimed would participate confirmed plans to join. Another watchdog, Follow the Money, then reported that CSG’s key Spanish ammunition factory had been suspended by NATO’s procurement agency six months before the listing.

Then, another bombshell: Czech investigative news outlet Seznam Zprávy reported that a disgruntled minority shareholder with extraordinary blocking rights had demanded a 1.4 billion euro buyout just days before the IPO.

None of that was clearly or comprehensively disclosed in the prospectus. In fact, CSG apparently considered disclosing the date of Kratochvíl’s request in its March annual report but later deleted it: Seznam Zprávy reported that it had discovered the old language in invisible ink, which CSG claimed was due to a software glitch. (Side note: In 2024, CSG CEO Strnad pressured the owner of Seznam to sell his media business — threatening “ruinous fines” from Czech authorities if he didn’t comply, according to Czech news outlet Respekt.1)

Notably, the date that CSG considered adding was January 24, 2026, four days after Kratochvíl claims he exercised his right. The difference matters a lot: The IPO was on January 23, meaning CSG still had time to disclose Kratochvíl’s put option to investors — but chose not to do so. In a comment, CSG says this was because Kratochvíl exercised the option improperly.

Hunterbrook Media’s investigation — based on a review of subsidiary filings across the Czech Republic, Slovakia, and Spain buried in CSG’s consolidated numbers — suggests the prospectus omissions go well beyond what’s already surfaced. What we found challenges not just CSG’s governance but its fundamental business model.

Underneath the prospectus calling CSG “a leading defence group” and “a top two European player” in ammunition, and the press releases touting “state-of-the-art” filling lines and “one of the most modern workplaces within Europe,” we found a hodgepodge of Soviet-era factories, a history of political patronage, and widespread reselling of old ammunition.

CSG admits in its prospectus that a portion of its ammunition production comes from “recommissioning” — lower quality ammunition like expired inventory it acquires, upgrades, and returns to the field. It doesn’t explain, however, what proportion of its total sales come from this approach. Hunterbrook’s investigation suggests it has been the vast majority, with CSG’s in-house manufacturing capacity responsible for a fraction of total output. CSG calls recommissioning a “nimble and entrepreneurial” approach — but later admits “limited visibility” on the global stock of ammunition to be resold.

CSG has promised to lessen its reliance on outside supply by vertically integrating the ammunition supply chain, with a goal of achieving production of 1.1 million large-caliber rounds by 2028, putting the upstart ammo dealer on a par with Rheinmetall’s multibillion euro production machine. But the gap between that promise and today’s reality is vast, our findings show — and it likely requires significantly more investment than CSG has made so far.

It’s an investment CSG may not have the cash to embark on: While CSG reported 1.6 billion euros in operating profit in 2025, it converted just 3.8% of that to operating cash flow. And the company sat on 3 billion euros in net debt as of the end of last year — nearly twice its operating profit.

CSG appears to be trying to obscure these facts by adorning its growth narrative with big promises and flashy announcements — while important details about its business are left out of the public record. But the numbers exist, across primary source and subsidiary-level documents, and they reveal a central insight: CSG is still running its business like a post-Soviet era conglomerate in the shadow-world of Eastern European defense markets, where it could brush off public scrutiny over questionable practices.

Hunterbrook found other examples of questionable ways CSG is still doing business — from cash disappearing into Strnad’s orbits, to a former minority shareholder and key operator of the company now facing criminal charges for corruption. None of these are in the prospectus or in the annual consolidated statements.

But CSG is now a publicly listed company on a global exchange, backed by some of the world’s biggest investment funds and with obligations to public shareholders. The practices that built the empire will face a very different level of scrutiny from regulators, investors, and journalists than the company may have grown accustomed to.

Hunterbrook is committed to providing that scrutiny. The following represents just some of what we have found in a matter of months. The reporting continues. Reach out at ideas@hntrbrk.com if you know more.

Can CSG Actually Make the Shells It Is Promising?

It would be the biggest deal in CSG’s history. And it was disclosed in December 2025, just weeks before the company’s IPO.

The announcement: ZVS Holding, a 50/50 joint venture between CSG and the Slovak government, had signed a contract with Slovakia’s defense ministry to provide up to 58 billion euros worth of medium- and large-caliber ammunition over the next seven years. The deal is not a firm purchase order but a framework agreement, meaning it sets the terms under which orders could be placed.

Slovakia, which is undergoing budget cuts in order to save 2.7 billion euros, assured the public that the deal wasn’t just for them. Slovak Defense Minister Robert Kaliňák claimed eight countries wanted to join: Croatia, Greece, Romania, Italy, Poland, the Czech Republic, Belgium, and the Netherlands. “We assume that specific contracts will be concluded within two to three months, and deliveries will start next year,” Kaliňák said in an interview with Slovak newspaper Pravda in December 2025.

It turned out the announcement may have been more hype than substance. None of the countries Kaliňák said would join the deal actually confirmed any intention to join. Slovak investigative news outlet ICJK, in cooperation with other journalists, contacted every named country, and while one indicated the proposal was still under consideration, others were outright dismissive of Kaliňák’s claims. “This is not currently relevant for the Netherlands,” the country told ICJK. CSG’s home country of the Czech Republic went a step further, blocking any consideration of the deal after determining it was akin to a no-bid contract, ICJK reported.

The participation of other countries is crucial for Slovakia to get special 1% financing through the EU’s flagship 150 billion euro Security Action for Europe program, which is possible only if at least two members participate. The exemption allowing single-country participation expires at the end of May 2026. If Kaliňák doesn’t secure a partner by then, low-cost EU financing for the ammunition purchases may become harder.

In an email to Hunterbrook, CSG emphasized the deal is a framework agreement with a “maximum potential value, not committed orders,” and that the deal is not dependent on any single EU funding mechanism. The company added, “serious discussions are ongoing with several potential customers” but declined to name them, citing commercial confidentiality.

But the eye-popping 58 billion euro figure drew attention to a more fundamental question: Even if the contract does convert, is CSG physically able to make the shells it’s promising?

CSG told Bloomberg in an email that the figure “reflects the maximum potential volume of deliveries over the seven-year period,” estimated based on the “total value of available production capacity.” Kaliňák vouched for the company’s ability to deliver on the deal, adding that the deal is based on what can be produced in three shifts. The company recently inaugurated a new filling line at its Dubnica nad Váhom plant that it claimed would help fulfill the Slovak deal and reach its production target of a million large-caliber rounds per year.

But CSG and ZVS allegedly dodged questions from the ICJK about the exact production capacities, citing “security and competition reasons.” (In contrast, Rheinmetall, CSG’s larger competitor, has been more explicit about its current and projected production capacity and timeline. See its March 2026 Investor Presentation, page 34, and its 2025 Annual Report, page 51)

Ambiguity grew when CSG announced that “operating conditions allow for a two-shift mode” at the newly updated plant — seemingly at odds with Kaliňák’s comment that the target production is achievable through a three-shift operation. A defense ministry spokesperson responded to an ICJK question on the discrepancy by accusing the outlet of having “fundamentally incorrect information” as before, and offering a vague rebuttal: “crucial components can also be produced in three-shift operation.”

Even with three shifts, however, it’s unclear CSG could produce anything close to 58 billion euros worth of ammunition.

In fact, it’s unclear how much CSG is producing at all. Or what proportion of CSG’s claimed 4.1 billion euros in revenue from M/L ammunition sales is coming from real in-house manufacturing.

Beyond the 58 billion deal, medium- and large-caliber ammunition is central to the group’s growth thesis, with the company arguing demand will continue to climb even after the war in Ukraine ends, driven by NATO’s stockpile refill requirements. It was the group’s “largest revenue driver,” contributing 61% of total revenues in 2025, according to the company. CSG reported a nearly 7 billion euro backlog on M/L caliber ammunition sales last year (and 15 billion euros across all segments), with an additional 27 billion euros in the pipeline for the whole group — including about 1 billion euros from the recent Slovak deal.

Hunterbrook’s investigation, however, raises fundamental questions about CSG’s production capacity — and how much the company can actually cash in on this opportunity.

Demystifying CSG’s Ammunition Production Scale

To start, it’s very hard to even understand precisely what CSG is claiming regarding its capacity to produce ammunition.

In one place in the prospectus, CSG says it can make 630,000 rounds; in another, 550,000 rounds. In an email to Hunterbrook, CSG affirmed it had “more than 600,000” of capacity.

More confusing is what those numbers represent.

In some places, including page 109 of its prospectus, CSG says this total refers just to large-caliber ammunition. In its email to Hunterbrook, CSG said it covered both medium- and large-caliber ammunition, but also, in that same email, CSG’s spokesperson directed Hunterbrook to look at page 109.

On page 109 of the prospectus, CSG also says it plans to increase its production to 1.1 million large-caliber munitions, substantially, by the end of 2028. If true, those figures would put it on a par with its much larger peer Rheinmetall, whose current production capacity and near-term goals are just slightly ahead of CSG’s.

The apparent discrepancy in CSG’s numbers may be small, but the precise definition of what these numbers represent is critical. 630,000 large-caliber rounds is a very different proposition than 550,000 medium- and large-caliber rounds combined — if you’re trying to understand CSG’s true ammunition production capacity.

Large-caliber rounds, especially 155 mm artillery rounds, are responsible for much of the company’s narrative; they’re the workhorse of Ukraine’s fight against Russia, the NATO-standard artillery in most critical need; and a big reason Rheinmetall saw its stock price increase more than 20 times after the invasion. An investigation by a consortium of journalists in 2024 put the total Western European artillery output at just 500,000 to 600,000 rounds per year — a fraction of Ukraine’s burn rate of about 2.4 million and the EU’s stated 2 million target. (Watch this YouTube video by The Wall Street Journal on the 155 mm ammunition shortage.)

155 mm rounds appear to go for anywhere between 3,000 and 8,000 euros apiece, reflecting a significant wartime premium; medium-caliber rounds are priced in the low hundreds.2 So the lion’s share of CSG’s revenue, very plausibly, comes from the sale of large-caliber rounds.

And a reader of the prospectus could be forgiven for thinking CSG makes around half a million 155 mm rounds per year — since the company says that 80% of its M/L production is 155 mm; and its total production capacity is around 600,000.

This is an impressive-sounding claim that, if true, makes CSG’s manufacturing footprint almost comparable to Rheinmetall’s.

But that’s far from clear.

Hunterbrook took three approaches to stress-test these claims. In none of them could we find evidence that CSG is producing anything close to the number of 155 mm rounds Rheinmetall is producing in-house.

And all of the approaches raised more questions than answers as to how to explain the roughly 4.1 billion euros CSG reported in M/L ammunition revenue in 2025.

Now, a caveat: Military production can be obfuscated. And we may have missed something — say, a secret, underground propellant manufacturer. Our calculations could also be wrong, and if you have better ideas for how to back into these numbers, please write us at ideas@hntrbrk.com.

But based on what is publicly available, CSG’s in-house production numbers, alone, cannot come close to explaining its rapidly growing revenue.

Method 1: Matching CSG’s Capacity Claim With On-the-Ground Realities

The idea here is to test CSG’s production claims by actually looking at CSG’s production capacity on the ground.

The prospectus makes two key claims: CSG has capacity for roughly 630,000 large-caliber rounds a year, and 80% of its M/L ammunition production is 155 mm. Read together, an investor might naturally conclude CSG is producing roughly half a million 155 mm rounds a year in-house. Eighty percent of 630,000 is 504,000.

But there’s a critical distinction buried in the language. The 630,000 figure refers to “capacity.” The 80% figure refers to “production.” Those aren’t the same thing. If CSG is running below full capacity, actual 155 mm output could be far lower than investors might assume.

Or, CSG might be counting production of components, say shell bodies, rather than fully-assembled rounds, as part of overall production capacity. The spokesperson said in CSG’s email to Hunterbrook that the term “in-house production capacity” is meant to refer “to the Group’s ability not only to carry out final assembly of large-calibre ammunition, but also to manufacture a substantial portion of its key components within the CSG industrial network.3”

CSG appears to be aware of the ambiguity: When Hunterbrook asked the company to confirm whether it can produce around half a million rounds of 155 mm per year, CSG refused, saying it does “not disclose detailed production capacity by specific ammunition types” and that “production capacity is not a fixed or single-dimensional metric.”

The refusal is notable given that CSG’s own prospectus disclosed both figures that produce that math.

So the question is, does CSG have the manufacturing footprint to produce anything close to half a million rounds of 155 mm ammunition in-house?

CSG doesn’t clearly disclose exactly which of its factories make ammunition, much less break down the full quantity by type of ammunition. And as mentioned, information across CSG’s various public statements — press releases, websites, and company disclosures to investors — is often contradictory, sometimes even across pages in the same report.

For example, in the prospectus, the company says there are 14 ammunition factories on one page (page 95), then says there are 12 on another (page 99).

But in the sustainability section of the 2025 annual report, where the company is legally required under EU laws to describe its sector and operations, including aspects of its value chain (pages 124–126), Hunterbrook counted only seven operating entities described as involved in manufacturing ammunition.4 Of those, three don’t seem to make fully assembled artillery ammunition, but rather components for ammunition — like shell casings or explosives — or only make ammunition for small firearms.5

That seemingly leaves only four factories that actually make finished M/L ammunition: the Dubnica factory that CSG co-owns with the Slovak defense ministry; VOP Nováky, a longstanding CSG ammunition factory in Slovakia; Fábrica de Municiones de Granada (FMG), a 700-year-old ammunition factory in Spain, acquired in 2020; and ZVI a.s., a Czech ammunition factory that CSG recently purchased.6

Of those four, Dubnica appears to be the only CSG facility the company has described as capable of producing fully assembled 155 mm rounds. CSG’s three other ammunition-producing plants do not seem to mention such capability: CSG’s website mentions the VOP Nováky plant making tank and mortar rounds, rockets, and ammunition components. It also says it’s involved in the “life cycle extension of ammunition of 57-155 mm calibre” — i.e. refurbishing; an ICJK investigation in 2024 said VOP Nováky only makes 155 mm training rounds, effectively meaning rounds with no explosives.

As for FMG, an article on its website dated November 2025 describes its main production as tank ammunition. It also mentions making mortar ammunition, as well as propellants and explosives — components needed for 155 mm artillery ammunition. The article doesn’t mention making 155 mm ammunition itself. ZVI only makes medium-caliber ammunition.

CSG finished upgrades at Dubnica in late 2025 that it says will allow the factory to produce up to 280,000 rounds in 2026. In 2025, however, total output at Dubnica was likely around 100,000 rounds per year, according to ICJK.

That suggests CSG’s in-house production capacity for 155 mm rounds at Dubnica is somewhere between 100,000 and 280,000.

That’s nowhere close to the half a million rounds investors might infer from the prospectus.

To be sure, in its email to Hunterbrook, CSG denied that Dubnica is the only location where “final assembly of 155mm ammunition takes place,” but declined to identify the other facilities, citing commercial and security reasons. CSG also explained that large-caliber ammunition is a “complex system,” and therefore, it “would not be accurate to attribute production of 155mm ammunition to a single site.”

Which raises another question: Does CSG include “recommissioning” in its production numbers?

CSG explains in its prospectus that one of its methods for producing ammunition is what it calls “recommissioning”: “The Group acquires lower quality large caliber ammunition from NATO member armies (such as expired inventory) and then inspects, upgrades and reassembles inventory for return to the field,” CSG writes.

Just how much recommissioning CSG does — relative to in-house production — is the key question, one the company did not answer directly in response to Hunterbrook’s questions. So we tried to figure it out.

Method 2: Subsidiary Revenue Analysis

To answer that question, Hunterbrook looked at the revenue reported by each of the subsidiaries that we identified as making ammunition to their respective state registries in the Czech Republic, Slovakia, and Spain. The idea is straightforward: If CSG is really manufacturing most of its ammunition in-house, the factories doing the manufacturing should be reporting revenue that, in aggregate, accounts for the bulk of the group’s ammunition sales, after accounting for the margin on final sale.

In its 2024 filing with the Czech registry — the latest available — ZVS reported 119 million euros in revenue, of which 85 million euros were likely from in-house production; the rest was labeled as revenue from goods resold or services. VOP Nováky reported 260 million euros in revenue in 2024, of which about 160 million euros were likely from in-house production. FMG reported 163 million euros in revenue. A portion of that is revenue from components of ammunition, like propellant charges, rather than finished ammunition, but FMG did not break out the revenue by these subitems.

ZVI’s filings with the Czech registry showed it was a loss-making company with 37 employees and revenue of less than half a million euros in 2024. CSG provided a pro forma table showing what the revenues would have looked like if the acquisition had occurred at the beginning of the reporting period, which shows just 15 million euros from several acquisitions clumped together, including ZVI — suggesting CSG’s expected revenue from ZVI in 2026 was a fraction of 15 million euros.7

In total, the combined revenue in 2024 reported by ZVS, VOP Nováky, ZVI, and FMG — net of revenue from components, services, or resold goods where those figures were reported — was no more than 409 million euros, according to a Hunterbrook analysis.

Now, CSG is a complex group with extensive intragroup sales: Factories don’t always sell directly to end customers. Instead, they sometimes sell to trading entities like Excalibur Army, which then resells to governments and militaries at the final market price. That means the revenue reported by the manufacturing subsidiaries may understate the consolidated revenue that in-house production ultimately generates — because the markup to the end customer is booked at the trading entity, not the factory.

To account for that, Hunterbrook applied the estimated 28% operating EBIT margin for the Defence Systems division — a generous assumption in some ways, since it ignores that some factories do sell directly to end customers and their revenue almost certainly already reflects those margins.8 After that addition, you get about 524 million euros for the estimated total revenue in 2024 from ammunition made by its factories. That’s roughly a fifth of the total M/L ammunition revenue of 2.5 billion euros reported by the company for 2024.

Hunterbrook used revenue data from 2024, because that’s the latest year for which subsidiary files were available from the Czech business registry.

Meaning, roughly 80% of CSG’s total ammunition revenue the company reported in 2024 seemingly can’t be accounted for by in-house production, and therefore likely came from third-party sources, according to Hunterbrook’s analysis.

CSG has since made meaningful progress to expand its in-house production capacity, including a new 155 mm filling line at Dubnica that was inaugurated in December 2025. Which brings us to the question: Just how much new capacity could CSG have brought online?

Method 3: Comparing CSG to Rheinmetall’s Manufacturing Footprint

This approach is premised on the assumption that if CSG can really produce around as much ammunition as Rheinmetall, then its investment in its manufacturing capacity should also roughly mirror Rheinmetall’s.

CSG appears like it actually had a bit of a head start. CSG said it could make less than 200,000 rounds a year in 2022; Rheinmetall says its numbers were about 130,000 (based on the company’s chart showing 70,000 155 mm rounds and 60,000 tank rounds).

Across the four existing CSG ammunition-producing plants Hunterbrook identified, CSG has since reported the completion of improvements costing a total of around 205 million euros: installing a new filling line and a new shell line at its ZVS plants, and improving automation at VOP Nováky. In 2024, CSG’s Defence division CEO mentioned “big plans” for significant investments in FMG, but it’s unclear what the improvements were or how much it would cost. CSG has also announced plans to invest in “extensive modernization and expansion of ZVI’s production.” But details on the plan are unclear.

Compare that to the nearly 500 million euros Rheinmetall spent on a greenfield ammunition plant in Germany completed last year alone, which is expected to make 350,000 155 mm shells per year by 2027 after a full ramp-up.

That’s one site. Rheinmetall also runs a Spanish firm it acquired for 1.2 billion euros in 2023, which analysts say added roughly 300,000 rounds per year. Then there is an additional retrofitted Berlin auto plant where more than 200 workers have been shifted from automotive pumps to forging 155 mm casings — plus new plants under construction in Romania, projects in Lithuania and the U.K., and expanded capacity in Hungary, South Africa, and Australia.

CSG and Rheinmetall are targeting similar production ramp-ups on similar timelines: In its prospectus, CSG compares its target to “substantially” achieve 1.1 million medium- and large-caliber production by the end of 2028 with Rheinmetall’s target to reach the same goal by 2027. But again, CSG appears to be blurring important facts.

Reading the fine print, CSG seems to be comparing its own 1.1 million medium- and large-caliber ammunition target with Rheinmetall’s 1.1 million 155 mm-specific target. According to Rheinmetall’s disclosures, the company is actually targeting around a total of 4.5 million rounds, including medium- and other large-caliber ammunition.

Moreover, CSG hasn’t explained exactly where the ammunition is currently coming from, or how or when it plans to nearly double production. When asked, CSG argued: “We do not disclose detailed production capacity by specific ammunition types. This is standard practice in the defence industry, and such granular data is not publicly provided by manufacturers.”

Instead, CSG offers a chart without precise labels, scale, or timeline details to suggest roughly half of its revenue is in-house currently, a number that will presumably grow to become the vast majority of revenue by an unspecified “mid-term” time horizon.

Achieving full in-house production requires vertically integrating manufacturing capacities for each of the components of a round. A complete 155 mm artillery round has four main components: the shell body, the explosive fill (typically TNT), a fuse, and a propellant charge; the propellant charge often relies on nitrocellulose as a precursor.

Of the components, TNT and propellants are the most constrained input in European ammunition manufacturing, according to industry experts. Rheinmetall’s CEO has repeatedly called propellant production “a strategic bottleneck” in the defense industry. CSG itself has acknowledged this, saying “the availability of such propellant charge systems is currently one of the main limiting factors in increasing artillery ammunition production capacity in Europe.”

Both companies have publicly committed to addressing the propellant and explosives bottlenecks by bringing the capabilities in house. CSG is guiding to flat margins this year, but increased margins in the midterm as the “benefits of vertical integration and production capacity-ramp up” kick in.

In March, CSG announced plans to invest 300 million euros in a joint venture with European propellant manufacturer Eurenco to build a new propellant plant in Slovakia. It also purchased a nitrocellulose plant in Walsrode, Germany, last year, as part of a 141 million euro acquisition9 and invested 50 million euros in a joint venture in Greece to restart TNT production.10

But again, the scale of CSG’s investment pales in comparison to its competitor, Rheinmetall. Rheinmetall announced in November that it is planning to invest 535 million euros in building a new nitrocellulose plant in Romania — nearly four times what CSG spent on Walsrode. Rheinmetall already has four other nitrocellulose sites globally. Its German plant is supposed to produce the full value chain, according to a 2024 press release — from projectile and fuse to explosive charge and propellant charge.

CSG and Rheinmetall’s total capital expenditures in 2024 and 2025 also reflect that gap, with Rheinmetall’s capex at nearly four times CSG’s in 2025. CSG appears to be gearing up to play catch-up, guiding to nearly triple the 2025 capex intensity in the next year.

To be clear, capex comparisons are imperfect. Rheinmetall’s higher spending might reflect predominantly other segments, as well as variables like the relatively high cost of Western European labor compared to Eastern European labor (as CSG pointed out in its comment to Hunterbrook). In other words, CSG could theoretically be getting more output per euro of investment, and it certainly could have benefited from the head start of inheriting Soviet-era factories.

But the gap — between CSG’s 205 million euros in total upgrades across four plants, plus a nitrocellulose acquisition and two joint ventures not yet operational, and Rheinmetall’s investments of billions of euros so far to expand ammunition production — implies it would be hard for CSG to catch up.

And while CSG does plan to increase capex massively, it would be with cash that CSG may not have.

In 2025, the company reported just 61 million euros in operating cash flow, despite 1.6 billion euros in operating EBIT — a 3.8% conversion rate. The entire cash from operating profit was wiped out by its working capital totaling 1.6 billion euros, up to 24% of revenue from 13% the year before. The company said it “deliberately” increased the working capital to fulfill “record order backlogs amid tight supply chain constraints and strong order inflows.”

That means, unless CSG can manage to materially reduce its working capital, funding a 638 million euro capex will almost certainly require outside capital or massive upfront payments; CSG’s debt already stood at 3 billion euros as of the end of 2025 — nearly double from the prior year. CSG said it expects working capital will stabilize in 2026, even as it highlights a 36% increase in backlog. For comparison, Rheinmetall reported a 66% cash conversion rate in 2025, and guided to at least 40% in 2026.

So, we’re back to our original question. How do we explain the gap between our estimated production revenue for CSG and CSG’s claimed 4.1 billion euros in M/L ammunition sales? Markups beyond the 27% EBITDA of the segment or non-production-related revenue like servicing can only explain a portion of that. The rest is likely coming from recommissioning.

The Big Business: “Buy a Warhead for CZK50… Sell It Abroad for CZK150“

“You buy one warhead for CZK50 and immediately sell it abroad for CZK150 or more,” Jaroslav Strnad explained to news outlet Respekt in 2018. It’s a good pitch!

Recommissioning can indeed be a profitable business; after all, it’s CSG’s founding trade — sourcing from depleted Eastern European stockpiles, refurbishing, and reselling at a markup.

Spain’s transportation minister Óscar Puente posted a document on X showing one example of CSG operating as a middleman for the importation of ammunition from India.

The problem is, there is a finite supply of ammunition to be resold. CSG described the recommissioning method in its prospectus as a “nimble and entrepreneurial approach” that allowed the company to meet urgent demand from the Ukraine war. But it also admitted there is “limited visibility on the remaining stocks of large-caliber ammunition for recommissioning available in the market in the medium-term.”

Countries like the U.S., South Korea, and India, which in the past have supplied 155 mm ammunition to NATO, are facing constraints: The U.S. is woefully behind its replenishment targets, with production ramp-up at less than half its target as of last year. South Korea cannot spare any more 155 mm shells without “risking shortages” given the North Korea threat, the Center for Strategic & International Studies wrote in 2024. India is scaling up production, but the Indian defense ministry was reportedly considering a ban on 155 mm shell exports due to escalating tension with Pakistan as of last year.

A 2024 Reuters investigation cited “a senior Czech official” putting the available global stock of large-caliber ammunition at 2 million rounds — which a research study from that year suggested would not last more than a couple of years.

And sourcing inputs from the outside is inherently risky. In May 2024, CSG CEO Michal Strnad acknowledged publicly that roughly half the components his company was purchasing from suppliers in Africa and Asia were not of sufficient quality to ship directly to Ukraine. “Every week the price is going up and there are big issues with the components,” he told the Financial Times.

An alarming complaint about quality reportedly arrived in the form of a letter from the Ukrainian to the Czech government, the German business newspaper Handelsblatt reported in 2024. “It’s every artilleryman’s nightmare: The shell doesn’t explode kilometers away in the enemy positions, but shortly after leaving the gun barrel. This is exactly what happened to soldiers in Ukraine — with artillery shells that were partly supplied as part of the Czech munitions initiative,” the article reads. The purportedly malfunctioning shells — including some apparently from Excalibur Army — “resulted in injuries to soldiers and damage to artillery systems, ” the Handelsblatt article reported, noting the cause was traced to outdated World War II–era designed detonators; five out of every 10,000 shells fired exploded prematurely.

That’s one premature detonation for every 2,000 rounds fired. For a battery firing hundreds of rounds per day in active combat, that could mean a premature detonation every few days.

Jana Černochová, who was then the Czech minister of defense, challenged the Handelsblatt article, saying any quality issues were rare. “We know that we don’t just buy new ammunition, we also buy older ones that may have potential problems,” she said.

CSG subsidiary Excalibur Army reportedly also acknowledged “minor issues” with shells.

A source close to the Ukrainian Army told Hunterbrook that the attitude among Ukrainian artillery units was that unreliable shells were still better than no shells.

In its email, CSG told Hunterbrook that third-party sourcing “was primarily linked to the initial phase” of the Czech ammunition initiative and that the company is now “predominantly supplying large-calibre ammunition from its own production.” The company offered no data to support that claim.

Production Bottleneck: Bottlenecked

Then there may be a more immediate problem: a potential bottleneck at FMG — casting a shadow over how much the company can actually vertically integrate its ammunition production.

Hunterbrook reviewed CSG’s annual reports and press releases and found no references to existing propellant charge production outside of FMG.11 The Eurenco propellant charge plant, announced earlier this year, won’t be operational until 2028 at the earliest, meaning there’s a two-year gap during which growth appears to be constrained by a single factory.

A single factory that NATO has blacklisted.

In July 2025 — six months before the IPO — NATO’s procurement agency, NSPA, suspended FMG from bidding for new contracts, according to Amsterdam-based watchdog Follow the Money and a consortium of reporters that included the French outlet La Lettre. The suspension had since been extended indefinitely as of March, according to Follow the Money. In an email response to Hunterbrook’s question on NSPA’s ban on FMG, a NATO official said, “The NATO Support and Procurement Agency (NSPA) will not comment on ongoing proceedings concerning the suspension of contractors suspected in fraudulent and corrupt activities related to NATO contracts.”

The company told Follow the Money that the suspension was “unfounded” based on an internal audit. CSG also said in a statement that it was immaterial, given the fact that the “NSPA represents one of many FMG customers” and that FMG can still sell to NATO member states, even if blocked from selling directly to NATO. When asked about the suspension by Hunterbrook, CSG said: “To the best of our knowledge the situation regarding FMG is not related to any wrongdoing by the company, but to an ongoing internal investigation within NSPA concerning one of its officials.”

CSG also characterized the FMG suspension as “a temporary and procedural measure” with “no material impact on the group’s business performance,” and argued that its integrated manufacturing model — in which entities like FMG supply components internally — is “a core strength of our industrial model, not a concentration risk.”

But while NSPA itself may not be a major direct customer for FMG, the suspension at least raises the question of how NATO will treat products sold by other CSG subsidiaries that incorporate FMG’s propellant charges. After all, FMG’s 2024 accounts show 85% of its revenue went to intragroup customers. If NATO’s position is that sales from those entities are also suspended due to the incorporation of FMG products, the effect of the suspension on CSG as a group could be more pronounced.

The Price of Urgency

A separate question for CSG is how much the current revenue reflects the premium the company charges on wartime shortage, and how much longer those conditions will last.

One of the key pipelines for CSG’s delivery to Ukraine, the Czech ammunition initiative, is facing funding shortfalls. Launched in 2024 to pool funding from various Western countries to procure ammunition from around the world through intermediaries like CSG, the initiative had delivered 4.4 million large-caliber rounds to Ukraine as of February, Czech President Petr Pavel told Czech news outlet Odkryto.cz. However, the new Czech Prime Minister Andrej Babiš — notably a far-right populist who has been labeled pro-Russia — has publicly said that despite his commitment to keeping the Czech initiative alive, he will no longer finance it. He also committed to scrutinizing procurement deals and said he “was uncomfortable with individuals earning excessive profits from the war,” according to the publication Liga.

Other key members of the initiative like Germany, the Netherlands, and Denmark are also drawing back support. As of February, the initiative had raised only 1.4 billion euros out of the 5 billion euros targeted for 2026 — according to a senior NATO official who spoke to Reuters.

The initiative has faced numerous criticisms around cost, quality, and delay. Investigative journalists have accused the Czech intermediaries of charging margins at least four times higher than what Ukrainian arms brokers charge for comparable procurement. Critics also questioned why the initiative excludes other major defense suppliers like Rheinmetall and Norway’s Nammo.

Documents obtained by the Radio Free Europe/Radio Liberty investigative unit Schemes, for example, show that in February 2024, CSG subsidiary Excalibur Army offered 155 mm artillery shells to the Czech government at 3,200 euros per unit. According to the investigation, “Around the same time, a Turkish manufacturer offered to sell seemingly similar M107 shells to Excalibur Army for 2,500 euros each” — a potential markup of roughly 30%.

CSG has publicly denied profiteering, calling the accusations “attacks based on half-truths and information taken out of the overall context.”

But the accusations are hardly new or limited to ammunition. In January 2023, the Kyiv Independent reported on a deal in which a broker earned 2 million euros — roughly 30% of the 6.8 million euro contract value — on a sale of 12,500 grenades destined for Ukraine from CSG’s export agency, Excalibur International, to a Dutch intermediary during the hectic days of the early invasion. Multiple European arms dealers told the Kyiv Independent the normal commission is between 5% and 10%.

Moreover, while competitors like Rheinmetall have a large backlog of long-term government contracts, CSG often charges customers the “spot” price for munitions, according to defense investors who spoke with Hunterbrook. In the near-term, this is a boon to margins, as CSG can essentially pick its price — like New York City bodegas selling umbrellas in the rain.

The problem with this business model is that it relies heavily on a continued shortage caused by perpetual conflict.

CSG’s underlying business, meanwhile, offers little margin of safety. Medium- and large-caliber ammunition delivered nearly two-thirds of CSG’s 2025 revenue, and the company’s flat-margin guidance for 2026 depends on that subsegment holding up. Land Systems, a subsegment responsible for 16% of revenue last year, is “expected to expand moderately.” The company expects “low to moderate margin improvement” in the Ammo+, responsible for 21% of revenue last year but saw its adjusted operating EBIT fall 58% year on year on a like-for-like basis on the back of what the company described as weaker commercial ammunition demand, rising input costs, and tariff headwinds.

So, this is the business reality. CSG appears to resemble the Cold War inventory-liquidation business it once was, with a manufacturing side business that seems much smaller than implied. The premium it charges is a wartime premium. The inventory it refurbishes may be running out. Its in-house manufacturer of a critical ammunition component is under indefinite NATO suspension. Its centerpiece growth contract depends on the participation of countries that say they hadn’t joined. And the cash it would need to close the capex gap with Rheinmetall might be cash it can’t generate.

Which brings us back to the IPO and the question that remains: How did CSG sell this story to investors in the first place?

The Operator

Part of the answer, again, may begin with what CSG left out.

Within weeks of the IPO, investigative journalists across Central Europe had identified serious omissions from the prospectus — including a put option exercised by a minority shareholder, Seznam Zprávy reported.

Hunterbrook dug deeper. The findings pointed to a company that appears to be concealing not just individual facts, but deeper aspects of its business — a sprawling network of intercompany transactions that obscure the way money flows and who really controls what.

And it seems to follow a pattern: Major milestones — from its first factory acquisitions to its largest government contracts — have often been accompanied by allegations of political favoritism, questionable business tactics, or conflicts of interest.

Undisclosed “Minority” Shareholders

CSG’s prospectus acknowledges, in generic legal language, that some subsidiaries have minority shareholders whose interests may differ from the company’s. It does not name all of them.

It also does not mention that one of them had already triggered a mechanism to force CSG to buy his stake — three days before the IPO. That shareholder is Petr Kratochvíl, according to a March 19 report from Seznam Zprávy.

Kratochvíl is not a passive investor. He ran Excalibur Army — CSG’s founding entity and most important subsidiary — for nearly 20 years, from 2005 until 2024. Until recently, he was also the chairman of the board of CSG’s Land System, the holding company that sits between CSG Group and Excalibur Army, company filings with the Czech registry show. He built “key parts” of the business alongside the Strnads, he said in an interview, and stood beside them through legal troubles, including a personal prosecution for purchasing anti-personnel mines. (He was later acquitted.)

Kratochvíl wants to sell his 10% stake in CSG Land Systems, he told Seznam Zprávy. He also has an 8.9% stake in MSM Group — another important subsidiary controlling CSG’s ammunition businesses.

On January 20 — the week of the IPO — Kratochvíl exercised a put option forcing CSG to buy his CSG Land Systems stake. His price: 1.4 billion euros, Seznam Zprávy said, citing an interview with Kratochvíl. CSG’s counteroffer: one-tenth of that.

Regardless of whose valuation is more accurate, the dispute is meaningful. While CSG Land Systems itself appears to be primarily a holding company, one of the assets it holds is Excalibur Army. In 2024, Excalibur Army reported 2.6 billion euros in revenue — two-thirds of the entire CSG group’s revenue for the year. Excalibur Army also held nearly 600 million euros in cash at that time.

Moreover, Kratochvil doesn’t just own 10% of the entity sitting atop CSG’s most important subsidiary. He also has “extraordinary rights.” His Class B shares carry four votes each, giving him veto power over changes to capital structure, articles of association, and corporate reorganizations, Kratochvíl told Seznam Zpravy in an interview.12 He also claims to have the right to a guaranteed buyback at market price.

In practical terms: Kratochvíl asserts he can block operational decisions at the entity that appears to control a subsidiary responsible for 578 million euros in cash and two-thirds of CSG’s revenue as of 2024.

The prospectus does not even name Kratochvíl. But if Kratochvíl’s valuation of his stake prevails, that’s a 1.4 billion euro liability that CSG may not have the cash on its balance sheet to pay.

In a March 2026 interview with Seznam Zprávy, Kratochvíl said he was promised there would be no IPO.

CSG told Hunterbrook in an email that the put option exists but that the auditors found “no liability (not even a contingent liability) that would need to be presented” in the annual report. Asked to confirm whether Kratochvíl exercised the put option days before the IPO, CSG replied that outside counsel confirmed Kratochvíl “did not effectively exercise his right regarding the minority stake before the IPO.” According to Seznam Zpravy, Kratochvíl delivered the notification twice on January 20: physically, via a courier to CSG headquarters, as well as digitally.

The omission of the details of Kratochvíl’s exercise of the put option in the prospectus doesn’t appear to be an anomaly. As Hunterbrook dug deeper, other omissions surfaced, revealing a pattern: a series of related-party transactions that seem to move cash to company insiders’ orbit — and sometimes, to their political partners — and a web of affiliated entities and shell companies that seem to play a questionable role in CSG’s acquisition practices.

Hidden Liabilities: Money Leaking to Strnad’s Orbit

Take, for example, the 275 million euros that Strnad’s personal vehicle likely owes CSG for which no evidence has been collected.

Ahead of the IPO, CSG transferred two dozen subsidiaries to Ytara SPV, an entity personally owned by Michal Strnad.13 The prospectus mentioned that non-core assets would be carved out but didn’t name Ytara as the recipient. The 2025 annual report clarifies that: 275 million euros worth of related-party receivables owed by Ytara from a disposal of subsidiaries.

Along with the disposal, 63.3 million euros in cash held by those entities left the group. CSG retrieved just a fraction of that, leaving CSG with negative cash proceeds of 56.9 million euros, the 2025 annual report further reveals.

Ytara SPV has since moved above CSG FIN in the corporate chain, according to CSG’s 2025 annual statement — meaning Strnad now controls the listed entity through a vehicle that’s absorbing assets from it.

Ytara’s articles of association as of December 4, 2025, filed with the Czech registry, give it broad authority, including to provide financial, trade, consulting, IT, and real estate assistance — notable capabilities for a holding company sitting between Strnad and the rest of the group that can facilitate upward cash movement without public disclosure.

In an email to Hunterbrook, CSG described the pre-IPO restructuring as “standard,” explaining the carved-out entities — which included “a Czech manufacturer of railway braking systems” and “an Italian producer of luxury sporting firearms” — were non-core. The company said “all related party transactions (including the 275 million euroreceivable from carved-out business) will be fully settled in cash.” Asked why CSG did not disclose that Ytara was owned by Strnad in the prospectus, the company said the information is publicly available in the Czech Commercial Register and reiterated that “all relevant related-party transactions were properly disclosed” in the prospectus.

“Disposing” of subsidiaries by moving them to the CEO’s personal vehicle isn’t the only way the company’s assets seem to flow to company insiders. As of 2024, Excalibur Army, perhaps CSG’s richest subsidiary, reported CZK 118 million (~4.7 million euros) in “other short-term receivables,” which consisted principally of receivables from Jaroslav Strnad, the company’s founder and current CEO’s father — a detail not disclosed as a related-party transaction at the group level. Excalibur Army also wrote off CZK 327 million (~13 million euros) in “time-barred receivables” — debts so old the company gave up collecting — raising the question of whether the same fate awaits the company’s claim on the senior Strnad.

For its part, CSG dismissed the 13 million euro write-off of time-barred receivables at Excalibur Army as an accounting necessity, saying it “does not represent the incurred losses but rather only theoretic risk.” The company did not address the 4.7 million euro receivable from Jaroslav Strnad.

Excalibur Army’s 2024 filing also lists a waste disposal company as a contract counterparty, with contracts dating back to 2015 on services including office rental and sewage operation.14 That company’s own filing shows 90% of the company was moved out of CSG into Jaroslav Strnad’s personal company, CE Industries, in 2019. CSG did not disclose this as a related-party transaction in its prospectus.

In another example of funds moving into Strnad’s orbit, the cash trail could be extended even further — to political partners. After leaving CSG in 2022, former Vice-Chairman Miroslav Dorňák acquired, at an undisclosed price, a real estate company to which CSG had previously transferred CSG’s real estate assets, also for an undisclosed price. The company then bought the building into which CSG moved its headquarters, according to multiple Czech press reports and registry filings.15 Essentially, CSG now pays rent to its former vice-chairman’s company.

In an email to Hunterbrook, CSG called the lease “a standard landlord–tenant relationship” and said it does “not comment on the business activities of third parties.”

A Czech registry filing shows Dorňák also owns a Czech crowdfunding platform called FinGood, which shares the same address as CSG’s headquarters.16 According to a March 2026 investigation by Seznam Zprávy, FinGood paid millions to Slovak Defense Minister Kaliňák’s legal consultancy firm, Kallan Legal, citing a former CEO and co-owner of Kallan and the current FinGood CEO who confirmed the payments. Seznam Zprávy alleged that CSG, including Michal Strnad personally, was involved in the management of FinGood, citing FinGood employees and internal emails obtained by the outlet.

Kaliňák, as noted earlier, is behind the recent 58 billion euro ammunition deal with CSG.

The Shadows Behind the Empire

CSG’s post-IPO disclosures may be lacking. But they’re less surprising when set against the company’s history — a three-decade record of political patronage, questionable acquisitions, and ties to figures that would give any compliance officer pause.

The company’s origin story is inseparable from Czech state connections. CSG was founded by the CEO’s father, Jaroslav Strnad, in the mid-1990s as a scrap dealer, collecting Societ-era military assets liquidated after the Cold War, refurbishing them, and reselling them at a markup. Jaroslav Strnad’s father-in-law served as deputy head of the material and technical support office at the Czech defense ministry — a role that Czech media reported helped steer Strnad toward his surplus military equipment deals.

Numerous accounts of CSG’s dealings highlight its ties to Miloš Zeman, who served as the president of the Czech Republic from 2013 to 2023. CSG’s domestic growth accelerated after Zeman took office in 2013.

Ahead of Zeman’s 2018 re-election, Jaroslav Strnad became Zeman’s largest campaign donor. Two Strnad-linked firms, DAKO-CZ and Composite Components, together accounted for two-thirds of all campaign contributions, according to OCCRP. Seznam Zprávy reported that his son, Michal Strnad, used a car pawn company to channel hidden cash to Zeman’s campaign: Internal documents recorded a CZK 950,000 transfer as directed by Strnad personally, routed through an intermediary entity called Car Service Group CZ.

Milestones in CSG’s growth during Zeman’s tenure, including major acquisitions and contracts, were often flagged by journalists and regulators as irregular. For example, an investigation by the Ukrainian outlet Gordon found the sale of an armored vehicle repair and production factory to the then-named Excalibur Army in 2013 “puzzling,” noting the state had spent substantial sums of money reconstructing it; a source told Euro that Strnad bought the plant “for a really good price.” Other coverage found CSG had received major orders from the government without a competitive tender for one truck contract in 2017.

Some investigations even led to formal arrests, only for the cases to be dismissed later. In 2015, defense ministry and army personnel were charged with transferring surplus Czech military property — specifically tank engines and spare parts — to Excalibur at below-market prices. The case later ended in acquittal, with Česká Justice reporting a “complete failure of the investigative bodies.” Notably, two of the acquitted defense ministry officials ended up working for Strnad.

Other cases are ongoing.

Marián Goga, former president and 10% owner of MSM Group — the parent company controlling CSG’s ammunition businesses, is currently facing criminal charges in Slovakia for “bribery and money laundering.17”

Prosecutors alleged Goga, in his former capacity as the head of CSG subsidiary MSM Martin, steered a public tender for recovery tanks through a bribery scheme in which MSM Martin paid 15 months of warehouse rent in a single day to an intermediary, who then transferred two below-market apartments to the son of the state strategic reserves chief. The intermediary reportedly committed suicide after the transfers became public.

MSM Martin, which is a codefendant, could face serious consequences if found guilty, CSG disclosed in the prospectus. Penalties could include “the forfeiture of assets,” a “financial penalty,” and “a prohibition on receiving EU funds” — in light of which MSM Martin “wound down all business activities and is currently a dormant entity with no significant assets.” Ahead of the IPO, CSG reorganized its structure to remove MSM Martin and Goga’s shares in MSM Group. It’s unclear how Goga was compensated.

CSG told Hunterbrook in an email that Goga “was indeed a former shareholder and executive director of MSM Martin” and was a shareholder of MSM Group “until January 2026, i.e. prior to the IPO.” But the company also confirmed that Goga “currently serves on the supervisory board of MSM Group” — a fact the prospectus omitted. CSG said Goga “has played a positive role in the development of the business and continues to support the company in this supervisory capacity,” and that the criminal proceedings “remain ongoing” with “no material developments.”

The pattern of strange deals extends beyond Europe. CSG’s announced contracts in Southeast Asia — a region the company has cited as a growth market — often flow to a single country, Indonesia, and substantially through a single intermediary. Since 2015, CSG subsidiaries have announced roughly $3.7 billion to $4 billion in Southeast Asian defense contracts, including an over 500 million euro KHAN ballistic missile package signed in November 2022, a $300 million Patriot II armored vehicle deal announced 13 days after the January 2026 IPO, and a $2.5 billion air-defense megapackage disclosed in April 2026.

Some of these deals are, themselves, reportedly funded by the Czech Export Bank — a state-owned institution backed by Czech taxpayers. The structure is circular: A Czech state bank lends to Jakarta so Jakarta can pay a Czech private company, and the resulting revenue inflates the backlog that helped underpin CSG’s 25 billion euro IPO valuation.

Then there is the question of where CSG’s original capital came from. Foreign Policy reported in 2018 that one of Jaroslav Strnad’s key investors was Alexei Belyaev, described as connected to Vladimir Yakunin, the sanctioned former head of Russian Railways and Putin associate. Corporate filings confirm that Belyaev and his partner Michal Lazar were co-investors in DAKO-CZ, a CSG subsidiary. Belyaev joined DAKO-CZ’s supervisory board in 2016.

A WikiLeaks-published U.S. Embassy cable identified Lazar as suspected of organized crime connections; the cable noted that Slovak police appeared “clearly uncomfortable” discussing him. “Strnad’s partner is widely reported to have a business association with senior Russian intelligence officers,” The American Spectator wrote.

Belyaev and Lazar’s company, Optifin Invest, reportedly became a major co-investor in CSG entities. A Czech security analyst told Ukrainian news outlet Gordon in 2020 that this structure was designed so that “Belyaev was the one who was coordinating the whole strategy from the very beginning” of CSG’s expansion.

Journalist Jaroslav Spurný told Gordon, citing internal sources, that “substantial amounts involving Russian financing” flowed through Creditas bank — which shared a Prague office building with CSG and had connections to First Czech-Russian Bank, described by Czech intelligence as an FSB, or Russian intelligence, instrument. CSG had a separate revolving loan from Creditas worth up to CZK 330 million, according to Czech news organization Euro. A former Czech intelligence chief told Gordon that CSG is “in bad financial state and these debts are all owed to the Russians.” CSG has denied the Russian financing allegations, and the Belyaev/Lazar involvement formally ended with a 2021 DAKO-CZ stake sale. CSG told Hunterbrook it “immediately ceased all activities in Russia” in 2022.

But the security concerns over CSG’s ties to Russia were apparently serious enough that a CSG subsidiary had issues receiving a security clearance from the Czech National Security Office — effectively barring CSG from supplying the Czech military. Defense Minister Karla Šlechtová confirmed: “If someone does not have the requisite security clearance, they do not meet the formal criteria to receive an order, and thus cannot receive it. This is exactly what happened with the Czechoslovak Group.”

The company did not appeal the decision in court — a choice a former head of Czech intelligence noted suggested CSG had “real problems” it did not want aired publicly, according to Gordon. Strnad “transferred all of his companies to his 25-year-old son Michal.” CSG denied any connection between the clearance issue and the transfer.

Jaroslav Strnad admitted as much himself, though he put it a bit more mildly: “I understand that my name is connected to a number of people and events that could negatively impact the company and its reputation,” he told Respekt in 2018.

These concerns resurfaced in 2024 — this time in the U.S. — when multiple Republican senators including JD Vance raised objections to CFIUS about CSG’s acquisition of Vista Outdoor’s ammunition unit, The Kinetic Group.

“We cannot afford for America’s supply of weapons to fall into the wrong hands,” Vance wrote in a January 2024 letter to U.S. Treasury Secretary Janet Yellen, requesting a security review of the sale. “Until it can be proven that this transaction will not jeopardize our national security, I respectfully urge you to deny the sale of Vista Outdoor’s Sporting Products business to the Czechoslovak Group.”

Vance’s letter raised concerns about ties to Putin’s inner circle, including the company’s sponsorship of a show in Moscow aimed at facilitating Russian access to European military technology. CFIUS ultimately approved the deal.

But CSG’s choice of personnel reinforces the pattern. In February 2024, CSG appointed Jan Hamáček as its director of external relations. Hamáček had previously served as Czech minister of the interior — and was investigated by the Czech National Center for Organized Crime after Seznam Zprávy reported that he had planned to travel to Moscow to exchange Czech silence on the 2014 Vrbětice ammunition depot explosions — which killed two Czech citizens and were attributed to GRU agents — for, among other things, a Biden-Putin summit in Prague.

Police ultimately dropped the case because Hamáček did not end up taking the trip to Moscow. But a company that hires an official investigated for involvement in a Russian intelligence operation to run its government relations is making a statement about its priorities — whether it intends to or not.

It’s an especially odd statement, of course, from a company whose riches — however ephemeral they may be — are a direct result of Russia’s invasion of Ukraine.

Read the full version of the investigation, including all of the footnotes, on our website at hntrbrk.com/csg.

Authors

Jenny Ahn joined Hunterbrook after serving many years as a senior analyst in the US government. She is a seasoned geopolitical expert with a particular focus on the Asia-Pacific and has diverse overseas experience. She has an M.A. in International Affairs from Yale and a B.S. in International Relations from Stanford. Jenny is based in Virginia.

Till Daldrup is an investigative journalist who joined Hunterbrook from The Wall Street Journal, where he focused on open-source investigations and content verification. In 2023, he was part of a team of reporters who won a Gerald Loeb Award for an investigation that revealed how Russia is stealing grain from occupied parts of Ukraine. He has an M.A. in Journalism from New York University and a B.S. in Social Sciences from University of Cologne. He’s also an alum of the Cologne School of Journalism (Kölner Journalistenschule).

Blake Spendley joined Hunterbrook from the Center for Naval Analyses (CNA), where he led investigations as a Research Specialist for the Marine Corps and US Navy. He built and owns the leading open-source intelligence (OSINT) account on X/Twitter, called @OSINTTechnical (over 1 million followers), which also distributes Hunterbrook Media reporting. His OSINT research has been published in Bloomberg, the Wall Street Journal, and The Economist, among other top business outlets. He has a B.A. in Political Science from USC.

EDITOR

Sam Koppelman is a New York Times best-selling author who has written books with former United States Attorney General Eric Holder and former United States Acting Solicitor General Neal Katyal. He helped build Fenway Strategies into one of the preeminent strategic communications firms in the country—with side quests speechwriting for Michael Bloomberg, running the surrogate remarks operation on the Biden-Harris campaign, and co-founding Mayday, which is now one of the leading information providers on how to access reproductive health care in states with bans. Sam has published in the New York Times, Washington Post, Boston Globe, Time Magazine, and other outlets — and occasionally volunteers on a fire speech for a good cause. He has a BA in Government from Harvard, where he was named a John Harvard Scholar and wrote op-eds like “Shut Down Harvard Football,” which he tells us were great for his social life.

Graphic

Dan DeLorenzo is a creative director with 25 years reporting news through visuals. Since first joining a newsroom graphics department in 2001, he has built teams at Bloomberg News, Bridgewater Associates, and the United Nations, and published groundbreaking visual journalism at The Wall Street Journal, Associated Press, The New York Times, and Business Insider. A passion for the craft has landed him at the helm of newsroom teams, on the ground in humanitarian emergencies, and at the epicenter of the world’s largest hedge fund. He runs DGFX Studio, a creative agency serving top organizations in media, finance, and civil society with data visualization, cartography, and strategic visual intelligence. He moonlights as a professional sailor working toward a USCG captain’s license and is a certified Pilates instructor.

Hunterbrook Media publishes investigative and global reporting — with no ads or paywalls. When articles do not include Material Non-Public Information (MNPI), or “insider info,” they may be provided to our affiliate Hunterbrook Capital, an investment firm which may take financial positions based on our reporting. Subscribe here. Learn more here.

Please contact ideas@hntrbrk.com to share ideas, talent@hntrbrk.com for work opportunities, and press@hntrbrk.com for media inquiries.

LEGAL DISCLAIMER

© 2026 by Hunterbrook Media LLC. When using this website, you acknowledge and accept that such usage is solely at your own discretion and risk. Hunterbrook Media LLC, along with any associated entities, shall not be held responsible for any direct or indirect damages resulting from the use of information provided in any Hunterbrook publications. It is crucial for you to conduct your own research and seek advice from qualified financial, legal, and tax professionals before making any investment decisions based on information obtained from Hunterbrook Media LLC. The content provided by Hunterbrook Media LLC does not constitute an offer to sell, nor a solicitation of an offer to purchase any securities. Furthermore, no securities shall be offered or sold in any jurisdiction where such activities would be contrary to the local securities laws.

Hunterbrook Media LLC is not a registered investment advisor in the United States or any other jurisdiction. We strive to ensure the accuracy and reliability of the information provided, drawing on sources believed to be trustworthy. Nevertheless, this information is provided “as is” without any guarantee of accuracy, timeliness, completeness, or usefulness for any particular purpose. Hunterbrook Media LLC does not guarantee the results obtained from the use of this information. All information presented are opinions based on our analyses and are subject to change without notice, and there is no commitment from Hunterbrook Media LLC to revise or update any information or opinions contained in any report or publication contained on this website. The above content, including all information and opinions presented, is intended solely for educational and information purposes only. Hunterbrook Media LLC authorizes the redistribution of these materials, in whole or in part, provided that such redistribution is for non-commercial, informational purposes only. Redistribution must include this notice and must not alter the materials. Any commercial use, alteration, or other forms of misuse of these materials are strictly prohibited without the express written approval of Hunterbrook Media LLC. Unauthorized use, alteration, or misuse of these materials may result in legal action to enforce our rights, including but not limited to seeking injunctive relief, damages, and any other remedies available under the law.

By: Jean Wang

Editor: Sam Koppelman

Based on Hunterbrook Media’s reporting, Hunterbrook Capital is long $DDOG and short a basket of comparable securities at the time of publication. Positions may change at any time. See full disclosures on our website.

Datadog told Wall Street it had landed an eight-figure deal with “one of the largest AI foundational model companies” — but didn’t say which one. Hunterbrook found Datadog’s monitoring software embedded across Claude sessions and products, with public bug reports dating back to January and a leaked source map suggesting a much deeper rollout may be ahead. If the biggest new logo in Datadog’s history is who we think it is, the consumption-priced contract could scale directly with Anthropic’s growth.

Anthropic has got that dog in them.

When Datadog ($DDOG) told analysts on its fourth-quarter earnings call in February that it had landed an “eight-figure annualized” deal with “one of the largest AI foundational model companies” — among the biggest new logos in the company’s history — CEO Olivier Pomel didn’t name the customer.

Datadog’s CFO David Obstler said the company modeled the contract with “very conservative” assumptions in 2026 guidance because of the unpredictability that consumption-based pricing introduces. A handful of analysts speculated about Anthropic being the customer. Neither company confirmed, though several Datadog employees liked a LinkedIn post whose author inferred an Anthropic partnership.

Hunterbrook Media dove deep into live browser traffic from Claude.ai and public bug reports filed against its developer tools to establish what had not previously been proven:

Datadog’s big new customer indeed appears to be the fastest-growing software company of all time — not bad for a deal with consumption-based pricing.

The evidence of an Anthropic partnership comes in two distinct forms.

The first is solid, universal, and trivial to verify: Datadog’s monitoring software is deeply embedded in Claude products, watching the sessions of millions of everyday users. This is incontrovertible — though comparatively modest financially.

The second is more circumstantial but suggestive of a much larger ticket: Anthropic appears to be quietly exploring using Datadog for all its essential telemetry — in a phased, experimental way that hasn’t been publicly disclosed, dating back to at least January.

One key clue came from a March 31 leak.

None of this is direct confirmation — and Anthropic may just be testing Datadog. But the evidence… certainly seems to amount to both bark and bite.

Datadog Likely Watches All Your Claude Sessions

Big Datadog is watching.

Hunterbrook’s engineering team conducted a live test using various Anthropic products.

What we found:

After loading Claude in your browser or desktop, Anthropic quietly ships packets of data capturing that experience — from how fast the page rendered to whether a script crashed — over to Datadog. That’s called telemetry. It’s how Anthropic engineers track whether things are working or not, and collect data to improve the experience of using Claude for all of its estimated 30 million users.

What Anthropic is trusting Datadog with is not trivial. A one-second slowdown or a silent error in the Claude composer can translate directly into churn and lost revenue. You can’t fix what you can’t see, and you can’t see tens of millions of browsers without piping their experience back to a central system.

And it’s not just product analytics: Anthropic classifies Datadog Real User Monitoring (RUM) as essential — crash reports, error stack traces, and performance timings — meaning that it’s enabled by default, and enterprise customers need to actively elect to turn it off.

Datadog sells one of the most widely used RUM tools on the internet — which has helped power Datadog to its $45 billion market cap. RUM runs inside the user’s browser and ships that data regularly back to Datadog, where Anthropic’s engineers use it to find bugs and measure speed.

A live inspection of Claude.ai conducted for this story found the Datadog RUM software development kit — a small bundle of code that the website loads into the visitor’s browser — initialized with the service name claude-ai, the environment set to production, and a sample rate of 100%. That means literally every consumer session on the chat product is captured by Datadog. The same SDK is bundled inside the Claude desktop app.

Early Signs of a Wider Rollout

The second finding is harder to interpret. But, taken together with the first, paints a clearer picture.

Anthropic appears to be rolling out Datadog much deeper than RUM — using it for more types of telemetry — in a phased way that is consistent with an account in active expansion.

The clearest indicator is Claude Code, Anthropic’s developer command-line tool.

As part of a module, its source code contains a hard-coded Datadog client token and a dedicated Datadog integration, where there is capability for Claude Code to send telemetry events to Datadog. This is distinct from Datadog’s browser RUM product, which Anthropic is already using: Claude Code shipped with wirings to send analytics and telemetry to Datadog. Whether each user actually sends data to Datadog is governed by what engineers call a feature flag — an on-off switch that lets a company turn a feature on for some users and off for others.

The switch is named tengu_log_datadog_events. A user-submitted configuration dump captured the switch turned on, paired with settings that bundle telemetry into batches before sending — the architecture of a production telemetry pipeline mid-rollout, not a one-off debug log. The flag is managed through a third-party A/B testing tool called GrowthBook, the standard mechanism companies use to roll new features out to small subsets of users at a time.

Public bug reports show the rollout is real, even if its breadth is unclear.

As early as January 6, a user noticed Claude Code calling datadoghq.com every few seconds throughout each session and complained that the calls were undisclosed; the issue was closed as “not planned.” On March 13, another user reported that during a memory problem, Claude Code generated 1.2 million error log lines in roughly 80 minutes — a loop of failed attempts to deliver telemetry to https://http-intake.logs.us5.datadoghq.com/api/v2/logs, Datadog’s log-ingestion endpoint.

A March 18 outage report showed the same Datadog endpoint failing in lockstep with api.anthropic.com, suggesting the connection is part of the production critical path for users who have it turned on. Anthropic enterprise feature Cowork is similarly instrumented: A March 26 bug report references its OpenTelemetry exporter attempting to ship events to https://otlp.us5.datadoghq.com/v1/logs, with event types named user_prompt, api_request, api_error, tool_result and tool_decision. This is Datadog’s instance of the OpenTelemetry protocol, which signals that Anthropic is exploring using Datadog’s infrastructure for the telemetry surface of an LLM application.

The most direct evidence emerged on March 31, when Anthropic seemingly accidentally shipped a sourcemap file inside a software update to Claude Code — a debug file that maps the product’s compiled code back to its original internal layout, effectively a blueprint of the source tree. Several independent researchers — including Karan Prasad and Engineer’s Codex — had pulled the file apart and posted their findings. Among the files they identified was a path indicating a dedicated Datadog integration module sitting alongside Anthropic’s other analytics services.

A Chinese explanation of how this works was posted to GitHub. Here is a data flow diagram created by Hunterbrook summarizing the findings:

Why Occam’s Razor Is Anthropic

Datadog’s earnings-call description matches Anthropic’s known stack. Pomel told analysts the customer — the largest “new logo” in Datadog’s history — was consolidating onto Datadog from a fragmented mix of “more than five open-source, commercial, hyperscaler, and in-house” tools. That description fits Anthropic. Grafana Labs, a prominent open-source observability vendor, publicly names Anthropic as a customer in its September 2025 annual recurring revenue announcement — supplying exactly the kind of open-source observability tooling Pomel said the customer was migrating away from.

The “new logo” qualifier — Wall Street shorthand for a brand-new enterprise relationship, not the expansion of an existing one — narrows the field further. OpenAI, long flagged by analysts as Datadog’s largest single customer, is by definition not a new logo.

Among AI labs operating at the scale implied by an eight-figure annual contract, that leaves Anthropic as perhaps the most plausible candidate.

The two companies have also been moving toward each other publicly for more than a year. At Datadog’s DASH 2024 conference, Anthropic President Daniela Amodei announced a joint product integration. Datadog now publishes a dedicated Anthropic monitoring product and a Cloud Cost Management integration for tracking Anthropic spend, and is itself a named member of Anthropic’s Claude Design research beta.

The relationship cuts against a specific bear thesis on Datadog. Some investors have argued that AI-native companies will eventually “vibe code” their way out of buying observability software, building in-house tools instead. (A July 2025 Tegus interview with an Anthropic employee propagated this thesis, claiming the company “didn’t even touch Datadog ever.” That was, of course, before the announced partnership.)

Anthropic, the frontier-AI customer most capable of doing exactly that kind of vibe coding — and, according to Pomel, already running in-house tools before the deal — seemingly chose instead to consolidate at least five tools onto Datadog.

It’s living proof that there are plenty of “pick-and-shovel” plays left in software: OpenAI and Anthropic aren’t necessarily going to vibe code their own observability stacks — at least not now. They have better things to do. They need something robust and rigorous, so they tapped Datadog.

Observability software is like soy sauce. Whereas a Japanese restaurant specializing in making a few complicated dishes–say, Netflix or Salesforce–might make their soy sauce in house to suit their unique needs, Chinese restaurants generally buy their soy sauce from three trusted brands. Much like a rapidly-growing AI company like Anthropic, Chinese restaurants tend to have huge menus and wide offerings; the last thing the kitchen should worry about is making its own soy sauce.

But you can make a lot of money selling soy sauce to Chinese restaurants!

The OpenAI contract, alone, represents “4-6% of DDOG’s revenue run-rate,” wrote Wells Fargo last year, estimating it would total roughly $200 million annualized.

Because the Anthropic contract appears to be consumption-priced, Datadog’s revenue from that deal increases with every increment of Anthropic’s usage curve.

Which, right now, looks like this:

Author

Jean Wang is a machine learning engineer and data scientist at Hunterbrook Media. Previously, Jean worked on the alternative credit and fixed income desk at Magnetar Capital, studied Math at Yale University, and served as an editor for the Yale Daily News.

Editor

Sam Koppelman is a New York Times best-selling author who has written books with former United States Attorney General Eric Holder and former United States Acting Solicitor General Neal Katyal. Sam has published in the New York Times, Washington Post, Boston Globe, Time Magazine, and other outlets. He has a BA in Government from Harvard, where he was named a John Harvard Scholar and wrote op-eds like “Shut Down Harvard Football,” which he tells us were great for his social life.

Hunterbrook Media publishes investigative and global reporting — with no ads or paywalls. When articles do not include Material Non-Public Information (MNPI), or “insider info,” they may be provided to our affiliate Hunterbrook Capital, an investment firm which may take financial positions based on our reporting. Subscribe here. Learn more here.

Please contact ideas@hntrbrk.com to share ideas, talent@hntrbrk.com for work opportunities, and press@hntrbrk.com for media inquiries.

LEGAL DISCLAIMER

© 2026 by Hunterbrook Media LLC. When using this website, you acknowledge and accept that such usage is solely at your own discretion and risk. Hunterbrook Media LLC, along with any associated entities, shall not be held responsible for any direct or indirect damages resulting from the use of information provided in any Hunterbrook publications. It is crucial for you to conduct your own research and seek advice from qualified financial, legal, and tax professionals before making any investment decisions based on information obtained from Hunterbrook Media LLC. The content provided by Hunterbrook Media LLC does not constitute an offer to sell, nor a solicitation of an offer to purchase any securities. Furthermore, no securities shall be offered or sold in any jurisdiction where such activities would be contrary to the local securities laws.

Hunterbrook Media LLC is not a registered investment advisor in the United States or any other jurisdiction. We strive to ensure the accuracy and reliability of the information provided, drawing on sources believed to be trustworthy. Nevertheless, this information is provided “as is” without any guarantee of accuracy, timeliness, completeness, or usefulness for any particular purpose. Hunterbrook Media LLC does not guarantee the results obtained from the use of this information. All information presented are opinions based on our analyses and are subject to change without notice, and there is no commitment from Hunterbrook Media LLC to revise or update any information or opinions contained in any report or publication contained on this website. The above content, including all information and opinions presented, is intended solely for educational and information purposes only. Hunterbrook Media LLC authorizes the redistribution of these materials, in whole or in part, provided that such redistribution is for non-commercial, informational purposes only. Redistribution must include this notice and must not alter the materials. Any commercial use, alteration, or other forms of misuse of these materials are strictly prohibited without the express written approval of Hunterbrook Media LLC. Unauthorized use, alteration, or misuse of these materials may result in legal action to enforce our rights, including but not limited to seeking injunctive relief, damages, and any other remedies available under the law.

By: Bethany McLean, Matthew Termine, JD Jean-Jacques

Editor: Jim Impoco

Based on Hunterbrook Media’s reporting, Hunterbrook Capital is short $HLNE and long a basket of comparable securities at the time of publication. Positions may change at any time. See full disclosures on our website.

“I literally think all the marks are wrong.”

A few weeks ago, Apollo’s John Zito found himself unexpectedly in the headlines for brutally honest remarks he thought were off the record about private market valuations.

As the co-president of Apollo’s asset management arm, Zito is a titan in the private credit and private equity markets. And his grim assessment added fuel to the fire that is racing through an industry whose marks are being more heavily scrutinized amid a “SaaSpocalypse” triggered by AI disruption.

The funny thing about the discourse, as Zito pointed out, is that so far, it has mostly focused on private credit, rather than private equity. This ignores a basic rule of finance: In any capital structure, debt is the safety net, and equity is the high-wire act.

Which is to say: If people are scared of the debt, they should be terrified of the equity. “I can’t compute, but I’m the dumb guy,” Zito said. “I don’t understand.”

Zito isn’t dumb.

And if the cracks he identified are indeed signs of a larger structural failure — as JPMorgan’s Jamie Dimon and former Goldman Sachs CEO Lloyd Blankfein have also signaled — one place they might manifest is private markets colossus Hamilton Lane ($HLNE).

This week, the private equity firm — which sells to retail investors — recorded its largest net outflow in recent history with investors pulling an estimated $172 million from the fund in March alone. Redemptions like these could create a feedback loop, through which the true value of Hamilton Lane’s assets, independent of its current mark, could be revealed.

A Gatekeeper to Private Equity

Headquartered in Conshohocken, Pennsylvania, Hamilton Lane is a roughly $4 billion market cap investment powerhouse that has spent three decades evolving from a niche research shop into a gatekeeper for the private markets. As of last year, it managed or advised on $1 trillion of assets, and it positions itself as the bridge between institutional-grade private equity and a new, hungry class of retail investors.

In the last five years, Hamilton Lane has made its business, which mostly comes from the assets it manages, increasingly dependent on buying assets known as “secondaries.” This is particularly true of the almost $6 billion Hamilton Lane Private Assets Fund, or PAF. In other words, Hamilton Lane has increasingly bought stakes in private equity portfolios from sellers who want to unload them.

This has looked like a fantastic business in part because of a specific accounting quirk first scrutinized by The Wall Street Journal: Hamilton Lane buys the assets at what it considers a discount and can then promptly mark these assets back up to the previous valuation, even if there were no buyers at that old price. We’ll refer to these as “day-one markups.”

Last year, Hamilton Lane brewed another accounting potion, changing the way the firm is compensated. It used to be paid incentive fees when gains were realized — meaning when the underlying assets were actually sold. Now, it can collect its fees regularly based simply on increasing its own marks on its investments, including those day-one markups. The change allowed Hamilton Lane to pull forward fees that may have taken years to accrue. Thanks to the new fee structure, Hamilton Lane took in $58 million of incentive fees from the Private Assets Fund in the year that ended in March 2025.

In a statement to Hunterbrook addressing the issue, the company said that “because semi-liquid funds typically reinvest proceeds rather than distribute them, traditional carry calculations based on distributions are less appropriate, making NAV-based carry a better fit.”

The firm made another change at the same time, one that has gone largely unremarked: It began excluding stock-based compensation from its fee-related earnings calculation, just like competitors like Blackstone ($BX), Apollo ($APO), and Carlyle ($CG) do. The combined effect of the two changes — adding performance fees based on unrealized gains as a revenue stream and stripping out stock-based compensation from expenses — moved Hamilton Lane’s reported fee-related earnings margin (which is an important component in how Wall Street values firms like Hamilton Lane) from roughly 32% to approximately 59% for the nine months ending in December 2025.1

The decision to exclude stock-based compensation isn’t all that unusual — but the timing is worth noting. About six months before Hamilton Lane changed its calculation of the metric, co-CEOs Erik Hirsch and Juan Delgado-Moreira each received 544,000 restricted shares — a grant valued at roughly $71 million apiece, with vesting tied to stock price targets.

While previous reporting by both The Wall Street Journal and the Financial Times identified the existence of the day-one markups and the firm’s new fee structure, our reporting goes a step further to reveal the true depth of the distortion. An analysis conducted by Hunterbrook shows that about a third of the total gains in valuations across PAF’s portfolio were largely generated by these accounting uplifts, rather than any actual appreciation of the companies owned.2 (In a statement to Huntebrook, Hamilton Lane said that more than 70% of secondary returns in the fund are not a result of the day-one markups, roughly confirming our estimate that one third of the returns is attributable.)

And the change to the way Hamilton Lane collects fees explains more than all of its recent fee-related earnings growth. In other words, without the change to collecting fees along the way, rather than growing fee-related earnings by 37%, earnings actually would have gone down over the last nine months of 2025.3

In good times, Hamilton Lane’s accounting alchemy can seem infinitely repeatable: Raise money, buy assets, mark up assets to charge performance fees, report high returns, use those high returns to raise more money from retail investors, use that money to buy more assets, and on and on.

But when investors begin to look behind the curtain, they may notice that some of their returns exist only in the realm of illusion — and that they are paying Hamilton Lane fees for those fantastical returns.

If the enchanted flywheel breaks, and Hamilton Lane can’t keep raising billions of dollars to buy (and then immediately mark up) new assets, the magic can start moving in reverse: Redemptions lead to forced asset sales lead to markdowns lead to pressure on reported returns lead to stalled fundraising, etc.

The only question, for now, is whether investors open that Pandora’s Box.

This is, oddly enough, an analogy Hamilton Lane has wholeheartedly embraced, albeit in a slightly different context.

In a recent 2026 market overview presentation that is an exercise in mixed metaphors and potentially unintended messages, Hamilton Lane’s former CEO, Mario Giannini, who still serves as executive co-chairman and on the investment committees of its various funds, dressed up as Gandalf from The Lord of the Rings. “What do you think Pandora imagined the moment she opened that box?” he asked. “A box that shouldn’t be opened is opened anyway and out pours … what exactly? Depends on who you ask.” He said investors should think of Hamilton Lane as the “Gandalf of the private markets Middle Earth,” sent to “guide and inspire” them on their journey.

The problem for Hamilton Lane itself is that in this business, eventually, reality has a way of being a Morgoth. Just like with private credit funds, investors are starting to worry about private equity, too — and could begin to demand their money back from PAF. Hamilton Lane’s stock is already down 50% from its peak amid this fear. And there are reasons to believe that it could fall much, much further.

Asked repeatedly about these issues, Hamilton Lane did not initially respond over a period of days; then, once receipt was confirmed, the company made legal threats to a Hunterbrook reporter, and, finally, provided answers to our questions. “The way we calculate fees is in fact the industry standard, and you can find examples of numerous other funds, including ones with strong secondary exposure,” said a spokesperson. Indeed, other funds, like Blackstone, Apollo, and Carlyle, collect performance fees that aren’t dependent on selling underlying investments.

The statement continued: “While the discounts in secondaries are intended to be additive to performance, in accordance with U.S. accounting standards, our analysis shows that the vast majority of long-term performance comes from underlying growth of the companies.”

Deference to Marks

Hirsch and Delgado-Moreira took over Hamilton Lane in 2024, more than three decades after Hamilton Lane’s founding by Leslie Brun, who came to the U.S. from Haiti as a young boy.

The original idea was to provide research for institutions that wanted to invest in private equity. Brun, who is now the cofounder, chairman, and CEO of Ariel Alternatives, departed in 2005 and seemingly has little to do with Hamilton Lane. Over the years, the business shifted away from just advising other investors, and toward managing capital directly. By the time Hamilton Lane went public in 2017, it managed $40 billion in assets, primarily by making investments in other funds or investing directly alongside those funds.

Just after the IPO, Hamilton Lane found itself enmeshed in the Abraaj scandal, where a Dubai-based private equity firm that at its peak managed almost $14 billion unraveled due to alleged fraud. Hamilton Lane had committed clients’ money to an Abraaj fund — despite warnings about valuation problems, according to The Economist.

On September 20, 2017, an anonymous whistleblower emailed Hamilton Lane executives, warning that Abraaj was overvaluing its holdings and that deals in its pipeline were dead. The email urged Hamilton Lane to do proper diligence before committing to Abraaj’s new Fund 6. Instead, Hamilton Lane employee Tarang Katira, who ran fund investments across Europe, the Middle East, and Africa, forwarded the whistleblower’s email directly to Abraaj founder Arif Naqvi, according to Bloomberg and The National reporting at the time.

Katira wrote to Naqvi that he was sending it so Naqvi would be aware the email was “out there and clearly targeting LPs.” Despite the warning, Hamilton Lane went on to commit $100 million to Abraaj Fund 6 — a commitment that reportedly encouraged other investors to put in roughly $900 million more. (Fund 6 never closed after the scandal broke, so those specific commitments did not result in losses; Hamilton Lane’s total Abraaj exposure across funds was approximately $382 million.)

The firm eventually shook off the Abraaj taint. But the episode illuminated something about how Hamilton Lane relates to other firms’ marks: with deference. The business it has built since is, in a sense, one where that deference is no longer a bug — where accepting the fund manager’s number has become the business model.

Hamilton Lane has doubled down on two strategies, in particular, that have helped catapulted the firm into the stratosphere.

The first was the launch, in 2019, of what are called evergreen funds. Traditional private equity funds raise money at the start of their life cycle. The money is usually locked up for about a decade. An evergreen fund, on the other hand, raises money by selling shares continuously. Existing investors are allowed to withdraw specified amounts at specified times. Hamilton Lane says that “upwards of 25%–30%” of its capital comes from institutions, but this liquidity-on-demand promise helped open the floodgates to retail investors.

In its recent presentation, Hamilton Lane’s Beth Nardi said that its evergreen funds are “really shaking things up.” The reason is that they are so easy for retail investors to use. “Prior to evergreens, investing in the private markets was, let’s be honest, a hassle,” added Brian Gildea, who runs the firm’s evergreen business.

Since 2020, the industry-wide evergreen business has more than tripled, to roughly $700 billion in global assets as of 2024. Bigger firms, like Blackstone, Apollo, and the ever-controversial Blue Owl ($OWL), are in the game, too, because everyone wants to get their hands on retail investors’ money — not to mention collect permanent fees, instead of having to launch new funds in order to keep the fees coming in.

Today, Hamilton Lane manages $17 billion of assets in various evergreen products, up from almost nothing in 2020 — and up 70% in just 2025. None of its competitors are as exposed to the retail business. Goldman Sach’s analysis shows that Hamilton Lane derives 34% of its management and incentive fees from retail and evergreen vehicles, versus 24% at Blackstone, 23% at Blue Owl, and single digits at Carlyle and Brookfield ($BAM).

The second big move was plowing all that retail money into the secondary market. This business has come into existence not because private equity is doing so well, but rather because it is struggling. Private equity firms are having a tough time cashing out in the usual ways, like selling a company or taking it public, and so a massive backlog of unsold companies has formed. According to McKinsey’s most recent Global Private Markets Report, over 16,000 buyout-backed companies have been held for more than four years — 52% of total buyout inventory, the highest share on record.

Investors are understandably antsy. So the way for existing investors to get out is selling their stakes to a new buyer, usually at a discount to their marks. In 2025, secondary transaction volume hit all-time highs, up 48% year-over-year. Hamilton Lane has been a huge buyer — at PAF, secondary stakes appear to account for almost half the fund’s value.

Where others see a crisis of liquidity, Hamilton Lane sees a heroic opportunity. The firm presents itself as the savior of the private equity ecosystem. After outlining the problems, Hamilton Lane’s co-head of secondary investments, Tom Kerr, who was dressed up as a knight, said, “Secondaries to the rescue!” He added, “Grab your shield, join the charge, let secondaries lead you to solid ground.”

The ground might be a tad shakier for buyers, however. Despite its medieval theme, Hamilton Lane’s video doesn’t mention the accounting sorcery laced through the company’s results. The day-one markups happen consistently — for example, last December, PAF acquired a stake in an A.Capital Partners vehicle for $50 million, and marked it up to $68.5 million within the same month. That same month, on December 31, a stake in a Hellman & Friedman fund was acquired for $20.3 million and marked up to $23.9 million the same day.

It’s not surprising that day-one markups account for about a third of PAF’s reported returns, according to an analysis conducted by Hunterbrook.

“The fair values of such investments as reflected in a fund’s NAV do not necessarily reflect the prices that would actually be obtained if such investments were sold.”

The practice of marking up assets on day one is allowed under accounting rules called a “practical expedient”: essentially a shortcut to make complex accounting easier. Hamilton Lane is not the only private equity player that takes advantage of this accounting practice: from StepStone ($STEP) to Apollo, funds utilize day-one markups of secondary interests. For example, the Apollo S3 Private Markets Fund, also an evergreen fund offered to retail investors, marks up stakes immediately.

At least three investor advocates have asked the Financial Accounting Standards Board to reconsider guidance that permits the practice. In one recent request, Mark J. Higgins of Enlightened Investor argues that markets have evolved since the existing guidance was issued. Indeed, while Hamilton Lane touts the depth of its relationships as a reason it can buy stakes at a discount, the secondaries business has become incredibly competitive. Investment banks like Evercore brag about their business advising buyers and sellers of secondary stakes.

If lots of people are fighting to buy something, it’s hard to believe anyone’s getting a steal. And if secondary transactions are occurring at ever-higher frequency and volume, why is there a need for expediency? Isn’t the price at which the secondary transactions occur a more accurate measure of value than the underlying fund’s NAV?

Some industry insiders defend the practice. At a recent SEC panel, Blake Nesbitt, the chief investment officer of the asset management firm Cliffwater, said the valuation process is not ideal — and leaves room for error — but noted that buyers don’t have the ability to delve into all of the underlying investments and figure out what each one is worth. For that reason, accepting a fund’s valuation can simplify an otherwise arduous task.

Others, including a former Hamilton Lane employee who spoke with Hunterbrook, argue that the actual sellers — investors in private equity funds — are willing to accept less than their stake might otherwise be worth in order to get liquidity. In its comment to Hunterbrook, Hamilton Lane says, “The simple fact is that investors in private equity funds will often sell their fund interests in the secondary market at a discount to their current value in exchange for liquidity prior to the end of a fund’s term. These transactions are negotiated between a willing seller and a willing buyer and are based on a record date valuation that is often months or quarters in advance of the transaction closing.” Which may be true, but does not account for the expanding depth of the secondary market itself and what that may mean for these “liquidity” discounts.

In Hamilton Lane’s market presentation, Gildea, co-head of evergreen portfolios, took a slightly different tack, saying it’s “clear” that the criticism that evergreen funds’ returns are not sustainable because of their use of secondaries to boost returns cannot be true, because to date, evergreen funds have outperformed. Well, yes, but every magic trick fools the audience until it doesn’t. If funds are outperforming based on marks that haven’t been tested, are they really outperforming?

Then there is the growing constipation in private equity: the industry’s increasing inability to exit the companies it owns. It’s hard to escape the logic that the gigantic backlog of unsold PE-backed companies reflects a growing disconnect between private equity marks and what buyers are willing to pay.

Add to all that the SaaSpocalypse. Software represents approximately 25% of total PE deal value over the past five years, and publicly traded software companies have plunged due to worries about the disruption from AI. Just as the buyout debt is being repriced by a skeptical market, the equity is facing a valuation reckoning that the industry’s internal marks have yet to acknowledge.

A March note from research firm KBW quantified Hamilton Lane’s exposure. After a trip with the company’s CFO and investor relations team, analyst Alex Bond reported that software represents 23% of total fee-paying assets under management and 30% of PAF.

Publicly, both private equity and private credit firms say their investments are doing fine. And one genuine advantage of private equity math, compared to private credit math, is that funds can be worth a great deal even if most investments are losers, so long as there’s one big winner that compensates for the wreckage. Power law and all that.

But it’s a little hard to believe that investors would be offloading stakes to Hamilton Lane at such steep discounts if the funds truly held such trophy assets.

And privately, you hear some doubts about the quality of these private market portfolios more generally. “Most of the businesses that were bought from 2018 to 2022 are lower quality than those companies,” Apollo’s Zito said in that supposedly off-the-record conversation, comparing private software businesses unfavorably to their public market counterparts. (Apollo shared in an earnings call that the company has 0% exposure to software in its private equity business.) “So I am concerned about many of [those] take-privates,” Zito noted.

Hamilton Lane’s own disclosures acknowledge the uncertainty: “Because there is significant uncertainty in the valuation of, or in the stability of the value of, illiquid investments, the fair values of such investments as reflected in a fund’s NAV do not necessarily reflect the prices that would actually be obtained if such investments were sold.”

A “Very Dangerous” Combination

While it’s true that other funds are also taking fees on unrealized gains, even industry veterans are taken aback by the combination of the evergreen fee structure and the reliance on day-one mark ups. Nesbitt, the Cliffwater CIO, called it “very dangerous,” noting that this structure effectively prioritizes the general partner’s fees over the investor’s reality. He called the arrangement “unheard of” in the history of private markets.

What isn’t widely understood is what a huge impact the accounting magic has had on Hamilton Lane’s earnings.

Overall, in the latest earnings call, Hamilton Lane reported $507 million in fee-related revenue for the nine months ending December 2025, with fee-related earnings up 37% year-over-year. But buried inside those stellar numbers is a newly created metric doing most of the heavy lifting.

Hamilton Lane’s total fee-related revenue is now the sum of standard management fees and something called “fee-related performance revenues,” or FRPR. In early 2025, Hamilton Lane said it “modified” its definition of fee-related earnings to exclude equity-based compensation and include FRPR. FRPR, says the firm, is “not dependent on realization events of the fund’s underlying investments.” Translation: These are fees Hamilton Lane collects based on markups that exist only on a spreadsheet, and which will only keep flowing if NAV keeps going up.

By simultaneously including this new metric and excluding stock-based compensation, the firm’s fee-related earnings grew by 37% year over year. Without it, growth would have actually been negative 16%.

In other words, more than 100% of the firm’s fee-related earnings growth came from technical adjustments, not any change to the fundamental value of the businesses it owns.

Troublingly, analysts, many of whom tell investors to buy Hamilton Lane’s stock, have priced Hamilton Lane as if this “new” revenue stream will grow smoothly for years. Goldman Sachs’ model, for instance, contemplates that after a blip in 2026 caused by a spike in redemptions, PAF’s value will appreciate at a steady 2.5% quarterly clip from March 2027 onward — roughly 10.4% compounded annually. You’ll be hard-pressed to find a 30-year window in private equity history in which quarterly marks were positive in every single quarter.

Hamilton Lane says “It’s important to note that we have a well-diversified business, providing solutions and services for a wide range of investors across the globe. If your calculation here was to simply remove FRPR, that would provide an inaccurate picture.” And in Hamilton Lane’s defense, many of its most aggressive practices have become the industry standard. But whether this means the practices are reasonable — or the private equity industry, on the whole, is more exposed in the event of redemptions than is widely understood — is the key question.

And Hamilton Lane’s practices are already starting to show some cracks. Funds that rely on day-one markups benefit from new money to buy more discounted assets, which creates more day-one markups, which keeps the reported performance high. But as the fund swells, the impact of new money — and, consequently, new day-one markups — is diluted, perhaps explaining why PAF’s performance has begun to sag.

Annual returns have compressed from 20.8% in fiscal 2022, when the fund managed $330 million, to 12.6% in fiscal 2025, when it managed $3.6 billion — a 10-fold increase in assets accompanied by a near-halving of returns. The since-inception figure of roughly 16% — which Hamilton Lane cited in its comment to Hunterbrook — is heavily weighted by those strong early years.

And, critically, Hamilton Lane hasn’t faced the pressure of forced liquidations. As of last year, money was still pouring into PAF: The fund reported about $146 million per month in average gross inflows during the first half of fiscal 2026, with $878 million in subscriptions against just $79 million in redemptions over that six-month period — an 11-to-1 ratio that is, by any standard, healthy. PAF today is not BREIT in December 2022. But that strength is also its primary vulnerability.

The entire model — the day-one markups, the FRPR, the reported IRR, the premium multiple the stock still commands — depends on a relentless stream of new money. Hamilton Lane’s next earnings report, scheduled for May 21, will give investors the first clean look at whether the ratio is holding. The math changes fast once it slows.

And the first sign that it may be slowing was revealed as this article was going to print. Public data from Hamilton Lane’s Global Private Assets Fund, or GPA — another flagship evergreen vehicle with $6.6 billion in assets — suggests that in March, the fund experienced an estimated net outflow of $172 million, its largest across the data Hunterbrook was able to review.4 It was only the second month of net outflows in the 18-month window for which data is publicly available — and it was nearly nine times larger than the first, a $19 million blip in September.

GPA is a different fund than PAF, but the dynamic is the same: The model depends on inflows exceeding redemptions. March suggests that equation may be starting to shift.

The Exit Doors

If there aren’t enough inflows to absorb the redemptions, then Hamilton Lane does have options—but none of them are good. They could use cash on hand to repay investors. They could add debt. They could tell investors they can’t have their money back, as multiple funds have done recently. Or they could start selling assets, thereby potentially triggering that vicious circle.

As Hamilton Lane itself warns in its fine print: “In a declining market, our evergreen vehicles may experience declines in value, and the pace of redemptions and consequent reduction in our assets under management could accelerate. Such declines in value may be both provoked and exacerbated by forced selling of assets, as further described below. Actions taken to meet substantial redemption requests could result in a material adverse effect on the fund’s investments, ability to make new investments or ability to achieve its investment objectives.”

And if Hamilton Lane starts seeing redemptions, it might take a long time to restore its brand.

“I don’t think memories are going to be short,” Dawn Fitzpatrick, the chief investment officer of Soros Fund Management, said in a recent interview about some of the practices she sees across the industry. “There’s going to be a culling of the alternative asset managers.”

Whether Pandora’s Box can be closed before the culling begins remains perhaps the industry’s most expensive question.

AUTHORS

Bethany McLean is the investigative journalist who exposed the notorious Enron fraud while a reporter at Fortune. She is a contributing editor at Vanity Fair who authored The Smartest Guys in the Room and several other books. She began her career as an investment banking analyst at Goldman Sachs before moving into journalism.

JD Jean-Jacques joined Hunterbrook from Goldman Sachs, where he worked as an investment banker. He was editor-in-chief of Howard University’s newspaper, The Hilltop, and wrote for The Exonian at Phillips Exeter Academy. Among other recognitions, he was a White House Correspondents’ Association scholar and was selected as student journalist of the year by The National Association of Black Journalists. He graduated from Howard with a B.A. in history.

Matthew Termine is a lawyer with nearly five years of experience leading the legal team at a mortgage technology company. In 2017, Matt was credited by the Wall Street Journal, among others, for identifying suspicious mortgage loan transactions that led to several successful criminal prosecutions, including that of a prominent political operative and the chief executive officer of a federally chartered bank. He is a graduate of Trinity College and Fordham University School of Law.

Editors

Jim Impoco is the award-winning former editor-in-chief of Newsweek who returned the publication to print in 2014. Before that, he was executive editor at Thomson Reuters Digital, Sunday Business Editor at The New York Times, and Assistant Managing Editor at Fortune. Jim, who started his journalism career as a Tokyo-based reporter for The Associated Press and U.S. News & World Report, has a Master’s in Chinese and Japanese History from the University of California at Berkeley.

Hunterbrook Media publishes investigative and global reporting — with no ads or paywalls. When articles do not include Material Non-Public Information (MNPI), or “insider info,” they may be provided to our affiliate Hunterbrook Capital, an investment firm which may take financial positions based on our reporting. Subscribe here. Learn more here.

Please contact ideas@hntrbrk.com to share ideas, talent@hntrbrk.com for work opportunities, and press@hntrbrk.com for media inquiries.

LEGAL DISCLAIMER

© 2026 by Hunterbrook Media LLC. When using this website, you acknowledge and accept that such usage is solely at your own discretion and risk. Hunterbrook Media LLC, along with any associated entities, shall not be held responsible for any direct or indirect damages resulting from the use of information provided in any Hunterbrook publications. It is crucial for you to conduct your own research and seek advice from qualified financial, legal, and tax professionals before making any investment decisions based on information obtained from Hunterbrook Media LLC. The content provided by Hunterbrook Media LLC does not constitute an offer to sell, nor a solicitation of an offer to purchase any securities. Furthermore, no securities shall be offered or sold in any jurisdiction where such activities would be contrary to the local securities laws.

Hunterbrook Media LLC is not a registered investment advisor in the United States or any other jurisdiction. We strive to ensure the accuracy and reliability of the information provided, drawing on sources believed to be trustworthy. Nevertheless, this information is provided “as is” without any guarantee of accuracy, timeliness, completeness, or usefulness for any particular purpose. Hunterbrook Media LLC does not guarantee the results obtained from the use of this information. All information presented are opinions based on our analyses and are subject to change without notice, and there is no commitment from Hunterbrook Media LLC to revise or update any information or opinions contained in any report or publication contained on this website. The above content, including all information and opinions presented, is intended solely for educational and information purposes only. Hunterbrook Media LLC authorizes the redistribution of these materials, in whole or in part, provided that such redistribution is for non-commercial, informational purposes only. Redistribution must include this notice and must not alter the materials. Any commercial use, alteration, or other forms of misuse of these materials are strictly prohibited without the express written approval of Hunterbrook Media LLC. Unauthorized use, alteration, or misuse of these materials may result in legal action to enforce our rights, including but not limited to seeking injunctive relief, damages, and any other remedies available under the law.

By: Sam Koppelman, Jenny Ahn, Matthew Termine, Alex Sternlicht, Michelle Cera

Editor: Jim Impoco

Hunterbrook Media’s investment affiliate, Hunterbrook Capital, has no positions related to this article at the time of publication. Positions may change at any time. See full disclosures on our website.

Something doesn’t add up at Lennar ($LEN), America’s second-largest homebuilder.

The numbers are bad: Margins have cratered, cash flow has dwindled, and the stock has crashed.

But it isn’t just that the numbers are bad — it’s that they are surprisingly bad. For 10 straight quarters, Lennar’s stock fell the day after reporting earnings.1 Wall Street keeps modeling the company’s financials, and keeps getting it wrong.

So what is everybody missing?

Lennar itself raised this question in a statement released around noon yesterday that appeared random to observers of the name — framed as a response to “several analysts and investors” asking the company about its strategy. The contents of the press release were anodyne, but the stock fell to multiyear lows within hours.

“If you’re explaining, you’re losing,” wrote analysts at Evercore in a research note about $LEN.

What Lennar did not mention: Hunterbrook had repeatedly reached out to the company since last week with specific, detailed questions about its accounting and financials. Lennar — which in its statement claimed to be “deeply committed to engaging with investors, analysts, and all stakeholders — ghosted that outreach.

Hunterbrook wrote Lennar again after the statement on Monday. Lennar’s response: Nada. The only answers to Hunterbrook’s questions Lennar provided seemed to be selectively disclosed to analysts at banks, rather than the public. (The corporate equivalent of responding to rumors on your Close Friends story.)

Lennar’s narrative: Its numbers have been disappointing because of soft demand and aggressive discounting amid a widespread housing downturn. Lennar famously slashed home prices further than any peer, a bet on volume over margins that management framed as a deliberate strategy. The company reiterated that margin impact is largely a reflection of the volume-based strategy defended in yesterday’s statement.

But if Lennar’s strategy is working, why the carnage?

Specifically, why has Lennar consistently tripped over Wall Street’s gross margin projections every quarter since 2023? Is it really just the discounting?

To solve this mystery of the vanishing margins and immense cash burn, Hunterbrook Media analyzed proprietary national real estate data, hidden variables across SEC filings, and obscure documents from Lennar counterparties — including the pitch decks that Lennar’s financing partner, Angelo Gordon, presented to a California county pension fund.2 We also spoke with multiple independent accounting experts, investment analysts, a land banker, and investors who have combed through Lennar’s financials.

The evidence pointed Hunterbrook toward an explanation so buried in the books it seems that many of those covering Lennar missed it.

The homebuilder last year completed an ambitious transformation: moving 98% of its pre-construction land off its books and into third-party “land banks.” Executives projected this “land-light” strategy would cost just 1 percentage point of gross margin.

But Hunterbrook’s analysis indicates Lennar’s pivot to land banking has locked the company into paying Wall Street firms more than $2 billion a year in annual fees just to maintain options on land it used to own, according to Hunterbrook’s estimates. That figure now rivals the company’s entire net income.

These fees are generally paid monthly or accrued over time until Lennar takes delivery of the properties. Either way, most fees seemingly aren’t immediately showing up as expenses. Instead, Lennar appears to be capitalizing some of these disbursements — recording billions in accumulated spending as though it is an asset.

“It’s delaying the inevitable,” said JP Krahel, chair of Loyola University Maryland’s accounting department. “Seems a little nuts.”

The costs have already begun to creep into margins, though they appear to be masked by aggressive cuts to construction expenses. But the worst is likely still ahead for Lennar. The land bank portfolio has grown every year since 2020, and the average lot likely sits in the bank for roughly two years before Lennar takes delivery.3

That means the income statement today seemingly reflects only the earlier, smaller vintages of land bank deals.

The later vintages — carrying years of accumulated fees — haven’t hit yet. And Lennar’s sales pace suggests some of the land is sitting in the banks significantly longer than originally planned, silently accruing fees until Lennar takes delivery.

In the meantime, Lennar’s leadership appears to be heading for the exits. The co-CEO, general counsel, head of IR, and several other executives, including ones tied closely to land-banking, have recently departed.

“A number of long-term Lennar associates have chosen to retire … and let the next generation shine,” said CEO Stuart Miller, who has been dropping guarded hints about how expensive the land deals have been. The asset-light transformation, he told analysts on the most recent call, is “getting more interesting by the day.”

“Interesting” is likely not the word Lennar homeowners would use to describe what has happened to their homes during the company’s push to save money wherever it can. In 2025, Hunterbrook published an investigation into quality issues from Lennar and other homebuilders — ranging from severe mold infestations to cracked foundations to violations of building codes, according to interviews with 22 Lennar homeowners.

“It’s like they give you the impression that they’re rapidly building because they have such a high demand,” a Lennar homeowner who asked to be referred to only by her last name, Marks, told Hunterbrook. She said Lennar sold her a “lemon” home full of defects in 2024. “Then once you buy and move in, you realize nobody’s moving into those other homes.”

Those empty homes may be signs of a broken business model.

The industry affectionately called Lennar’s volume-based growth engine the “Lennar Machine.” But the machine has become a treadmill that Lennar no longer controls: The company appears to be churning out more and more homes from the option pipeline to stop the fees from accruing while the land sits in the bank. But in order to sell these homes in a weak housing market, Lennar is slashing effective prices, inclusive of incentives, destroying margins already buckling under the weight of the fees.

It’s a lose-lose for both Lennar and for the people like Marks surrounded by empty lots, watching their home values — and the stock price of the company that sold them the dream — plummet.

How’d Lennar get here?

The Big Bet

Lennar no longer owns the land it plans to build on — not until the moment it’s ready to start construction.

Over the past seven years, the company has executed on an extreme version of a big bet made by many major homebuilders — selling or transferring nearly all of the land it once owned to land developers or financial firms. In place of the land itself, Lennar contracted with these firms for options to purchase land. With remarkable speed, Lennar pulled off this “land light” strategy, going from owning roughly 75% of its land in the fourth quarter of 2018 to just 2% in the fourth quarter of 2025, with the rest controlled through options.

Optioning land allows homebuilders to free up massive amounts of capital that would otherwise be tied up with owned land. By keeping these properties in what is called a “land bank,” builders can secure a pipeline of land to buy when they need it, just in time for home construction. They don’t have to own all that land while they develop it or wait until the market is right.

Experts told Hunterbrook that the idea was to make Lennar’s balance sheet look more like that of NVR ($NVR), a smaller rival that Wall Street rewards with a higher multiple for keeping its books light. In March 2025, Miller told investors the land-light strategy would only carry a cost of roughly 100 basis points, or 1 percentage point, of margin — a small price to pay for a Wall Street makeover and attendant higher valuation.

But that’s not what happened.

Unlike NVR, which options land from local developers in a network that took decades to build, Lennar wanted a faster, more institutional setup. It turned to private equity firms like TPG’s Angelo Gordon and spun off its legacy land into a real estate investment trust (REIT) named Millrose Properties ($MRP). Miller called the company’s transition “a strategic national process,” though he later admitted, “We’ve put these things together pretty quickly.”

The main issue is that institutional land banks aren’t cheap. First, Lennar pays those firms deposits — about 5% to 20% of the land’s value — pretty standard for builders who option land from developers.

Second, what’s notable about the institutional land banks that Lennar borrows from is that they also charge option fees, which essentially amount to interest payments. To Millrose, Lennar pays an average of 8.5% annually. The number is likely higher for other counterparties like Angelo Gordon, which reported a 15% internal rate of return on its first Lennar land bank, according to a document obtained by Hunterbrook.

Contrast this with the NVR model Lennar seems to be emulating. NVR largely enters into land purchase contracts with developers for finished lots and pays a one-time deposit at roughly 10% of the land’s purchase price, which goes toward the principal. And the company doesn’t appear to pay recurring option fees. Which means, essentially, NVR’s cost to option land beyond interest is simply the impairment write-downs of the initial deposit if the builder decides not to purchase the land. In all of 2025, NVR paid $75.9 million total in impairments, nowhere near the amount Lennar may owe in fees.

Hunterbrook estimates those option fees will cost Lennar billions annually upon maturity, rivaling the company’s entire net income.4 The 1-percentage-point margin drag management promised appears to be way off.

Delaying the Inevitable

There seem to be two ways Lennar pays option fees. 1) With Millrose, Lennar pays in regular monthly installments — 8.5% on the value of the finished land. 2) With Angelo Gordon, the deals may be structured more like a PIK loan5 — with the option fees rolling into the purchase price when Lennar finally takes delivery of the lot, or when Lennar abandons the project, according to Angelo Gordon’s pitch decks, master agreement, and a video of Angelo Gordon’s presentation to a California county pension fund in 2024.6

It’s not clear precisely how these deals work, as the details surrounding them are largely kept private. But Hunterbrook’s reporting suggests this structure enables Lennar to pull off an accounting two-step.

During the period in which Lennar is paying out options fees, the company does not appear to expense them. Instead, Lennar seems to be parking these costs on the balance sheet under “deposits and pre-acquisition costs” until they purchase the lot — or take an impairment on it if they decide to walk — at which point these expenses eventually flow through to Lennar’s COGS.

This approach — which may be perfectly legal! — enables Lennar to present better earnings today, at the expense of worse COGS tomorrow.

Why do we think this?

One seemingly reasonable assumption is that if the deposit and pre-acquisition costs line really reflects money Lennar puts down to genuinely purchase land, then deposits would rise and fall roughly in step with the number of homesites Lennar controls. More land, more deposits and pre-acquisition costs. Less land, fewer deposits and pre-acquisition costs.

That’s how it works at NVR, the company Lennar reportedly models itself after. Its deposits track homesites almost perfectly. Over the course of 2025, NVR’s controlled homesites ranged from 155,000 to 169,250. Its deposit line moved roughly in lockstep: from $727 million to $851 million, holding steady at roughly $4,700 to $5,000 per homesite. No mystery.

At Lennar, the two numbers have decoupled.

Between February 2025 and February 2026 — the four quarters following the Millrose spinoff — Lennar’s controlled homesites fell by roughly 47,000, from 533,000 to 486,000. At Lennar’s average deposit and pre-acquisition costs — of roughly $12,900 per homesite as of November 2025 — that reduction theoretically should have pulled the deposit line down by approximately $600 million.

Instead, the deposit line grew by $1.6 billion — from $5.2 billion to $6.8 billion.

That’s a $2.2 billion gap between what the deposit line theoretically “should” have done and what it actually did. Even accounting for normal churn — Lennar continuously options new land to replace lots it takes down — a gap of that magnitude points to something other than new deposits driving the growth, likely a combination of capitalized option fees, infrastructure spending, and other pre-acquisition costs. (After all, the number of homesites are down!)

Now, there may be reasonable explanations for the gap.

Lennar develops the lots while they sit in the bank. Laying roads and utilities. Grading. Permitting and entitlements. These costs could add up quickly, depending on how raw the land was when Lennar optioned it.

That’s what Evercore offered as a possible explanation: infrastructure costs. In Texas, for instance, Municipal Utility Districts (MUDs) require builders to fund infrastructure upfront and later reimburse them through bond issuances; analysts at Evercore ISI, in the note published the same day as Lennar’s statement, estimated MUD spending alone accounts for over $300 million of the deposit line’s recent growth. (Lennar’s MUDs raise their own series of questions.)

But even Evercore’s own analysis said the “concerns around capitalized” options fees are “not without merit,” noting that Lennar’s non-Millrose land bank exposure is two to three times the size of its Millrose exposure and that the margin headwind could be well above the company’s 100-basis-point estimate. Evercore called on Lennar to provide more disclosure about what’s in the line — a request the company has so far declined to fulfill, though its next SEC filing, a 10-Q, which appears to be imminent, may offer answers.

And experts who spoke with Hunterbrook believe the most plausible primary source for the unexplained billions of dollars in this line item over the years is the capitalized option fees Lennar has been paying out to its land banking partners. Lennar’s management also confirmed to at least one investor — and seemingly the sell-side — that the deposit line includes option maintenance fees.

Krahel and other accountants interviewed by Hunterbrook questioned Lennar’s approach. “I don’t capitalize an electric bill because the second I stop paying it, the lights go out,” said Krahel, the chair of the accounting department at Loyola University Maryland.

He offered a second analogy: a pawn shop. If you pay $50 a month for a shop to hold your watch, those payments don’t reduce the ultimate price you pay. “Those monthly payments don’t offset the value of the gold watch when I come buy it back,” he said. “That’s rent expense. That’s really what this is.”

The issue, Krahel explained, is that the option fees Lennar is capitalizing aren’t actually an asset. They’re not paying down principal and reflecting the value of new land ownership. They’re just expenses, like interest. The money leaves Lennar and goes to other firms. Period. He described the accounting rules governing these kinds of transactions as “fundamental principles,” but it’s unclear whether Lennar is following them.

“Generally speaking, you cannot capitalize an expenditure if you’re spending it on something you don’t control,” Krahel explained. But if Lennar indeed does “control” the properties on which it is paying management fees, Krahel argued, then that land shouldn’t be off its books. “They’re trying to eat their cake and have it too,” Krahel said. “If they control it, it should be on their balance sheet.”

To be clear: Lennar’s accounting may be permissible. Under GAAP, payments to obtain an option on real property can be capitalized, and Lennar could argue that because the fees become part of the eventual lot purchase price, they are properly recorded as pre-acquisition costs. Reasonable accountants disagree on where the line falls — and Lennar said in its statement that its disclosures and accounting treatment “have been carefully vetted.”

But the financial consequences are the same regardless of how the fees are classified.

Whether Lennar records them as an asset today or an expense today, the cash has already left the building. And when the lots are finally taken down, the accumulated cost should hits the income statement either way. The accounting treatment determines when investors see the damage — not whether the damage exists. In the meantime, reported earnings look relatively intact, the balance sheet line keeps growing, and investors are left evaluating a company that appears to earn $2.1 billion a year but is quietly spending what appears to be a comparable amount to maintain its land pipeline.

Then, when Lennar takes down a lot from a land bank, the accumulated fees don’t get paid back. They become a cost, flowing straight into cost of goods sold and cutting directly into gross margins. And if Lennar abandons a lot, it takes an impairment instead.

“You’re feeding everybody on the pathway to the takedown moment thinking everything’s better than they are,” Krahel said. “And then — hello, surprise — here are expenses that were sitting on the balance sheet instead of having flowed through the income statement.”

The Inevitable: Cost Recognition

That “surprise” is no longer theoretical. Hunterbrook’s reporting indicates that it has started to bleed onto Lennar’s income statement as the first wave of optioned lots have been taken down — even if these fees were once marked as an asset.

It’s not immediately clear where these fees show up in Lennar’s earnings, because the company does not itemize them in its cost of goods sold (COGS).

But you can roughly back into the option fees as part of those COGS — and margin hit — through a few different, if imperfect, proxies.

One is through a comparison with D.R. Horton.

D.R. Horton, which targets very similar buyers at around the same price points, had comparable cost structures as Lennar from from 2020 to 2022. And both companies significantly cut down on their construction costs, which, in theory, should have reduced their COGS.

But starting in 2023, Lennar’s COGS as a share of revenue began pulling away, culminating in a roughly 2.8 percentage point gap in FY2025 — approximately $800 million to $1 billion in extra costs. That’s almost certainly a reflection of Lennar’s rising land costs, not construction costs (land and construction costs are the two most important homebuilding costs), because Lennar’s construction costs haven’t only been falling; they’ve been doing so at nearly twice the rate of D.R. Horton.7

This is, perhaps, an explanation for why Lennar appears to be among the only major homebuilders that actually almost lost money in the last 12 months.

And this may only be the beginning, because much of the fees Lennar has accumulated likely haven’t reached the income statement yet. The land-bank portfolio has grown every year since FY2020, with vintages growing over time — from 119,283 controlled homesites to 496,250 by FY2025. That means, assuming Lennar’s optioned land sits roughly two years in the bank before takedown, the income statement today reflects just the carrying costs on the earlier, smaller vintages. Meaning, the large tranches optioned in recent years — including over 100,000 homesites Lennar added to its portfolio last year alone — are still accruing fees in the background.

Based on this vintage approach, Hunterbrook estimates that last year, Lennar paid about $1.5 billion in option fees, or an average of roughly $20,000 per home sold.8 In an extremely simplified calculation, Hunterbrook assumed that the option rate on all the lots was 8.5%, the rate Millrose charges — almost certainly an underestimate given the higher market rates that Lennar likely paid to third-party lenders before Millrose was created in 2025. (This model also assumes Lennar has been taking down all homesites according to schedule; it doesn’t take into account a potential reality where Lennar’s takedowns have actually skewed toward the end of the term or renegotiated past the original term — events which would add to the accrued option fees.9)

If our analysis is right, as the portfolio matures and Lennar begins taking down lots with the full weight of accumulated carry costs, option costs as share of revenue will continue to climb. By FY2028, option costs on its portfolio could be roughly $2.5 billion or more, Hunterbrook estimates, potentially bringing down margins to a level barely enough to clear Lennar’s overhead. (Note: These truly are just estimates. The company provides very little information on this, so we backed into it. If you have a better way to calculate it, let us know at ideas@hntrbrk.com.)

The upshot is Lennar may end up taking a much bigger gross margin hit than the company telegraphed when it pivoted to the land-banking model — and Lennar’s margins are already much worse than peers.

In 2025, Lennar previewed just 100 basis points of margin as the cost of optioning land. That may have been true if you were just looking at the 2024 estimated takedown numbers, reflecting a little over half of Lennar’s current portfolio. Option fees net of the cost of owning land — using its corporate debt rate of around 5% as a proxy — would have gotten you close to that 100 bps spread. But by the time Lennar’s current portfolio matures, that spread will have increased to 300 bps, we estimate.

And while a few hundred basis points doesn’t sound like much — a small price to pay for the perks of a land-light strategy, an easily absorbable margin cost if you can keep your sales prices and volume pace healthy — that doesn’t tell the whole story.

The fundamental problem with Lennar’s new model is that slowing down has gotten much more expensive. When Lennar owned its land, the cost of delaying construction was roughly 5% interest. Now it’s at least 8.5% — and the land bank contracts don’t wait. Lennar seemingly had to renegotiate as much as $2.4 billion worth of options last year, and was “compelled to takedown” an additional $663 million from option land in 2025 — even as it sat on roughly 5,000 completed homes it couldn’t sell, a 72% increase from the year before.10 To move those homes, Lennar has needed to slash prices, squeezing margins already buckling under the option costs.

Stijn Van Nieuwerburgh — a real estate and finance professor at Columbia Business School — described the model to Hunterbrook as involving layered, and partly hidden, leverage. On the balance sheet, he argued, a “land-light” strategy can look less risky, but the economic exposure can end up resembling the older model that homebuilders such as Lennar are trying to move away from.

“It’s leverage on top of leverage,” Van Nieuwerburgh said. “In an upcycle, leverage helps you; in a downcycle, it can be devastating.”

“The optionality is supposed to reduce risk. But if the contracts embed enough financing and enough effective commitments, it can end up doing less risk reduction than advertised.”

But Lennar can renegotiate or walk, right? Isn’t that the point of the “land-light” model?

At least according to the terms Hunterbrook reviewed, there doesn’t seem to be a lot of wiggle room built into Lennar’s arrangement — although “flexibility” was supposed to be a key feature of the land-light structure.

Lennar’s portfolio with Millrose is based on a “cross-termination pooling” structure that keeps Lennar from being able to walk away from a bad community in one geography without triggering forfeitures across an entire pool of communities. Meaning, Lennar likely has to sit on optioned lots for a long time in markets where it makes little sense to start building — say in Florida, where home prices have dropped as much as 16% in some parts.

Angelo Gordon deals don’t look much more flexible for Lennar either.

According to Angelo Gordon’s pitch decks, the project cycle is limited to a 40-month period where Lennar has to takedown a certain number of lots each month on a rigid schedule. If Lennar falls behind that schedule, Angelo Gordon has the right to immediately terminate, keep the deposit, and sell the fully improved lots to a competing builder.

“There’s really no flexibility,” Angelo Gordon managing director Bryan Rush assured SBCERA, San Bernardino County employee’s pension fund, in 2024. SBCERA later approved a $150 million investment in Angelo Gordon’s land bank.

Angelo Gordon’s leash?

Angelo Gordon considers Lennar’s 19.9% upfront deposit — nearly twice the 10% Millrose charges — “first loss,” meaning Lennar’s deposit would take the first hit if deals go south. It’s another way Angelo Gordon seems to protect its downside, while leaving Lennar with the risks of forfeiting the entire deposit if projects stall, according to Angelo Gordon’s pitch to the pension fund.

And, presumably, those protections really matter for California. Because that state has seen what can happen when it doesn’t have them.

In January 2007, the massive state pension fund CalPERS invested roughly $970 million for a 62% financial interest in a Lennar land venture near Los Angeles.

Seventeen months later, the joint venture filed Chapter 11 bankruptcy. CalPERS lost its investment: $922 million, one of the worst single losses in the pension fund’s history.

Today, Angelo Gordon is channeling money from another California public pension fund, SBCERA, to fund the Lennar land acquisition.

But this time, in the event of a crisis, it seems like the role of bagholder would be reversed. It is Lennar, not the pension fund, that posted the first-loss deposit.

Asked how she would feel if Lennar had to take billions worth of impairments on its land bank investments, Marks — the resident who claims she was sold a “lemon” by Lennar — was blunt.

“Good,” she said, “lose money, like the rest of us homeowners.”

Authors

Jenny Ahn joined Hunterbrook after serving many years as a senior analyst in the US government. She is a seasoned geopolitical expert with a particular focus on the Asia-Pacific and has diverse overseas experience. She has an M.A. in International Affairs from Yale and a B.S. in International Relations from Stanford. Jenny is based in Virginia.

Matthew Termine is a lawyer with nearly five years of experience leading the legal team at a mortgage technology company. In 2017, Matt was credited by the Wall Street Journal, among others, for identifying suspicious mortgage loan transactions that led to several successful criminal prosecutions, including that of a prominent political operative and the chief executive officer of a federally chartered bank. He is a graduate of Trinity College and Fordham University School of Law.

Sam Koppelman is a New York Times best-selling author who has written books with former United States Attorney General Eric Holder and former United States Acting Solicitor General Neal Katyal. Sam has published in the New York Times, Washington Post, Boston Globe, Time Magazine, and other outlets — and occasionally volunteers on a fire speech for a good cause. He has a B.A. in Government from Harvard, where he was named a John Harvard Scholar and wrote op-eds like “Shut Down Harvard Football,” which he tells us were great for his social life.

Alexandra Sternlicht is an award-winning journalist and Columbia Business School MBA candidate. She previously reported on technology at Fortune, where her investigations were cited by the U.S. Supreme Court, lawmakers and major media outlets. She’s also worked at Forbes and The New York Times. Alexandra is an alum of Columbia Journalism School’s Knight-Bagehot Fellowship and graduated magna cum laude from the University of Pennsylvania.

Michelle Cera trained as a sociologist specializing in digital ethnography and pedagogy. She completed her PhD in Sociology at New York University, building on her Bachelor of Arts degree with Highest Honors from the University of California, Berkeley. She has also served as a Workshop Coordinator at NYU’s Anthropology and Sociology Departments, fostering interdisciplinary collaboration and innovative research methodologies.

Editor

Jim Impoco is the award-winning former editor-in-chief of Newsweek who returned the publication to print in 2014. Before that, he was executive editor at Thomson Reuters Digital, Sunday Business Editor at The New York Times, and Assistant Managing Editor at Fortune. Jim, who started his journalism career as a Tokyo-based reporter for The Associated Press and U.S. News & World Report, has a master’s in Chinese and Japanese History from the University of California at Berkeley.

Graphic

Dan DeLorenzo is a creative director with 25 years reporting news through visuals. Since first joining a newsroom graphics department in 2001, he has built teams at Bloomberg News, Bridgewater Associates, and the United Nations, and published groundbreaking visual journalism at The Wall Street Journal, Associated Press, The New York Times, and Business Insider. A passion for the craft has landed him at the helm of newsroom teams, on the ground in humanitarian emergencies, and at the epicenter of the world’s largest hedge fund. He runs DGFX Studio, a creative agency serving top organizations in media, finance, and civil society with data visualization, cartography, and strategic visual intelligence. He moonlights as a professional sailor working toward a USCG captain’s license and is a certified Pilates instructor.

Hunterbrook Media publishes investigative and global reporting — with no ads or paywalls. When articles do not include Material Non-Public Information (MNPI), or “insider info,” they may be provided to our affiliate Hunterbrook Capital, an investment firm which may take financial positions based on our reporting. Subscribe here. Learn more here.

Please contact ideas@hntrbrk.com to share ideas, talent@hntrbrk.com for work opportunities, and press@hntrbrk.com for media inquiries.

LEGAL DISCLAIMER

© 2026 by Hunterbrook Media LLC. When using this website, you acknowledge and accept that such usage is solely at your own discretion and risk. Hunterbrook Media LLC, along with any associated entities, shall not be held responsible for any direct or indirect damages resulting from the use of information provided in any Hunterbrook publications. It is crucial for you to conduct your own research and seek advice from qualified financial, legal, and tax professionals before making any investment decisions based on information obtained from Hunterbrook Media LLC. The content provided by Hunterbrook Media LLC does not constitute an offer to sell, nor a solicitation of an offer to purchase any securities. Furthermore, no securities shall be offered or sold in any jurisdiction where such activities would be contrary to the local securities laws.

Hunterbrook Media LLC is not a registered investment advisor in the United States or any other jurisdiction. We strive to ensure the accuracy and reliability of the information provided, drawing on sources believed to be trustworthy. Nevertheless, this information is provided “as is” without any guarantee of accuracy, timeliness, completeness, or usefulness for any particular purpose. Hunterbrook Media LLC does not guarantee the results obtained from the use of this information. All information presented are opinions based on our analyses and are subject to change without notice, and there is no commitment from Hunterbrook Media LLC to revise or update any information or opinions contained in any report or publication contained on this website. The above content, including all information and opinions presented, is intended solely for educational and information purposes only. Hunterbrook Media LLC authorizes the redistribution of these materials, in whole or in part, provided that such redistribution is for non-commercial, informational purposes only. Redistribution must include this notice and must not alter the materials. Any commercial use, alteration, or other forms of misuse of these materials are strictly prohibited without the express written approval of Hunterbrook Media LLC. Unauthorized use, alteration, or misuse of these materials may result in legal action to enforce our rights, including but not limited to seeking injunctive relief, damages, and any other remedies available under the law.

By: Blake Spendley Carter Johnston

Editor: Sam Koppelman

Hunterbrook Media’s investment affiliate, Hunterbrook Capital, has no positions related to this article at the time of publication. Positions may change at any time. See full disclosures on our website.

Nine months before Iran reportedly began mining the Strait of Hormuz, a U.S. Navy director responsible for mine countermeasures delivered a candid assessment before an audience of international naval observers: The Navy’s next-generation mine-hunting systems suffered from unreliable unmanned vehicles, critical single-point failures, and sonar that couldn’t see.

Hunterbrook Media has obtained a copy of that briefing and its associated slides.

The picture they paint is harrowing. The U.S. stares down the barrel of a bleak scenario: a trio of flawed ships responsible for keeping open a global shipping artery that carries one-fifth of the world’s oil supply.

It wasn’t always this way.

In January, the U.S. Navy loaded four decommissioned Avenger-class minesweepers onto a heavy-lift cargo ship in Bahrain and sent them home for scrap. The wooden-hulled ships had spent the entire post-Cold War era keeping the Persian Gulf’s sea lanes clear — the only vessels in the American fleet at the time purpose-built to find and destroy naval mines. Their fledgling replacements — modern Independence-class Littoral Combat Ships appended with new unmanned mine countermeasures — had just arrived in theater. The transition was complete.

Two months later, Iran is reportedly laying mines in the Strait of Hormuz. For the United States, the artery is blocked.

And the U.S. Navy has three ships that haven’t been battle-tested standing between an Iranian minefield and the passage responsible for much of the world’s oil, fertilizer, and other critical products.

The current American mine countermeasures force positioned near the Persian Gulf is made up of Littoral Combat Ships the USS Canberra, USS Tulsa, and USS Santa Barbara. All three are Independence-class vessels — aluminum-hulled trimarans originally designed as multi-mission platforms, now carrying the Navy’s new Mine Countermeasures Mission (MCM) package in a largely fixed configuration. The modular concept that was supposed to be the LCS’s selling point — swap packages in, swap them out, reconfigure the ship for whatever the mission demands —has been quietly abandoned.

What each ship brings to a minefield: one MH-60S Seahawk helicopter carrying an airborne mine detection system using lasers and a mine neutralization system; two unmanned surface vehicles that can tow either the AN/AQS-20 side-scan and volume-search sonar or the unmanned influence sweep system; and the crew to run it all.

On paper, the package covers the full kill chain — find, classify, identify, neutralize.

The system reached initial operational capability in September 2022, but reportedly did not deploy on an actual ship until Canberra and Santa Barbara sailed in February 2025. Tulsa followed in May. Kansas City received its package that same month. The entire operational history of the LCS MCM mission package, from first deployment to today’s crisis, is measured in months.

For comparison, the ships they replaced had been doing this work for decades.

The Avenger class entered service in the 1980s. Their fiberglass-coated wooden hulls were designed specifically to reduce magnetic signatures — a basic requirement when you’re driving over mines that detonate based on what they sense passing above them. The LCS is a metal ship, and is significantly larger than an Avenger-class vessel, limiting how close it can safely operate in known or suspected minefields. In return, it sports a slightly more capable self-defense suite against surface and air threats, equipped with a 57 mm gun and a SeaRAM launcher. But it still needs escort protection during the slow, exposed work of sweeping and hunting. At the moment, as Reuters reported, the Navy has refused to provide escorts in the Gulf.

The Washington Institute estimated years ago that clearing the Strait of Hormuz of mines could require “up to 16 MCM vessels.” The Navy has seven. Iran has an estimated 5,000 to 6,000 mines and, according to U.S. intelligence in conversations with CNN, still retains “80% to 90% of its small boats and miners.”

The brief covered a recent MCM Advanced Tactical Training program, the final pre-deployment mine warfare assessment for LCS crews.

Some key findings:

Unreliable unmanned systems. Each Fleet-class USV mission requires over four hours of “pre-mission maintenance” and “1.5 hours of GPS/sonar calibration once launched,” according to the presentation. Multiple hunt missions were conducted where the sonar simply failed to record data — and crews didn’t know until the post-mission analysis. This is especially damaging during reacquire-and-identify missions, exactly the kind of work needed to clear a minefield.

Operators have responded by shortening mission times, which defeats the purpose of using unmanned vehicles in the first place. One pre-deployment exercise with the USS Tulsa off the coast of San Diego resulted in a runaway MCM USV near Mexico’s territorial waters that could not be recovered by the mothership LCS. “Literally, the practice minefield I use is 1 mile north of the US-Mexico maritime border, and there’s a good chance that that UUV drifts or decides to go off on its own. I’m going to get demarched by the Mexican government,” said the leader of the U.S. Navy’s Mine Countermeasures Technical Division. The USVs themselves act as a handicap to minesweeping, with a short bandwidth range forcing the mothership LCS to operate near or inside minefields to maintain visual range to the USV’s antennas.

Visual identification doesn’t work. U.S. MCM doctrine requires a camera to visually confirm mines — the AQS-20 has to drive directly over a bottom mine. But even the relatively clear waters off Southern California have defeated this approach. In the turbid, shallow, current-swept waters of the Persian Gulf, the problem would be far worse. The officer’s conclusion: The Navy needs to adopt high-granularity sonar identification, as other navies already have.

Critical single-point failures. The platform lift between mission bay and hangar, the BIT test laptops for the USV/ALMDS/AMNS, the twin boom extensible crane, and the payload handling systems are all single-point failures with no spares or redundancy aboard. If any one of these breaks, operations stop. When describing the deployment arm, the Navy mine countermeasures lead said, “It is a troubling system. It is highly complex for what it does, and when it breaks, I’m out of a job, I’m out of a mission.”

Multi-mission dilution. The LCS was designed as a multi-mission platform. The addition of Naval Strike Missiles and pressure to support visit, board, search, and seize operations means crews have less time to build and maintain MCM proficiency. “So now my ship with an LCS mission package may not necessarily be practicing MCM.” The LCS platform is also being experimented on as a long-range strike platform. The director’s own conclusion: The LCS will always struggle to match a dedicated MCM vessel.

Meanwhile, Hunterbrook sent a source Philadelphia to check out the old minesweepers — where the only war they are proximal enough to intervene in now is the cheesesteak rivalry between Pat’s and Geno’s.

The Pentagon declined to comment.

Authors

Blake Spendley joined Hunterbrook from the Center for Naval Analyses (CNA), where he led investigations as a Research Specialist for the Marine Corps and US Navy. He built and owns the leading open-source intelligence (OSINT) account on X/Twitter, called @OSINTTechnical (over 1 million followers), which also distributes Hunterbrook Media reporting. His OSINT research has been published in Bloomberg, the Wall Street Journal, and The Economist, among other top business outlets. He has a B.A. in Political Science from USC.

Carter Johnston is a second-year undergraduate student at George Washington University studying International Affairs and National Security. He is a freelance defense journalist covering the Air Force and Navy for The Air Current and Naval News since 2024.

Editor

Sam Koppelman is a New York Times best-selling author who has written books with former United States Attorney General Eric Holder and former United States Acting Solicitor General Neal Katyal. Sam has published in the New York Times, Washington Post, Boston Globe, Time Magazine, and other outlets — and occasionally volunteers on a fire speech for a good cause. He has a B.A. in Government from Harvard, where he was named a John Harvard Scholar and wrote op-eds like “Shut Down Harvard Football,” which he tells us were great for his social life.

Hunterbrook Media publishes investigative and global reporting — with no ads or paywalls. When articles do not include Material Non-Public Information (MNPI), or “insider info,” they may be provided to our affiliate Hunterbrook Capital, an investment firm which may take financial positions based on our reporting. Subscribe here. Learn more here.

Please contact ideas@hntrbrk.com to share ideas, talent@hntrbrk.com for work opportunities, and press@hntrbrk.com for media inquiries.

LEGAL DISCLAIMER

© 2026 by Hunterbrook Media LLC. When using this website, you acknowledge and accept that such usage is solely at your own discretion and risk. Hunterbrook Media LLC, along with any associated entities, shall not be held responsible for any direct or indirect damages resulting from the use of information provided in any Hunterbrook publications. It is crucial for you to conduct your own research and seek advice from qualified financial, legal, and tax professionals before making any investment decisions based on information obtained from Hunterbrook Media LLC. The content provided by Hunterbrook Media LLC does not constitute an offer to sell, nor a solicitation of an offer to purchase any securities. Furthermore, no securities shall be offered or sold in any jurisdiction where such activities would be contrary to the local securities laws.

Hunterbrook Media LLC is not a registered investment advisor in the United States or any other jurisdiction. We strive to ensure the accuracy and reliability of the information provided, drawing on sources believed to be trustworthy. Nevertheless, this information is provided “as is” without any guarantee of accuracy, timeliness, completeness, or usefulness for any particular purpose. Hunterbrook Media LLC does not guarantee the results obtained from the use of this information. All information presented are opinions based on our analyses and are subject to change without notice, and there is no commitment from Hunterbrook Media LLC to revise or update any information or opinions contained in any report or publication contained on this website. The above content, including all information and opinions presented, is intended solely for educational and information purposes only. Hunterbrook Media LLC authorizes the redistribution of these materials, in whole or in part, provided that such redistribution is for non-commercial, informational purposes only. Redistribution must include this notice and must not alter the materials. Any commercial use, alteration, or other forms of misuse of these materials are strictly prohibited without the express written approval of Hunterbrook Media LLC. Unauthorized use, alteration, or misuse of these materials may result in legal action to enforce our rights, including but not limited to seeking injunctive relief, damages, and any other remedies available under the law.

Authors: Jenny Ahn, Blake Spendley, Sam Koppelman

Editor: Jim Impoco

Based on Hunterbrook Media’s reporting, Hunterbrook Capital is long $HIMX and short a basket of comparable securities at the time of publication. Positions may change at any time. See full disclosures on our website.

Until that moment, it had been a sleepy earnings call.

After detailing the inventory levels of car dashboards and the “seasonal softness” of tablet displays, Himax CEO Jordan Wu sprinkled breadcrumbs that lead straight to the center of the AI revolution.

For over 20 years, Himax Technologies ($HIMX) has been a quiet semiconductor play — a reliable manufacturer of low-margin chips that tell pixels what color to display. To the market, Himax is a relic of the liquid crystal display (LCD) era: the firm that powered the digital photo frame on your grandmother’s mantle and the Face ID on the iPhone X.

But on February 12, Wu signaled a massive opportunity years in the making.

He described an optical product for “the GPU market” that Hunterbrook’s research indicates could one day eclipse his company’s entire $832 million revenue base. He also confirmed a mass-production ramp for a “leading brand’s” smart glasses entering the market this year.

He named neither customer, but the breadcrumbs may be enough.

At Hunterbrook Media — with the help of Citrini Research — we followed a trail of patents, investments, and supply chain shifts. The upshot: The two customers Himax did not name appear to be the two most powerful technology companies in the world. NVIDIA and Apple.

And while investors are still apparently pricing Himax like a commoditized component supplier, the company appears to be emerging as a supplier of a critical optical component powering the next generation of AI data centers.

Getting Light Off the Chip

Last March, Jensen Huang told the world that the networking fabric inside a million-GPU AI factory would consume an amount of electricity rivaling GPUs themselves.

“Energy is our most important commodity,” the NVIDIA CEO said.

Today, fiber optic connections between a network switch and GPU servers rely on a pluggable transceiver, a separate converter box that translates electrical signals to light. A network switch — think of the router in your home, but for AI data centers — routes data between GPU servers.

Each box burns roughly 30 watts. A million-GPU factory requires roughly six transceivers per chip, consuming 180 megawatts of power. To put that in perspective, before a single AI calculation, the networking alone consumes enough electricity to power 150,000 homes.

Huang’s proposed fix: co-packaged optics, or CPO. That’s an optical engine that converts electrons to photons right in the network switch. This eliminates the need for an external transceiver, slashing the power needed for networking by about two-thirds.

The optical engine is made of silicon chips produced by TSMC ($TSM), the same company that manufactures NVIDIA’s AI chips. The engine doesn’t generate light itself. Instead, it encodes data on a continuous laser beam from an external source. The external laser source modules are built by companies like Lumentum ($LITE) and Coherent ($COHR).

Earlier this month, NVIDIA put its money where its mouth was. The company announced $4 billion in investments split between laser suppliers Lumentum and Coherent, with multibillion-dollar purchase commitments attached. Huang called this investment essential to building “gigawatt-scale AI factories.”

The problem: The optical engine emits its data-carrying light straight up — for one, because it’s easier to test that way.

So something then has to bend each laser 90 degrees — and couple the light into dozens of fibers thinner than a human hair that carry data sideways across the data center.

That something is a prism and a grid of tiny lenses that catch each beam and directs it into the fiber.

Until this point, it has been unclear who would build NVIDIA’s essential light-bending devices.

Our reporting indicates it’s Himax.

How Himax Bends Lasers

Rather than grinding each lens one at a time, Himax stamps batches of them onto a glass wafer in one shot — like pressing a waffle iron into batter — in a process called nanoimprint lithography. Lenses on one side, prisms on the other, all perfectly aligned, according to Himax’s manufacturing patents Hunterbrook reviewed. Himax also uses the same batch process to make the baseplates with grooves shaped like a V that holds each fiber in place.

This isn’t the only way to do it, but “really the best method for doing it,” according to a former Lumentum engineer, referring to the lens-prism scheme. That’s because when you’re dealing with dozens of fibers packed tightly together, mechanically aligning each one becomes a nightmare, he explained to Hunterbrook.

Himax’s array of microlenses and prisms and V-groove baseplates are then integrated into a finished fiber array unit (FAU) by a Taiwanese manufacturer Fiber Optic Communications (TPEX: 3363), according to a FOCI patent Hunterbrook reviewed. FOCI also engineers the intricate connection between the FAU and the optical engine.

This joint Himax-FOCI design and manufacturing process will allow Himax and FOCI to enter the supply chain for TSMC’s COUPE architecture, the optical engine that will power NVIDIA’s CPO platform, according to legendary Apple analyst Ming-Chi Kuo of TF International Securities writing on Medium in December 2024.

Himax and FOCI are finalizing this process “in close collaboration with our anchor customer,” Himax CEO Wu said in a recent earnings call. Wu added Himax is able to manufacture them at mass scale already, and FOCI recently raised $100 million to buy equipment to prepare for mass production.

Shortly after Kuo published his analysis, Digitimes cited TF analysts — without naming them — as confirming Himax as the sole microlens array supplier for COUPE’s first and second generations.1

FOCI and Himax are not entirely independent: In June 2024, Himax invested $16 million for a 5.3% equity stake in FOCI. That stake is now worth around $90 million after a roughly 400% appreciation in FOCI’s share price.

But while FOCI’s valuation has multiplied, Himax’s has largely stagnated — with investors seemingly unaware of its critical role in NVIDIA’s pipeline.

“I think it’s 0% in the stock,” said the physician-turned-investor behind the new Substack Latent Value, which debuted with deep dive research about Himax.

He pointed to Himax’s history of sharp re-ratings once a new product cycle starts. $HIMX hit over $15 during the Google Glass cycle, then surged again on Microsoft HoloLens news.

“It’s at the bottom of the cycle just on the legacy stuff. So that’s what makes it really compelling — even on the legacy business,” he said.

If the Nvidia partnership is confirmed, he says, “the numbers get silly.”

He compared Himax to other companies in NVIDIA’s supply chain like Lumentum and Coherent, similar photonics stocks that have multiplied in value. “Once you’re proven to be in the supply chain for AI, as we’ve seen, stocks can easily 10x and 20x.”

The Patent Trail

The physical evidence goes beyond analyst conjecture.

Four patents — two filed by Himax, one by FOCI, one by TSMC — describe what appears to be the same FAU design and manufacturing process for CPO application. The documents were first unearthed by Latent Value.

A Himax patent published in January details the nanoimprint lithography process for manufacturing the fiber array.2

Then there is TSMC’s December 2025 patent for a FAU inspection tool built specifically for CPO application. The patent describes the schematic of the part being tested — without naming any supplier. But it’s the exact geometric structure that Himax describes: a prism/mirror on one face, microlenses on the other, sitting above a baseplate with fibers. The shape matches. And TSMC’s decision to build a dedicated inspection system for this specific component suggests it’s planning to use it at scale.

And then there is a FOCI patent, filed in late 2024, that adds the finishing touch to what the FAU-COUPE integration might look like.

It describes how to assemble the components of the FAU — the lens array, the prism that bends light, and the V-groove plate — all things that Himax makes. More critically, it describes exactly how the FAU will connect to the COUPE.

The connection is done through the “chip end lens group,” which has a lens part that:

• Sits at the bottom of the FAU assembly (chip-facing side)

• Can “change the path of the light beam through total reflection” and makes it “reflect and travel upward”

• “Can be connected to a chip.”

That’s exactly the way TSMC describes its CPO-FAU integration in a recent research paper (also on its website) explaining the optical plumbing on its COUPE platform.

The paper describes a component called COIT — short for Complimentary Optical Interconnect at Top — inside the FAU, equipped with the “high efficiency 90-degree beam deflector” that “couples light to fiber arrays” in the FAU. This is in line with Himax’s design. And, according to TSMC, the COIT is optically coupled with microlenses inside the COUPE.

The paper also says that this method is preferable to the alternative in which the beam is emitted from the side of the optical engine — suggesting TSMC’s decision to choose the light-going-up version was not made in a vacuum, but with a specific FAU in mind to make this version work.

The former Lumentum engineer interviewed by Hunterbrook reviewed the patents — and came away with the conclusion that “there is a high probability that Himax and FOCI are part of the FAU supply chain.”

He cited three main benefits of its technology.

First: “It is a batch process. There are multiple units on a glass wafer that can be processed together.” Second, this process has a lower cost than the traditional lithography process. Third, he added, the optical block using prisms also has higher precision than a mechanical process.

He suspects Himax “is also making some of those microlenses being used in the FAU. That’s why they wrote the wafer level testing patent.”

The patents also reveal intriguing — if speculative — clues to the Himax, FOCI, TSMC supply chain centering on what may be a custom FAU with an unusual specification: a 22-lens array.

In Himax’s January manufacturing patent, a figure illustrating the setup shows the optics block with exactly 22 lenses, if you count the circles representing the lenses.

And a separate patent filed by Himax for testing its prisms says the prism under test “may also be equipped with multiple lenses, such as 16, 22, or 50 lenses, but the number is not limited thereto.”

Twenty-two is an strange number in fiber optics, suggesting it might be custom design. Fiber counts are standardized around specific multiples like 12 or 16; both are numbers you’d see in product specs. Higher-density arrays generally scale in increments of those base units — 24, 32, 48, 72.

But there it is again, in the TSMC’s patent — the closest thing to a smoking gun, though still speculative. TSMC’s inspection schematic explicitly depicts 22 channels, labeled “ch1” through “ch22.”

Twenty-two channels appearing in both patents may be a coincidence. But it’s interesting because it appears to describe the same part being inspected at two points in the supply chain: by Himax before shipping, and by TSMC after assembly, before it reaches a GPU.

Another plus factor: The company’s short distance from TSMC, as it flexed on a skyline photo of its wafer fab campus posted on its website, suggests the parts wouldn’t be traveling very far. FOCI is just a seven minute drive away as well.

How big is the opportunity?

Morgan Stanley projects 5,000 next-generation NVIDIA Rubin Ultra racks shipping in 2027 and 28,000 in 2028.

Even using conservative pricing assumptions, the FAU market scales from hundreds of millions in 2027 to billions of dollars in 2028, according to Citrini Research’s analysis. And NVIDIA’s Feynman architecture — the generation planned for after Rubin — is expected to multiply the unit count further.

The Apple Opportunity

On the earnings call last month, Wu, the Himax CEO, disclosed that a top smart glasses brand would enter mass production by the end of 2026.

Himax has described three core technologies that could become essential to this space: its WiseEye AI sensing processor, its tiny projection display known as LCoS microdisplay, and its compatibility with waveguide — the optical layers that make see-through displays possible.

Wu has not named the smart glasses brand. He might not have needed to.

DigiTimes reported in early February that Apple’s expected smart glasses entry is reshaping the Taiwanese augmented reality (AR) optics supply chain, with multiple suppliers increasing capex in anticipation. The Apple analyst Ming-Chi Kuo, known for predicting the company’s releases, anticipates mass production beginning soon en route to a 2027 launch, a similar timeline to the one reported by Bloomberg.

Kuo projects Apple’s first smart glasses will ship 3 million to 5 million units in the launch year. These would likely be audio only, not requiring extensive purchases from Himax. And then, the potential transformative opportunity would arrive in 2028, when Kuo says Apple is expected to launch display-equipped AR glasses using LCoS microdisplays and waveguides. That’s where Himax would come in.

Himax’s relationship with Apple would not be new.

In 2017, Himax was reported (see this, this, this, and this) as a key supplier of the wafer-level optics that power Face ID on the iPhone X — the tiny dot-projector lenses that map a user’s face in three dimensions.

In the smart glasses supply chain, Himax may occupy a specific niche.

Karl Guttag is a former Texas Instruments Fellow with 40 years of experience in display technology, who founded an LCoS company and now runs KGOnTech, a popular independent technical blog in the AR/VR industry. He’s tracked Himax since the Google Glass era, when his identification of Himax as its LCoS supplier allegedly moved the company’s market cap by more than $100 million in a single day.

He described Himax to Hunterbrook as “a little bit the McDonald’s of the LCoS world” — not the highest quality, but able to produce massive volume. “They’ve never had the best LCoS,” Guttag said. “They’re kind of the low-cost guy.”

Historically, Himax has partnered with top companies on microdisplays, including the original Microsoft HoloLens and Google Glass and other AR platforms. More recently, at CES 2026, Himax and smart glasses maker Vuzix ($VUZI) unveiled an optical reference design pairing Himax’s ultra-compact Front-lit LCoS microdisplay with Vuzix’s waveguide.

If involved with Apple, Himax would “probably” be supplying “the LCoS and the waveguide,” Guttag conjectured. Himax’s wafer fab facilities — the same ones that fabricate microlens and V-groove baseplates using nanoimprint lithography — have already churned out thousands of waveguide displays in the past, and Wu recently talked on earnings calls about using its facilities for fabricating waveguide optics for other companies’ designs. But Guttag also questioned the fit: “Himax has never demonstrated the highest quality LCoS nor the highest quality waveguides,” he told Hunterbrook.

Guttag noted that Apple may be driven by capacity goals: Himax is “in the best capacity situation” among LCoS makers, he said. He suggested Apple’s choice may reflect pragmatism over perfectionism: “Maybe they’re more controllable, or they had the capacity.”

Although there is no direct confirmation that Himax has once again partnered with Apple, there’s circumstantial evidence — including on LinkedIn.

One profile of interest allegedly belongs to someone describing themselves as a Himax senior product manager with the title “Advanced Optics & NPI Integration | Apple Project Lead.” This individual — whose identity is labeled as “verified” on LinkedIn — claims to have previously led the Meta Quest 3 lens program, scaling production to over 7 million units.

Their recent posts describe shipping ahead of customer deadlines and resolving late-stage manufacturing defects, language consistent with a program approaching production ramp.

Then, as Hunterbrook prepared to publish this article, they posted again: “Visiting Apple Park this week for technical discussions and collaboration.”

Hunterbrook geolocated the image and tracked the aircraft, EVA’s B-16716, revealing that the senior product manager appears to have flown from Taipei to San Francisco on Monday, March 9.3

Asked whether it could confirm its relationships with Apple and Nvidia, Himax did not respond to Hunterbrook’s request for comment, and TSMC declined to comment. The purported employee on LinkedIn who posted the flight did not respond to direct messages. And Hunterbrook has not been able to match their identity on any other social media platforms — though we did find a master’s thesis on a Taiwanese academic research database from a student of the same name from the same graduate school at the same rough time period this purported Himax employee lists on LinkedIn. LinkedIn also claims the account was “verified” with a government ID.

And on Monday, the employee posted again: “Day 1 at Apple Park — a full day of technical discussions and collaboration, from 11am to 9pm. Productive start to the week.”

One of the images attached to the post? A selfie in front of Norman Foster’s signature Apple Park design, including the internal arbor of the headquarters, known as “the spaceship.” This area is closed to the public.

We geolocated the image. It’s clearly inside of the secure perimeter.

“I’m amazed that somebody would post something like that,” said Guttag.

Authors

Sam Koppelman is a New York Times best-selling author who has written books with former United States Attorney General Eric Holder and former United States Acting Solicitor General Neal Katyal. Sam has published in the New York Times, Washington Post, Boston Globe, Time Magazine, and other outlets. He has a B.A. in Government from Harvard, where he was named a John Harvard Scholar and wrote op-eds like “Shut Down Harvard Football,” which he tells us were great for his social life.

Jenny Ahn joined Hunterbrook after serving many years as a senior analyst in the US government. She is a seasoned geopolitical expert with a particular focus on the Asia-Pacific and has diverse overseas experience. She has an M.A. in International Affairs from Yale and a B.S. in International Relations from Stanford. Jenny is based in Virginia.

Blake Spendley joined Hunterbrook from the Center for Naval Analyses (CNA), where he led investigations as a Research Specialist for the Marine Corps and US Navy. He built and owns the leading open-source intelligence (OSINT) account on X/Twitter, called @OSINTTechnical (over 1 million followers), which also distributes Hunterbrook Media reporting. His OSINT research has been published in Bloomberg, the Wall Street Journal, and The Economist, among other top business outlets. He has a B.A. in Political Science from USC.

Editor

Jim Impoco is the award-winning former editor-in-chief of Newsweek who returned the publication to print in 2014. Before that, he was executive editor at Thomson Reuters Digital, Sunday Business Editor at The New York Times, and Assistant Managing Editor at Fortune. Jim, who started his journalism career as a Tokyo-based reporter for The Associated Press and U.S. News & World Report, has a master’s in Chinese and Japanese History from the University of California at Berkeley.

Graphic

Dan DeLorenzo is a creative director with 25 years reporting news through visuals. Since first joining a newsroom graphics department in 2001, he has built teams at Bloomberg News, Bridgewater Associates, and the United Nations, and published groundbreaking visual journalism at The Wall Street Journal, Associated Press, The New York Times, and Business Insider. A passion for the craft has landed him at the helm of newsroom teams, on the ground in humanitarian emergencies, and at the epicenter of the world’s largest hedge fund. He runs DGFX Studio, a creative agency serving top organizations in media, finance, and civil society with data visualization, cartography, and strategic visual intelligence. He moonlights as a professional sailor working toward a USCG captain’s license and is a certified Pilates instructor.

Hunterbrook Media publishes investigative and global reporting — with no ads or paywalls. When articles do not include Material Non-Public Information (MNPI), or “insider info,” they may be provided to our affiliate Hunterbrook Capital, an investment firm which may take financial positions based on our reporting. Subscribe here. Learn more here.

Please contact ideas@hntrbrk.com to share ideas, talent@hntrbrk.com for work opportunities, and press@hntrbrk.com for media inquiries.

LEGAL DISCLAIMER

© 2026 by Hunterbrook Media LLC. When using this website, you acknowledge and accept that such usage is solely at your own discretion and risk. Hunterbrook Media LLC, along with any associated entities, shall not be held responsible for any direct or indirect damages resulting from the use of information provided in any Hunterbrook publications. It is crucial for you to conduct your own research and seek advice from qualified financial, legal, and tax professionals before making any investment decisions based on information obtained from Hunterbrook Media LLC. The content provided by Hunterbrook Media LLC does not constitute an offer to sell, nor a solicitation of an offer to purchase any securities. Furthermore, no securities shall be offered or sold in any jurisdiction where such activities would be contrary to the local securities laws.

Hunterbrook Media LLC is not a registered investment advisor in the United States or any other jurisdiction. We strive to ensure the accuracy and reliability of the information provided, drawing on sources believed to be trustworthy. Nevertheless, this information is provided “as is” without any guarantee of accuracy, timeliness, completeness, or usefulness for any particular purpose. Hunterbrook Media LLC does not guarantee the results obtained from the use of this information. All information presented are opinions based on our analyses and are subject to change without notice, and there is no commitment from Hunterbrook Media LLC to revise or update any information or opinions contained in any report or publication contained on this website. The above content, including all information and opinions presented, is intended solely for educational and information purposes only. Hunterbrook Media LLC authorizes the redistribution of these materials, in whole or in part, provided that such redistribution is for non-commercial, informational purposes only. Redistribution must include this notice and must not alter the materials. Any commercial use, alteration, or other forms of misuse of these materials are strictly prohibited without the express written approval of Hunterbrook Media LLC. Unauthorized use, alteration, or misuse of these materials may result in legal action to enforce our rights, including but not limited to seeking injunctive relief, damages, and any other remedies available under the law.

By: Michelle Cera, Sam Koppelman, Andrew Ford, Laura Wadsten

Editor: Jim Impoco

Hunterbrook Media’s investment affiliate, Hunterbrook Capital, does not have any positions related to this article at the time of publication. Positions may change at any time. Full disclosures on our website.

When Hunterbrook Media reported that RadNet ($RDNT) appeared to artificially inflate its same-center sales growth, the company didn’t dispute the findings. It just stopped releasing the number altogether, according to its latest annual SEC filing, released on Monday.

In December 2025, Hunterbrook published a deep dive led by Bethany McLean into RadNet, the radiology roll-up that has repositioned itself as an AI growth story. A central finding: The company’s same-center sales growth — a metric Wall Street pointed to as evidence of RadNet’s special sauce — appeared inflated by the undisclosed consolidations of nearby imaging centers.

This week, RadNet released its latest annual report. The same-center metric … disappeared altogether.

The contrast is stark. In RadNet’s 2024 10-K, “same-center” or “same center” showed up 18 times. This phrase appears a grand total of zero times in RadNet’s 2025 10-K filing.

Notably, RadNet’s most recent unaudited statement, provided in a press release, does still refer to same-center procedure volumes — reporting 9.6% growth in advanced imaging and 4.5% overall in the fourth quarter.1 But those numbers do not allow investors to quantify how much revenue the company is generating year over year on a center-by-center basis.

The change raises an obvious question: If RadNet’s same-center growth were as strong as management repeatedly told investors, why stop reporting it?

It’s the latest example of a company whose accounting practices Hunterbrook has found to be creative — and, potentially, concerning. And now, Hunterbrook’s findings are being corroborated from the inside.

We spoke with multiple former RadNet accounting employees who described an accounting department under pressure to present favorable numbers — and a leadership team whose goal, as one former employee put it, was ”to present a better financial picture than what’s truly accurate.”

Radnet did not reply to Hunterbrook’s request for comment. But at an investor conference this afternoon, RadNet’s CFO Mark Stolper was explicitly asked for his response to allegations in an unnamed report — presumably Hunterbrook’s — that scrutinized same-center growth.

“So just give me your perspective on that report, and just anything you’ve done to maybe change how you present information, to make yourself, inoculate yourself from future [inaudible] on that front,” the moderator said.

Stolper did not address the changes in RadNet’s 10-K and tried to brush aside the concerns raised by Hunterbrook.

“If we do close down a center and we move that volume to an existing center, we keep that volume in the prior period. So we’re not gaming the system at all, which I think was the concern from that report that came out,” Stolper replied.

“So we feel very, very comfortable in how we look at same center performance, and it’s been done consistently with the same team since the beginning of time,” he said. “And ultimately, you know, our GAAP revenue, our audited GAAP revenue and financial statements reflect that growth.”

How RadNet Juiced Its Numbers

Our original investigation reconstructed RadNet’s shifting footprint with archived web data, SEC filings, and independent financial analysts.

The reporting found that RadNet had been closing imaging centers located within a 15-minute drive of other RadNet sites, likely consolidating patients into surviving locations. These closures were excluded from the same-center calculation — but the revenue that transferred to nearby RadNet locations was not.

The result was a metric that made RadNet look like it was growing organically at 6% to 10% per year. A financial analysis shared with Hunterbrook estimated that after removing the artificial boost from consolidations, real same-center growth was closer to 2.5% to 3% — roughly in line with what RadNet’s own CFO, Mark Stolper, described as “normalized” growth before the company’s AI rebrand.

@hunterbrookmediaTwo imaging centers on paper but only one dot on the map. We followed the trail through RadNet’s financials—and its AI narrative isn’t the only thing that raises questions. Read our investigation at the link in the bio

Tiktok failed to load.

Enable 3rd party cookies or use another browser

On this week’s earnings call, RadNet was asked by an analyst about its growth outlook.

Stolper responded by citing “same-center” performance, claiming it is typically modeled at 3% to 5% — and citing a long-term baseline of 2% to 4%, which often jumps higher.

But he never provided the same-center revenue growth metric RadNet just deleted from its 10-K.

“A Dumpster Fire”: Inside RadNet’s Accounting Department

Since the December investigation, Hunterbrook has spoken with multiple former RadNet accounting employees who worked in the company’s Los Angeles headquarters. Their accounts paint a picture of an accounting department with a chaotic internal culture and a leadership team focused on optics over accuracy.

A former accounting manager who worked at RadNet for two years described “adjustments made on a quarterly basis” to management estimates to make the company’s financial picture look better than it was.

The former employee said that on one occasion, when a manager refused to book a $2.1 million entry to cover underestimated audit fees, leadership simply bypassed them to find a more compliant manager. “Management estimates shouldn’t be changed whenever they feel like it,” the former manager said, adding that estimates were changed “quite frequently and quite drastically.”

The same former employee described a top-down corporate culture where financial statements would change because “adjustments would come through from the highest of the high up and then we just had to book them.”

“And there would be no questions asked…it would get booked and then the numbers would change with no other reason or cause,” they added.

Asked how RadNet’s practices compared to what they’d experienced at another public company, the former employee called it “night and day.” At RadNet, they said, there were “last-minute adjustments to get things to work.” They described the overall tone from leadership as “atrocious,” adding, “They have a goal and that’s to keep the company going, but they also will do whatever it takes to have that happen.”

Another former employee in RadNet’s accounting department, who worked closely with the Digital Health team, described the transactions between divisions as deeply confusing.

“There were so many intercompany transactions that I didn’t even know how they were even reconciling it,” they said. Invoices were miscoded, expenses were booked to closed imaging centers, and managers were rotating so quickly that institutional knowledge was lost, according to the interview.

“Nobody really knew what to, like, do,” they said, adding, “It was a dumpster fire.”

They recalled confusion around imaging centers closing and being absorbed, as well as pressure to cater to directors and VPs. “We were sometimes bullied into doing things that we didn’t particularly think was right,” they said, noting that directors finagled reports in an effort to “roll up better.”

This former employee also described being asked to make minor adjustments to run rates: “I remember actually the VP was saying, telling me, we have to get this right because our share price is kind of going down, too.”

They also described being told to reclassify certain expenses as capitalized assets — a move that would reduce reported expenses and increase reported assets. “Assets going up and expenses going down sounds like a great thing to have happen,” they said, adding that the Digital Health team “always wanted it to be like these things aren’t expenses but capitalized assets.”

The former employee said that at the executive level, leadership “did not care about how it was done as long as it was getting done. They don’t care that it was done right. They just cared that it was being done beneficially for the company … or maybe the best presentation for the company.”

Both former employees recounted a mass exodus of accounting staff, including managers, in 2025.

The Story That Doesn’t Add Up

The decision to quietly drop same-center reporting fits a pattern Hunterbrook has documented: a company whose disclosures shift in ways that make independent verification harder, not easier.

RadNet’s 2025 10-K does show Imaging Center segment revenue of approximately $1.99 billion, up 10.9% from $1.79 billion in 2024 — but investors can no longer see how much of that growth came from existing locations versus acquisitions and consolidations. The company attributes growth to higher fees per procedure, increased procedure volumes, and incremental volumes from acquired centers. But without quantified same-center data in audited financials, it’s impossible to distinguish organic performance from roll-up math.

The same-center question is just one thread in a broader pattern our December investigation uncovered. RadNet’s much-hyped Digital Health division — the foundation of the company’s AI rebrand — generated less than 5% of total revenue in 2025. A former accounting employee was skeptical of this rebrand: “Especially when a company’s kind of touting that they’re going to be the revolutionary change for AI,” they mentioned, questioning whether RadNet “is even capable of implementing something like that in a way that is per generally accepted accounting principles.”

Hunterbrook also reported that RadNet’s disclosures around its center count were riddled with inconsistencies. Its 2024 10-K said 44 centers were opened through internal development; the earnings release said nine. One quarterly filing reported 398 centers; a later filing restated the same period at 375, with no explanation.

Before its AI rebrand, RadNet traded at roughly 8 to 10 times adjusted EBITDA, in line with peers. It now trades at around 18 times — and closer to 24 times once stock compensation and non-capitalized R&D are factored back in. If you strip out the core imaging business at even a generous multiple, the implied valuation of the Digital Health unit tops $3 billion — a staggering price for a division whose products are struggling to find buyers outside of RadNet’s own walls.

Authors

Laura Wadsten is an investigative journalist specializing in healthcare. She began her career reporting on antitrust and healthcare as a Correspondent for The Capitol Forum, a premium financial publication. Laura was a Hodson Scholar and Editor-in-Chief of The News-Letter at Johns Hopkins University, where she earned a B.A. in Medicine, Science & the Humanities.

Sam Koppelman is a New York Times best-selling author who has written books with former United States Attorney General Eric Holder and former United States Acting Solicitor General Neal Katyal. Sam has published in the New York Times, Washington Post, Boston Globe, Time Magazine, and other outlets — and occasionally volunteers on a fire speech for a good cause. He has a B.A. in Government from Harvard, where he was named a John Harvard Scholar and wrote op-eds like “Shut Down Harvard Football,” which he tells us were great for his social life. Sam is based in New York.

Andrew Ford is an investigative journalist who exposed systemic flaws and prompted reforms in healthcare, business, policing, and state government. His reporting was published by ProPublica, USA Today, The Arizona Republic, Asbury Park Press, and Florida Today. He holds a journalism bachelor’s from the University of Florida and received a data analytics master’s from Georgia Institute of Technology. He is based in Phoenix, AZ.

Michelle Cera trained as a sociologist specializing in digital ethnography and pedagogy. She completed her PhD in Sociology at New York University, building on her Bachelor of Arts degree with Highest Honors from the University of California, Berkeley. She has also served as a Workshop Coordinator at NYU’s Anthropology and Sociology Departments, fostering interdisciplinary collaboration and innovative research methodologies.

Editor

Jim Impoco is the award-winning former editor-in-chief of Newsweek who returned the publication to print in 2014. Before that, he was executive editor at Thomson Reuters Digital, Sunday Business Editor at The New York Times, and Assistant Managing Editor at Fortune. Jim, who started his journalism career as a Tokyo-based reporter for The Associated Press and U.S. News & World Report, has a master’s in Chinese and Japanese History from the University of California at Berkeley.

Hunterbrook Media publishes investigative and global reporting — with no ads or paywalls. When articles do not include Material Non-Public Information (MNPI), or “insider info,” they may be provided to our affiliate Hunterbrook Capital, an investment firm which may take financial positions based on our reporting. Subscribe here. Learn more here.

Please contact ideas@hntrbrk.com to share ideas, talent@hntrbrk.com for work opportunities, and press@hntrbrk.com for media inquiries.

LEGAL DISCLAIMER

© 2026 by Hunterbrook Media LLC. When using this website, you acknowledge and accept that such usage is solely at your own discretion and risk. Hunterbrook Media LLC, along with any associated entities, shall not be held responsible for any direct or indirect damages resulting from the use of information provided in any Hunterbrook publications. It is crucial for you to conduct your own research and seek advice from qualified financial, legal, and tax professionals before making any investment decisions based on information obtained from Hunterbrook Media LLC. The content provided by Hunterbrook Media LLC does not constitute an offer to sell, nor a solicitation of an offer to purchase any securities. Furthermore, no securities shall be offered or sold in any jurisdiction where such activities would be contrary to the local securities laws.

Hunterbrook Media LLC is not a registered investment advisor in the United States or any other jurisdiction. We strive to ensure the accuracy and reliability of the information provided, drawing on sources believed to be trustworthy. Nevertheless, this information is provided “as is” without any guarantee of accuracy, timeliness, completeness, or usefulness for any particular purpose. Hunterbrook Media LLC does not guarantee the results obtained from the use of this information. All information presented are opinions based on our analyses and are subject to change without notice, and there is no commitment from Hunterbrook Media LLC to revise or update any information or opinions contained in any report or publication contained on this website. The above content, including all information and opinions presented, is intended solely for educational and information purposes only. Hunterbrook Media LLC authorizes the redistribution of these materials, in whole or in part, provided that such redistribution is for non-commercial, informational purposes only. Redistribution must include this notice and must not alter the materials. Any commercial use, alteration, or other forms of misuse of these materials are strictly prohibited without the express written approval of Hunterbrook Media LLC. Unauthorized use, alteration, or misuse of these materials may result in legal action to enforce our rights, including but not limited to seeking injunctive relief, damages, and any other remedies available under the law.

1) Hunterbrook Media’s Blake Spendley is doing extraordinary work covering the news in Iran — follow him on X at @OSINTtechnical.

2) The new CTO of Big Bear AI announced his resignation over the weekend, weeks after a Hunterbrook Media story on his increasingly vitriolic LinkedIn posts amid competition from OpenAI, Anthropic, and xAI. Read our reporting.

By: Jeremy Arnold

Editor: Jim Impoco, Sam Koppelman

Hunterbrook Media’s investment affiliate, Hunterbrook Capital, has no positions related to this article at the time of publication. Positions may change at any time. Our contributor has made extensive filings with the SEC and FINRA for potential securities fraud and anti-touting violations. If you’re an affected investor or former employee of any of these companies and want to tell us about it, write ideas@hntrbrk.com.

For the version of the article with detailed footnotes, read it on our website.

In 2012, Congress and the Securities and Exchange Commission passed a law that would open the doors to equity crowdfunding. For the first time, buying shares in private startups wasn’t just for the wealthy and financially sophisticated. Restrictions designed to protect amateur investors were relaxed as part of a mission to “democratize” finance. No longer would ordinary people be forced to watch from the sidelines as the next SpaceX or Instagram rocketed to the moon.

Or so the sales pitches went. What actually followed was exactly what skeptics predicted: a parade of failing companies raising hundreds of millions from unsuspecting investors through campaigns built on misdirection and deceit. Success stories proved rare. Disasters became routine.

A Hunterbrook Media investigation reveals how a network of distressed startups, a hungry marketing firm, and some of America’s most trusted financial media brands — wittingly or not — ended up separating small investors from their money. A remarkable number of respected outlets, from blue-chip financial newsletters to The Wall Street Journal, appear to have hosted misleading or downright false ads, calling into question their due diligence and adherence to disclosure obligations. The result looks like an unhappy masquerade: troubled companies partnering with slick marketers to hide their rotten balance sheets behind the prestige of reputable mastheads, and news outlets indirectly promoting the very companies they should have been scrutinizing.

The little guy finally had his seat at the table. The game was rigged.

By following the trail from the narrative factory to the newsletter inbox, a clear pattern emerges of how these companies manufactured the appearance of success to mask the reality of failure.

Here’s how it worked:

• Manufacturing the pitch. DealMaker Marketing Services , an arm of financial firm DealMaker, collected fat fees for helping early-stage companies — including Boxabl, HeartSciences, EnergyX, M2i Global, Pacaso, RAD Intel, and Miso Robotics — craft ads and landing pages that stretched the truth, sometimes beyond the breaking point.

• ‘Trust-washing’ These ads appeared in premium slots in newsletters with millions of subscribers — like Morning Brew, Robinhood’s Sherwood News, and The Wall Street Journal’s 10-Point.

• Last ones in the pool. While readers were frequently told that they’d be investing alongside venture capital firms and other financial professionals, these companies largely turned to retail investors after deteriorating circumstances forced their hands. The amateurs essentially bailed out the pros, buying heavily-restricted shares at jaw-dropping and opaque valuations.

While we’re publishing now to call attention to the scale of the problem, we believe we’ve captured here only a fraction of all concerning cases. In our review of additional examples, we struggled to find any that didn’t exhibit similar concerns.

Pay-to-Print Coverage

Hunterbrook approached all eight media outlets named here for comment. Only two replied. Their statements are included below.

Intermediaries like DealMaker didn’t — and couldn’t — pull this off alone. Instagram ads and landing pages can only go so far. Media legitimization is crucial.

Say you’re reading your morning newsletter from a trusted source. There’s a section titled “Presented By” that otherwise looks the same as the rest. Is this an ad? Who wrote it? At the least you might assume that the outlet in question did meaningful due diligence on your behalf — as some of them claim to do.

Few outlets count on this simple trust more than Morning Brew. A hotshot daily newsletter network with over 4 million claimed subscribers, its rapid rise led to an acquisition by publishing giant Axel Springer, parent company of both Politico and Business Insider. Hunterbrook’s analysis found that Morning Brew ran these crowdfunding ads in 165 editions of their flagship newsletter in 2025 alone. It also continued to run them regularly even after we approached for comment — to which we still have yet to receive a reply.

But it wasn’t just them. We found dozens of concerning examples across newsletters from Sherwood News (operated by a subsidiary of investment company Robinhood), The Wall Street Journal, 1440 (over 4.6 million claimed subscribers), TLDR (1.6 million) and Tangle News (460,000).

Tangle, an independent publication, provided Hunterbrook with a long and earnest response. Its founder affirmed that it did conduct both internal and referral-based due diligence. This doesn’t appear to have extended to comparing the provided ad copy to issuers’ disclosure documents. Tangle is reviewing its policies.

Benzinga, which boasts about 25 million monthly readers, went even further with a promotional post, seemingly presented as its own independent research, that was mostly just the fundraising company’s marketing boilerplate. Exec Sum, a financial newsletter run by Twitter/X personality Litquidity, did the same for its more than 350,000 subscribers.

In response to a request for comment, Benzinga shared the following statement: “We value Hunterbrook’s reporting on a variety of investigative topics like insider trading and other financial improprieties. We always endeavor to improve practices in conjunction with our legal teams, regulators, issuer clients, and general public inquiries.”

The effect of this legitimatization is evident in the numbers: Of the $75 million that startup EnergyX raised in 2024, DealMaker attributed “over $50 million” to these sponsored ads.

The pitches tended to follow a common theme: making small investors feel like they were joining something larger than just an investment. In a panel interview with DealMaker’s CEO and three fundraising founders, Morning Brew’s cofounder Alex Lieberman waxed poetic on this point:

“There are a lot of times where the word community is used but it isn’t necessarily authentic community,” he said. “Making retail investors part of your community, via investment in your business, [that] is the purest form.”

Two Truths and Some Lies

Note: We’ve included three short explainer sections as an investing and legal crash course. Seasoned market players can skip down to the deal profiles.

There are two classic ways to raise money: sell a better mousetrap or tell a better story. But what happens when you only have the story — and it’s not an especially good one? What if growth has stalled, losses have piled up, and professional investors have pulled back?

It’s no crime to fail in business. Startups should try hard things, with informed investors buying as much risk as they can stomach. What matters is what companies do after things go awry. If they turn to retail investors for a bailout, what story do they tell them?

We found that these companies’ legal disclosure filings were largely honest and typically painted a bleak picture of their position and prospects. But placed between investors and these unwieldy documents were layers of ads and landing pages — written in plain English, with far rosier outlooks, and, sometimes, outright falsehoods.

Regulatory Primer

U.S. securities law divides investors into two classes:

• Accredited investors, who meet special criteria around wealth, income, and financial licenses. The government assumes they have more investing experience, along with more loss cushion.

• Non-accredited (“retail”) investors, who may individually have deep financial expertise, but who on aggregate are considered a class worth protecting from potentially predatory deals.

In light of historical grift, the government has long restricted non-accredited investors from buying certain securities — like penny stocks and most private company shares — without special waivers or certifications. Uncle Sam prefers at least some guardrails.

The JOBS Act, passed in 2012, partially reversed this trend by permitting companies to directly approach retail investors with Kickstarter-style campaigns offering shares instead of early products. This legislation — including later amendments — gave companies two new options to mix-and-match:

• Regulation CF campaigns can raise up to $5 million per rolling year

• Regulation A+ campaigns can go as high as $75 million per rolling year

Because these campaigns target less affluent and sophisticated audiences (who can invest up to certain caps), the ads that issuers and their paid partners may run for these campaigns are subject to disclosure requirements and other restrictions:

• Ads for offerings made pursuant to Reg CF give the issuer a choice: They can either provide the details of the offering, but then only extremely limited factual details about the company and its business; or they can describe the business, but not the specific terms of the offering.

• Though Reg A+ campaign ads may include sales copy alongside offering terms, they are of course subject to general anti-fraud rules, such as Section 17(a) of the Securities Act, which prohibits “obtaining money or property” by means of false or misleading statements.

• Under either program, the content must make clear that it is a paid promotion, and, according to at least one lawyer we spoke to, must also include the amount of compensation received.

Dozens of the ads we reviewed for this investigation seemingly failed to follow the first and second rules, and, while most contained some marker identifying them as paid content, the level of clarity of those disclosures varied widely. And no ads disclosed the specific compensation received for their placement.

Brenda Hamilton, a securities lawyer and founder of Hamilton & Associates Law, described this shortcoming to Hunterbrook bluntly: “The public is entitled to know the information was bought and paid for and for how much.”

A Scorecard

To illustrate what these investors were truly buying, each deal profile covers a few key metrics.

1. Revenues

Did the company have meaningful paying customers yet?

While startups are often unprofitable in their early days, it’s important to know if they’re actually serving customers at all. Many of these marketing campaigns relied heavily on projections, obscuring multiple years of operations with little or zero sales traction.

2. Book Value

How much was the company worth on paper?

Accountants use the term “net tangible book value” to capture what a company’s hard assets are worth, minus all its financial liabilities. While this is always an incomplete picture — as it doesn’t factor expected growth — it provides a useful baseline.

3. Offered Valuation

What true price were new investors buying in at?

The SEC doesn’t require these companies to give investors a straight answer here. While their full legal disclosures must provide the raw inputs, investors are left to do the math themselves. This number is far more crucial than the offered price per share, as comparing a company’s offered valuation to its book value uncovers how much of a markup investors are paying for uncertain growth.

While infant startups still building their foundations can reasonably charge substantial markups, higher multiples become harder to justify with each passing year that a company burns cash without getting closer to that growth materializing — or, as in some cases here, after revenues fall sharply.

Deal Profiles

Each of the deal profiles below proceeds in three sections — first, we provide our scorecard metrics; second, we describe the company and the content of its crowdfunding ads; third, we identify examples of false or misleading claims made either in those ads or related marketing materials. Hunterbrook approached all seven featured companies regarding its findings. Only one replied. Its comments have been included below.

Methodology: All book values were calculated as of the start of each company’s 2025 financial year. As none are turning a net profit, these figures largely reflect the cash they’ve accumulated from investors. The listed offered valuations are pre-money and don’t factor bonus shares.

#1 – Boxabl

Stats: 2024 Revenues: $3.4 million. Book Value: $52.6 million. Offered Valuation: $3.5 billion. Markup: 66.6x.

Boxabl makes modular homes that can be collapsed for shipping like hulking IKEA boxes. The company claims to build each base-model Casita in “nearly four hours” using its “advanced assembly line” system, and boasts a reservation list of more than 190,000.

The company’s legal filings, however, tell a less optimistic story. Only 7,732 of those deposits are backed by cash deposits of any size, and few depositors have ever actually completed a purchase. Boxabl sold just eight Casitas over the first nine months of 2025, all while writing off an additional 68 of its hundreds of stuck inventory units as either too damaged or now obsolete.

The company’s early operations also ran into a critical snag: It hadn’t secured required state approvals. Even today it can only deliver standalone units to 15 states , some of which require further local permissions.

The company nonetheless still plans to go public via a SPAC merger at a $3.5 billion valuation.

Concern #1

• Claim: Boxabl sold recent investors on a “first of its kind” 2022 partnership with homebuilding giant D.R. Horton, which allegedly included a “phase 1 order for 100 Casitas.”

• Reality: No Boxabl SEC filings reviewed by Hunterbrook have listed D.R. Horton as a delivery recipient. As a Hunterbrook analysis only identified 15 total units shipped to unlisted buyers, this order appears to have been either cancelled or heavily trimmed. And not for lack of inventory, as over 60% of Boxabl’s built homes remain unsold.

Concern #2

• Claim: Potential investors were told that Boxabl had delivered over 600 units by early 2025.

• Reality: According to its own SEC filings, through Nov. 10, 2025, Boxabl had delivered only 293, of which 207 went to two early buyers in 2022.

#2 – HeartSciences

Stats: 2024 Revenues: $4,000. Book Value: -$1.4 million. Offered Valuation: $5.0 million. Markup: ∞.

HeartSciences ($HSCS) has partnered with multiple hospital research teams to bring AI into the business of ECG heart readings. The company was founded in 2007 and has yet to commercialize its technology.

When will it finally go to market? Despite telling investors as late as September 2025 that it was still targeting initial FDA approval “this year,” it had yet to even file its application when those statements were made. It only did so in mid-December, months too late for a possible 2025 approval.

Concern #1

• Claim: “Wall Street just raised its price target. Some analysts were projecting a $14 one-year share price target for HeartSciences (Nasdaq: HSCS). Now, they think it’ll hit $15.”

• Reality: HeartSciences’s website lists three analysts that cover its stock. All three work for firms that have had financial relationships with the company. Of 12,135 Wall Street analysts evaluated for their prediction quality, the trio rank 11,978th, 12,068th, and 12,066th. Also, none of them raised their target to $15 as claimed. That was an old evaluation, which had been downgraded five months before this ad went out.

Concern #2

• Claim: “With one-year price targets as high as $14, buying our investment unit for $3.50 gives you over 500% upside potential at that target.”

• Reality: Beyond the controversial nature of the target-setting, this math includes a basic error: At a $14 share price, the potential net return (factoring the included warrant) was about half of the quoted 500%. Though this is also academic, as the alluded-to analyst has since trimmed his target to $9.20 — which is still more than three times higher than the stock’s current trading price.

#3 – RAD Intel

Stats: 2024 Revenues: $1.2 million. Book Value: -$1.8 million. Offered Valuation: $149.7 million. Markup: ∞.

RAD Intel, “the ChatGPT of marketing,” spent $2.9 million on its own sales and marketing in 2024. The payoff? About $1 million in revenue over the first half of 2025 — on which it lost $6.2 million. The company operates without a permanent office and appears to make most of its money from selling shares to new investors.

Up to $11.8 million of its current fundraising will go to cashing out existing shareholders, including significant holdings from the company’s CEO, president, and board of directors.

(Captured from Morning Brew’s Nov. 10, 2025 edition.)

Concern #1

• Claim: RAD Intel’s investor portal included a press quote that it sourced to Rolling Stone. It also used the magazine’s logo next to a broader claim about “buzzing” press coverage.

• Reality: The unlinked excerpt quotes from sponsored editorial content, not actual reporting. Rolling Stone itself has seemingly never written about RAD Intel. A separate glowing snippet from Fast Company was similarly from a paid advertorial.

Editor’s note: Though RAD Intel never replied to Hunterbrook’s request for comment, it subsequently made quiet changes to its current offering website to clarify where it was quoting from advertorials.

Concern #2

• Claim: Multiple crowdfunding ads claimed that RAD Intel was “backed” by both Adobe and Fidelity Ventures.

• Reality: Adobe gave the company an $8,000 grant to fund design work, but is neither an investor nor backer. Fidelity Ventures, a former venture capital firm, also never invested. Rather, two funds managed by Fidelity Investments Canada, which is not a venture capital firm, joined a 2018 bridge round for a Canadian startup that was later folded into RAD Intel.

Editor’s note: The company similarly updated this language after Hunterbrook’s reachout.

#4 – M2i Global

Stats: 2024 Revenues: $0. Book Value: -$2.5 million. Offered Valuation: $105.1 million. Markup: ∞.

Mineral Metals Initiatives (M2i Global) is more a constellation of ideas than a company. Originally an app development firm, its only reported revenues — around $4,400 total — came from “AI generated tattoo ideas.” It now has vague plans of helping America pull together a critical mineral stockpile.

Though M2i co-released a press release in August 2025 announcing “a transformative public-private initiative to develop and operate the United States’ first Strategic Minerals Reserve,” its CEO clarified a day later, in a YouTube video, that this wasn’t “a done deal.” Months later, it still had yet to sign a storage lease.

M2i ($MTWO), which currently trades as a penny stock, has filed for an NYSE American listing via a reverse merger with an aviation company that, rather poetically, claims to be developing AI for evaluating financial filings. This merger may be complicated by the fact that, in December 2025, M2i lost to a default judgment in a lawsuit it claimed to have not known about. The court awarded the plaintiffs $18 million and 100 million of the company’s shares. That same month, M2i withdrew its Regulation A+ offering after raising $723,000.

Concern #1

• Claim: “Now, backed by federal partnerships and a recent $850m deal, this move sets M2i Global up for national scale.”

• Reality: Hunterbrook found no evidence that M2i has any observable federal partnerships. Additionally, the “$850m deal” was a right for M2i to buy copper offtake up to a fixed amount. It’s also likely moot: M2i’s counterparty, NT Minerals, announced the sale of its interests in the underlying mining area almost seven weeks before this ad went out.

Concern #2

• Claim: “We project our scrap and recycling operation will grow revenue from $3.5M to $8.3M within two years.”

• Reality: While web archives show this claim circulating as early as Jan. 21, 2025, the company’s total 2025 revenues through Nov. 30 were $0. It also continues to own no such operation.

#5 – Pacaso

Stats: 2024 Revenues: $126.5 million. Book Value: $60.3 million. Offered Valuation: $1.5 billion. Markup: 24.9x.

Pacaso sells real estate units that it stresses aren’t timeshares. Its “fractional ownership” solution merely allows buyers to acquire one-eighth shares in vacation homes that Pacaso manages on their behalf.

From a revenue standpoint, this business peaked in 2022. As for profitability, that depends on whether you place weight on Pacaso’s emphasized gross profits (over $120 million to date) or its actual bottom line ($214.8 million in accumulated net losses over the past 4.5 years, with an ongoing cash burn of about $1 million every eight days.)

Over the first six months of 2025, Pacaso spent $9.4 million on ads convincing retail investors to buy in — all while its revenues fell by $23.6 million compared to the same period in 2024.

Concern #1

• Claim: “2024 was a record setting year for Pacaso, reaching a total of over $110M in gross profits and 41% YoY growth.”

• Reality: Beyond the misleading nature of gross profit headlines for a cash-incineration business, this $110 million-plus figure is (non-obviously) cumulative. Pacaso only booked $21.5 million in gross profits in 2024 itself, which was not a record. It also turned this gross profit into a net loss of $31.4 million.

Concern #2

• Claim: “Meet $PCSO. They aren’t publicly traded, but Pacaso just reserved their Nasdaq ticker, $PCSO.”

• Reality: Hunterbrook found that The Wall Street Journal ran this line seven times in its 10-Point newsletter. What’s omitted: Tickers are reserved via a free online form and indicate neither concrete plans to pursue a public listing nor likely success in doing so. Investors were buying illiquid shares in a company with crashing revenues and mountainous losses, not advance tickets for a traditional IPO.

#6 – EnergyX

Stats: 2024 Revenues: $385,000. Book Value: $64.6 million. Offered Valuation: $2.1 billion. Markup: 32.4x.

EnergyX is one of several firms racing to develop direct lithium extraction (DLE) technology to replace traditional brine pool evaporation. The company’s first demonstration plant entered pilot operations in the fourth quarter of 2024.

Though EnergyX’s CEO, Teague Egan, claimed in late 2023 that the company would alternate its fundraising between crowdfunders and professionals, it has raised just $262,000 through isolated private sales since, compared to over $100 million from three retail campaigns.

A 2025 DealMaker case study quotes Egan as having spurned venture capitalists who’d argued for lower valuations. This can be read as a self-believing founder fighting for a vision that only he could see. It can also be read as professional money managers being more skeptical of his projections.

Retail investors aren’t experts in industry due diligence. When they read in a Morning Brew ad that EnergyX is a good buy because General Motors alone expects to need “414,469 tons” of lithium carbonate to meet its 2035 EV target, they’re unlikely to realize that this is enough for roughly 7 million EVs. Or that GM — excluding its Chinese joint venture — sold under 4 million total vehicles in 2025, an estimated 5% of which were EVs. They might also not notice that the price of lithium carbonate remains roughly 70% below its 2022 highs.

Concern #1

• Claim: “EnergyX’s patented tech extracts lithium 300% better than conventional methods.”

• Reality: This doesn’t seem mathematically possible. Lithium recovery rates for modern evaporation ponds start at 40% and average about 50%. While yields are lower in unusually magnesium-rich fields in Bolivia, EnergyX’s bid to operate there was rejected in 2022 in favor of competing firms also using DLE technologies. Even if the company extracted at 100% efficiency, doubling the typical 50% baseline would be a maximum improvement of 100%, not 300%.

Concern #2

• Claim: In an October 2025 investor webinar, Egan hyped EnergyX’s steady share price performance: “We went from $8 to $9 and then $9 to $9.50. The shares are at $10 today.” A few minutes later he added that early crowdfunding investors had gotten in at 41 cents per share.

• Reality: There’s a subtle bait-and-switch here. EnergyX “arbitrarily” increased its own valuation in between some of its fundraising rounds, then used share splits to smooth out the impact to its share price. For example, its earliest crowdfunders paid $4.90 per share. Those shares were then split 12-to-1, leaving investors with 12 shares for each one they initially purchased. This is how Egan arrived at a retroactive 41-cent price. But when discussing the more recent increase from $9.50 to $10, Egan left out that those who paid $9.50 ended up with twice as many shares after the March 2025 split. So those paying $10 at the time of the webinar were effectively paying more than double per share, not just 50 cents more. The selective disclosure strategy seems to have worked. Per a DealMaker case study, these webinars raised several million dollars in new investments.

#7 – Miso Robotics

Stats: 2024 Revenues: $384,676. Book Value: $5.5 million. Offered Valuation: $282 million. Markup: 51.6x]

Miso Robotics makes robot arms that fry fast food on behalf of human workers. Its sole marketed product today is the Flippy Fry Station.

“It looks like their main business operation is fundraising and not robot development,” said Tim McLaughlin, a Miso investor. Hunterbrook reviewed an email he sent the company in 2024 to cancel his investment. Miso wasn’t obligated to refund him, and he said the firm never replied. It has gone on to raise over $35 million more since.

Concern #1

• Claim: “Demand for the new third generation of our AI-powered Flippy Fry Station robot is real. Our first production run sold out in a week.”

• Reality: Miso didn’t sell the Flippy Fry Station at the time. It leased the units, with “no upfront costs.” It also had a maximum of three in the field when this sell-out claim first circulated. Over a year later it still had just 14 of these units deployed. It also announced that its pilot with Jack in the Box ended last November with the removal of its final remaining unit.

In a detailed statement, Miso’s CEO, without disputing Hunterbrook’s analysis of the number of units, argued that “sold out” was appropriate given lease commitments for “all available inventory.”

Concern #2

• Claim: In the “2025 onwards” section of its roadmap, as a separate item from its Flippy expansion, Miso’s Innovation Lab was listed as “expanding experimental products for future growth.”

• Reality: While asking investors to credit it for this “future growth” potential, the company simultaneously disclosed in its legal filings that it had discontinued all non-Flippy Fry Station products and R&D for cost discipline. For emphasis, it added that it would only reconsider if and when “a restaurant partner [commits] to also sharing the development costs of a specific product.”

In his statement, Miso’s CEO outlined incremental technical and support improvements being made to and for the Flippy Fry Station, which Hunterbrook understands to be a single, already-existing product.

The Connecting Thread: DealMaker

In 2018, two corporate fundraising attorneys decided to spin up a new firm, DealMaker, to focus on these new crowdfunding programs. They saw a future in helping startups tell their stories better, especially to retail investors looking for difference-making companies that might notice and value them. As its CEO, Rebecca Kacaba, put it, “People want to make investments in companies that they believe in, that they believe are making a great difference in the world.”

While this is a lovely sentiment, DealMaker makes no bones about who the company actually serves. DealMaker’s customers are the startups, and if those customers want funding then it’s the firm’s job to help. Kacaba was clear on this in a 2024 interview: “We’re not an investor-focused platform.” The firm’s job is to appeal to investors, not to advocate for them.

DealMaker’s strategy worked, with over $2 billion raised for their clients to date across all offering types, including from “hundreds of thousands of individuals and families.”

One key to DealMaker’s success: slick ads. In 2022, it acquired Austin-based agency Ridge Growth, which now operates as its in-house marketing arm. All clients that use DealMaker for their fundraising campaigns have the option of paying extra for help crafting their campaign content — from website copy to email pitches to the sponsored ads supplied to media partners like Morning Brew.

Editor’s note: DealMaker Marketing Services was contracted by all the above issuers, with DealMaker Securities also serving as a broker-dealer.

While DealMaker itself routinely disclaims responsibility for the accuracy of the campaigns it designs, its own client contracts suggest it is deeply involved with the content of the materials:

Again quoting DealMaker’s CEO:

“Our digital marketing team … they do this stuff all the way down to the analytics. They’re monitoring the reactions on the first emails you send, and tweaking the messaging and understanding how to position it to get that deal done.”

But how far did the team tweak? It’s one thing to put your client’s best foot forward, and another entirely to misrepresent key claims to an audience you know is mostly unsophisticated investors.

Especially when you know how ugly the returns are likely to be.

A Legacy of Losses

When the SEC finalized the new rules for these crowdfunding programs in late 2015, DealBook coverage from the New York Times included a dark assessment:

“Ninety-nine percent of these deals will prove to be unprofitable,” said Andrew Stoltmann, a lawyer who specializes in securities fraud. “This is a disaster waiting to happen.”

Young companies need efficient pathways to raise money. There was nothing wrong with giving them more and better options. The foreseeable disaster was adverse selection: Why would a healthy company want the hassles of crowdfunding — e.g., slow financing, bad industry optics, and unwelcome public disclosure requirements — if professional investors were fighting to give them checks? And if the best companies were unlikely to have interest, which companies would?

We saw the same pattern over and over again in our investigation: a company started off with a credible idea, raised money from sophisticated investors, and then failed to deliver results. It was often only then that they found a sudden interest in giving retail investors a seat at the table.

Some of these companies may yet turn the corner and return some money, though history indicates that most will not. Crowdfunding investors can typically only sell their shares when a company goes public or gets acquired, which only 1.2% of all issuers had done as of April 2024.

Even these exceptions are still mostly unhappy stories. To avoid cherrypicking, Hunterbrook analyzed a list from a University of Southern California researcher of Regulation A and CF startups that had filed for a public listing by the end of 2022. Some never crossed the finish line, and those which did all fared poorly. The best of them saw its valuation collapse by 90% before a late crypto pivot partially reversed its fortunes.

DealMaker itself has a similar record. Its two self-touted public listings from 2024 — Sky Quarry ($SKYQ) and Autonomix ($AMIX) — are now down 87.7% and 99.6% respectively from their debut prices. Hunterbrook was only able to identify four other DealMaker issuers that went on to have an exit event. Newsmax ($NMAX) is already trading down 40.5% from its 2025 IPO, Knightscope ($KSCP) and Trust Stamp ($IDAI) are both down over 99%, and Monogram was acquired for a loss of 44.3% along with a promise of uncertain future payments. Despite its name, DealMaker appears to still be chasing a deal that proves successful for its investors.

Of course, staying private doesn’t necessarily mean failure. What about the companies that haven’t gone public or been acquired but really have met their loftiest growth expectations? A recent Morning Brew ad for Pacaso leveraged just such a success story, celebrating a “400x buyout offer” for early crowdfunding investors in the fintech startup Revolut.

While Revolut wasn’t a DealMaker client, there’s a larger issue with this reading: These lucky few investors didn’t actually get to sell all their shares. They were stuck with largely paper gains for years as professional investors and company insiders sold ahead of them. They learned the hard way that most of these deals are like lobster traps: While everyday investors don’t need to wait to get in, they must often wait a long time to get out — if they ever can.

Authors

Jeremy Arnold is the publisher of The Save Journalism Committee, a newsletter focused on journalism reform. He has also worked as a communications consultant for Silicon Valley startups that fundraised on their real merits. He went to school in Canada “and got very good grades.”

Additional research by Daniel Ptashny.

Editors

Sam Koppelman is a New York Times best-selling author who has written books with former United States Attorney General Eric Holder and former United States Acting Solicitor General Neal Katyal. Sam has published in the New York Times, Washington Post, Boston Globe, Time Magazine, and other outlets — and occasionally volunteers on a fire speech for a good cause. He has a B.A. in Government from Harvard, where he was named a John Harvard Scholar and wrote op-eds like “Shut Down Harvard Football,” which he tells us were great for his social life. Sam is based in New York.

Jim Impoco is the award-winning former editor-in-chief of Newsweek who returned the publication to print in 2014. Before that, he was executive editor at Thomson Reuters Digital, Sunday Business Editor at The New York Times, and Assistant Managing Editor at Fortune. Jim, who started his journalism career as a Tokyo-based reporter for The Associated Press and U.S. News & World Report, has a master’s in Chinese and Japanese History from the University of California at Berkeley.

Graphic

Dan DeLorenzo is a creative director with 25 years reporting news through visuals. Since first joining a newsroom graphics department in 2001, he has built teams at Bloomberg News, Bridgewater Associates, and the United Nations, and published groundbreaking visual journalism at The Wall Street Journal, Associated Press, The New York Times, and Business Insider. A passion for the craft has landed him at the helm of newsroom teams, on the ground in humanitarian emergencies, and at the epicenter of the world’s largest hedge fund. He runs DGFX Studio, a creative agency serving top organizations in media, finance, and civil society with data visualization, cartography, and strategic visual intelligence. He moonlights as a professional sailor working toward a USCG captain’s license and is a certified Pilates instructor.

Hunterbrook Media publishes investigative and global reporting — with no ads or paywalls. When articles do not include Material Non-Public Information (MNPI), or “insider info,” they may be provided to our affiliate Hunterbrook Capital, an investment firm which may take financial positions based on our reporting. Subscribe here. Learn more here.

Please contact ideas@hntrbrk.com to share ideas, talent@hntrbrk.com for work opportunities, and press@hntrbrk.com for media inquiries.

LEGAL DISCLAIMER

© 2026 by Hunterbrook Media LLC. When using this website, you acknowledge and accept that such usage is solely at your own discretion and risk. Hunterbrook Media LLC, along with any associated entities, shall not be held responsible for any direct or indirect damages resulting from the use of information provided in any Hunterbrook publications. It is crucial for you to conduct your own research and seek advice from qualified financial, legal, and tax professionals before making any investment decisions based on information obtained from Hunterbrook Media LLC. The content provided by Hunterbrook Media LLC does not constitute an offer to sell, nor a solicitation of an offer to purchase any securities. Furthermore, no securities shall be offered or sold in any jurisdiction where such activities would be contrary to the local securities laws.

Hunterbrook Media LLC is not a registered investment advisor in the United States or any other jurisdiction. We strive to ensure the accuracy and reliability of the information provided, drawing on sources believed to be trustworthy. Nevertheless, this information is provided “as is” without any guarantee of accuracy, timeliness, completeness, or usefulness for any particular purpose. Hunterbrook Media LLC does not guarantee the results obtained from the use of this information. All information presented are opinions based on our analyses and are subject to change without notice, and there is no commitment from Hunterbrook Media LLC to revise or update any information or opinions contained in any report or publication contained on this website. The above content, including all information and opinions presented, is intended solely for educational and information purposes only. Hunterbrook Media LLC authorizes the redistribution of these materials, in whole or in part, provided that such redistribution is for non-commercial, informational purposes only. Redistribution must include this notice and must not alter the materials. Any commercial use, alteration, or other forms of misuse of these materials are strictly prohibited without the express written approval of Hunterbrook Media LLC. Unauthorized use, alteration, or misuse of these materials may result in legal action to enforce our rights, including but not limited to seeking injunctive relief, damages, and any other remedies available under the law.

By: Matthew Termine, Sam Koppelman, Michelle Cera

Editor: Jim Impoco

Based on Hunterbrook Media’s reporting, Hunterbrook Capital is short $HTGC and long a basket of comparable securities at the time of publication. Positions may change at any time. See full disclosures on our website.

• Hercules Capital ($HTGC) is among the most software-exposed business development companies (BDCs), yet it trades at some of the richest valuations. About 35% of the value of the company’s loan portfolio — roughly $1.5 billion — is in software debt, meaning that after accounting for leverage, 70 cents of every dollar of net asset value is backed by software. Despite billions worth of such debt across the industry falling into distressed territory, as of its most recent filings Hercules still marks its software book at100 cents on the dollar. And true software exposure may be higher than 35%. Hunterbrook Media found the firm categorizes companies like Houzz and AppDirect outside of software despite their products being explicitly software-based.

• A growing share of income is phantom. Income from a type of debt called payment-in-kind (PIK) loans doubled in two years to $55.9 million in 2025, while the actual cash collected on PIK loans collapsed 73% year over year to just $4.9 million. PIK loans enable borrowers to “pay” interest by adding debt, but the lender does not actually receive interest payments along the way: The outstanding PIK receivable at Hercules has nearly tripled in recent years to $109.1 million — interest that has never been paid in cash by borrowers.

• The dividend’s real cushion is thinner than it appears. Full-year NII of $1.91 per share covers the $1.60 base distribution at 119%, but Hercules also pays a $0.28 supplemental drawn from a finite reserve of past earnings. Including the supplemental, total distributions of $1.88 leave just three cents of margin. On a cash basis — stripping out PIK and other noncash items — Hercules’ own figures show the base distribution is covered at only 105.7%. If Hercules takes write-downs, even the base dividend may come under pressure.

• Hercules runs a premium-dependent flywheel that could reverse catastrophically. Like Strategy (formerly MicroStrategy) with bitcoin, Hercules issues stock above NAV that funds accretive growth. The premium has already compressed from about 60% to about 25%. If it hits zero, equity issuance turns destructive, loan growth can stall, and the dividend may come under pressure — the same kind of doom loop that hit competitor TriplePoint ($TPVG).

• The portfolio doesn’t cycle as fast as management claims. Despite describing 18-to-24-month turnover — assuaging fears surrounding the terminal value of its borrowers — Hunterbrook found that a significant share of funding events since 2022 went to existing borrower relationships, based on an analysis of 16 quarters of earnings-call disclosures, including some relationships that have lasted almost a decade. Hercules told Hunterbrook in an email that 63% of new fundings went to new portfolio companies.

• A former Hercules finance team member described an overstretched valuations process. A team responsible for preparing valuations that fed directly into financial statements consisted of just four people in a single reporting line responsible for dozens of companies, they said, with few checks or cross-team review — a contrast to the multilayered processes the same employee now sees at another public company. Another former employee described an ad hoc deal sourcing process, in which Hercules would follow name-brand VCs into deals — a process the former employee likened to the swarm around Sam Bankman-Fried’s FTX.

• The company’s headline credit metric may be an invention that flatters its track record. Hercules’ cumulative net realized loss rate of 3 basis points nets equity and warrant gains from IPO-era exits (DocuSign, Palantir, Pinterest) against loan losses. We couldn’t identify any other major BDCs using this methodology.

• Hercules Capital provided Hunterbrook Media with a written response to questions. The company said that its portfolio valuations as of December 31, 2025, do not reflect 2026 volatility; that its PIK exposure is at the median of similarly sized BDCs; that its base dividend coverage is among the highest in the industry at 119%; that the majority of new fundings since 2022 went to new portfolio companies; that its valuations process is “rigorous and cross-functional” involving PwC audits, SOX compliance, and board oversight; and that its credit metric is required by GAAP. Hercules’ responses are incorporated throughout this article.

SaasPocalypse: Now

The software sector has lost an estimated trillion dollars in market value so far in 2026. It’s the largest, fastest destruction of software market cap in years. And private-credit firms — like Blue Owl Capital ($OWL) and Ares Management ($ARES) — have fallen with the borrowers of their managed funds.

“Private Credit Is Headed for a Software Shock,” read a Bloomberg headline on February 2. Three weeks later, the shock penetrated the public consciousness when Blue Owl — one of the largest private-credit managers in the world, with over $300 billion in assets under management — halted quarterly redemptions at one of its retail-focused business development corporations.

BDCs lend money to mid-sized and smaller companies, often companies that can’t easily access traditional bank financing. The BDC’s shares can be publicly traded, giving everyday investors access to private credit.

According to a Bloomberg index, $18 billion in broadly syndicated software loans have plunged into distressed territory — trading below 80 cents on the dollar — in a matter of weeks. UBS recently estimated private-credit default rates could soar to 15%. KKR’s BDC just cut its dividend earlier in the week.

But according to the books of lender Hercules Capital ($HTGC), everything is … totally fine?

The firm values its software loan portfolio, comprised of loans to smaller, often unprofitable startups, at more than 100 cents on the dollar — even in its 10-K released on February 12.

In a written response to Hunterbrook Media, Hercules stated that its figures are as of December 31, 2025, and “therefore do not reflect any potential pricing volatility that may have occurred in 2026.1”

Hercules describes itself as the largest nonbank venture lender in the United States. The firm has committed over $25 billion since launching in 2003. The founder, Manuel Henriquez, resigned as CEO in 2019 after being charged in the college admissions fraud Operation Varsity Blues.2 Hercules has been led since by Scott Bluestein, who also serves as its chief investment officer.

In total, Hercules has software debt valued at $1.5 billion on its books, according to its financial statements: roughly 35% of the value of its total loan portfolio. That means 70 cents on every dollar of its net asset value (NAV) is backed by a software loan — because Hercules, like other BDCs, uses leverage to fund its portfolio, in its case at roughly a 1:1 ratio.

According to JPMorgan’s recent assessment of the sector, this makes Hercules the most software-exposed lender of any meaningful size in America. And these numbers likely underestimate true exposure. Hunterbrook Media analyzed each loan Hercules disclosed in its most recent annual report and found that the firm assigns certain businesses that describe themselves as software companies to categories outside of software.

Take Houzz, which declares on its homepage that it offers, “Powerful software for construction and design.”

Another company that Hercules classifies as unrelated to software is AppDirect. That company describes itself as “the leading platform for selling, buying, and managing recurring technology services.” In other words: It’s a middleman for software companies, as the company lays out pretty clearly on its website.

And Hercules has grown more exposed to software over time.

In the firm’s 3Q25 investor deck, on a slide about rebalancing risks based on market conditions, Hercules reveals it has gone from 24% software exposure to 35% software exposure — a nearly 50% increase — in just four years. Its next-largest category: Drug Discovery & Development.

And yet Hercules trades at around a 25% premium to book value — the highest of any software-exposed BDC in the country depicted in a recent JPMorgan North American Credit Research report.

Charting premium and software exposure, Hercules sits alone in the upper-right corner: where substantial software concentration and a significant premium intersect.

A graph of premium to NAV (a proxy for valuation) relative to software exposure based on the JPM coverage universe. Note: BDCs JPM categorizes as having 0% software exposure — such as $MAIN — were excluded. Price per share data as of February 26, 2026.

Hunterbrook found every other BDC with more than 20% software concentration in its loan portfolio in the JPMorgan coverage universe trades at a discount to NAV or a far smaller premium. TriplePoint Venture Growth ($TPVG), a smaller venture lender, trades at an approximate 35% discount to NAV.

Why is one of the most exposed lenders the most expensive? And what happens if it loses its premium to NAV?

The Saylor Spiral

The inherent risk of the Hercules valuation is perhaps best explained by the Infinite Money Glitch of Michael Saylor’s Strategy (née MicroStrategy).

Here’s how it worked: Strategy traded at a massive premium to the value of its bitcoin holdings — at one point, investors were reportedly paying nearly $2.80 for every $1.00 of bitcoin on the balance sheet. That premium was the engine. Because the stock traded above the value of the underlying asset, every time Saylor issued new shares, he was effectively raising capital at a massive markup. He’d sell the expensive stock, buy more bitcoin, increase the bitcoin-per-share, and the stock would go up — which widened the premium — which let him issue more stock. Flywheel. Magic. Infinite money glitch.

Then bitcoin cracked. The premium vanished. By November 2025, Strategy fell below its net asset value. The stock, which peaked above $450, cratered 70% — the worst performer in the entire Nasdaq 100 in 2025. And here’s the doom loop: Once the premium disappears, the flywheel reverses. Issuing stock below NAV destroys value for existing shareholders. So you can’t issue stock as easily. So you can’t buy as much additional bitcoin. So the growth narrative dies. So the stock falls further. So the premium compresses more.

Hercules runs the same flywheel — just with software loans instead of bitcoin.

When Hercules trades at 1.25x NAV, every dollar of new equity it raises comes at a 25% premium to book value. That’s accretive to existing shareholders and provides cheap capital for new loans. Premium stock price → cheap equity → fund new loans → generate income → support dividend → maintain premium. Elegant.

Hercules noted to Hunterbrook that it has “conservatively raised less than $250 million of new public equity per year since FY22” and has also raised nearly $940 million in unsecured debt to fund growth since 2024. The company emphasized that as an internally managed BDC, “Hercules shareholders do not pay a management fee and Hercules management does not earn more by simply increasing Hercules’ asset base through the issuance of new shares.” Hercules said it primarily issues equity to maintain a conservative debt-to-equity ratio within regulatory limits and below levels required by its four investment-grade ratings. The company also pointed to the growth of Hercules Advisor-managed private funds as an additional source of capital for growth.

But as at Strategy, the flywheel spins in both directions.

If Hercules stock falls to NAV or below, the company can’t issue equity. And without equity issuance, Hercules can fund fewer new originations and extensions to existing portfolio companies, because BDCs are statutorily limited in how much leverage they can take on relative to the cash on their balance sheets.

This is the precise sequence that hit one of Hercules’ closest competitors. TriplePoint Venture Growth lost its premium and then cut its dividend. Hercules has maintained both its dividend and a premium to NAV — but the premium has narrowed from about 60% to about 25% in under a year.

If it hits zero, the Saylor spiral could begin.

The Marks Haven’t Caught Up

Hercules values its $1.5 billion software portfolio at around par — which is to say: The company doesn’t expect meaningful losses. But the borrowers behind those loans operate in a sector that has just experienced a systemic pullback.

And nobody knows what software companies will be worth when the debt comes due.

As JunkBondInvestor, a widely followed buy-side credit analyst, wrote this week: “What is a software company worth in five years? Not what it earns today or next quarter. Not what the sponsor paid for the business. What it’s actually worth when the AI labs are telling their investors they plan to replace Salesforce, Workday, and Adobe … Nobody can answer that question. Nobody.”

And the collateral underlying these loans — software code, brand equity, customer relationships — could ultimately be worth essentially nothing.

When Hercules’ software borrower Khoros hit a rough patch, Hercules was forced to write a $61 million loan down to $18 million. With software companies, generally, there is no factory to liquidate, no meaningful real estate to foreclose on. The collateral is code and customers. When the latter evaporates, the former can be rendered worthless.

Historically, the majority of Hercules’ portfolio companies have repaid their debt. But amid the sector decline, a borrower-by-borrower review of Hercules’ debt positions in 55 software firms as disclosed in Hercules’ most recent financials reveals a portfolio with pockets of pressure.

• Sisense, a business intelligence platform, has cut its workforce in half, suffered a severe data breach, and doesn’t seem to have publicly raised external funding since early 2020.

• Marigold Group’s debt is structured as a payment-in-kind (PIK) loan at a very high rate — Hercules collects zero cash interest — after the company already sold off its enterprise business.

• Ceros, a no-code content platform, faces existential AI disruption. Similar names, like Wix and GoDaddy, have been killed in the public markets, each down more than 50% in the last year.

There are strong companies on the Hercules books as well.

Armis, the cybersecurity platform, is being acquired by ServiceNow for $7.75 billion in cash — a deal expected to close in the second half of 2026 that will almost certainly result in full loan repayment plus a meaningful return on the firm’s $2 million equity investment. Carbyne, a 911 emergency communications platform that Hercules helped fund, is being acquired by Axon for $625 million, with the deal expected to close in early 2026. AlphaSense, the AI-powered market intelligence platform, raised $650 million at a $4 billion valuation in 2024.

But the sourcing process for those deals was not exactly scientific, according to a former Hercules analyst who worked on deal sourcing.

“I got into the meeting with my manager and I was like, ”How do you do sourcing? Like, what’s the juice? Like, what’s your secret sauce?’ And he was like, ‘Go on the website for Google Ventures and just see what they invest in and just copy it. I don’t want anything else.’”

“That’s actually a really bad way to do your due diligence,” they added. “And that’s actually a way that a lot of these, you know, like, FTX and all these things, that’s how it happens because once these companies get one brand name on their investor list, right? Like Sequoia or something like that? Everybody else just freaking flocks to them.”

And once Hercules makes the loans, the valuation process itself may warrant scrutiny. A former member of Hercules’ finance team w described a small, overstretched team with few checks in place.

“There wasn’t a whole lot of fail-safes or backup checks,” they said. “We would have kind of one stream of resources and backups,” compared to other public companies where “there are multiple streams and cross-checking and cross-team development.”

According to this former employee, a team responsible for preparing certain valuations that fed directly into the financial statements consisted of four people arranged in a single reporting line — an analyst, a senior, a manager, and a director — handling an important subset of valuations for Hercules.

“I don’t think the team got the support that it needed,” they said. “It was a small team and it was stretched quite thin.”

Review was concentrated at the manager level. By the time work reached the director, the former employee recalled seeing few comments or corrections. The former employee, who now works in a similar role at another public company, described a starkly different environment.

“There is a strong push to do things the right way, to reinvent, to make sure that we’re double-checking, triple-checking.” Not so at Hercules.

Hercules told Hunterbrook that the valuations in its February filing are as of December 31, 2025, and don’t reflect volatility in the first quarter of 2026 — leaving open the possibility that the numbers will look different when it files for Q1. It also described its valuations process as “a rigorous and cross-functional exercise involving our deal teams responsible for portfolio origination and monitoring, credit team and finance team, who re-underwrite each position in the portfolio on a quarterly basis.”

The company noted it is subject to an integrated external audit by PwC, internal audit under SOX regulations, an annual SOX review, and has adopted Investment Company Act Rule 2a-5 for valuation governance. Hercules said its independent board of directors approves all fair value marks, and that the company uses two external valuation specialists each quarter to independently value a rotating portion of the portfolio, as “overseen by the Audit Committee.”

A Metric That May Mask the Truth

The secret sauce of Hercules’ investor pitch is an interesting statistic: a cumulative net realized loss rate of only 3 basis points across $21.6 billion in total commitments since inception.

Hunterbrook reporting suggests this metric may be an invention unique to Hercules. It nets equity and warrant gains — windfalls from IPO-era exits like DocuSign, Palantir, and Pinterest — against loan losses. It is as if a bank reported its mortgage default rate after subtracting gains from its stock portfolio. We couldn’t identify any other major BDCs using this methodology.

Ares Capital — the BDC affiliate of Ares Management, which holds a portfolio almost seven times the size of Hercules — reports its equity gains as a separate line item and leads with standard credit metrics like nonaccrual rates and interest coverage ratios.

Strip out the warrant gains and the picture changes meaningfully.

The company’s cumulative net realized loss position was negative $134.7 million through 3Q25.

And the safety net that made the metric work is nearly gone.

The warrant and equity portfolio has shrunk from 11% of total assets in 2020 to about 4% today. Going forward, debt losses hit the bottom line with virtually no equity gains to soften the blow.

The former member of the finance team called the practice of using this metric “quite odd.”

“I don’t think that they should be netting that against their equity and warrants,” they said, adding that as a public company, Hercules should be able to articulate “a good rationale as to why they are doing that” if it diverges from how peers in the industry handle the same reporting.

Hercules said it is required by the Investment Company Act of 1940 and GAAP to report net realized gains and losses on investments including equity and warrants, and that “excluding these equity gains and losses from the calculation would not be in compliance with GAAP.” The firm said it has used consistent methodology under independent accounting oversight for more than 21 years.

A Growing Share of Income Is Phantom

Then there’s the rising tide of “phantom income.”

In 2025, Hercules booked $55.9 million in payment-in-kind interest income. It collected $4.9 million in cash — an amount equal to roughly 9% of the PIK income booked during the year, compared to 36% the year before.

Payment-in-kind interest, or PIK, is a mechanism that allows borrowers to defer cash interest payments by capitalizing the interest owed onto the loan’s principal balance. The lender records it as income. But the cash may never arrive. For a venture lender concentrated in software, rising PIK is a credit canary: When borrowers stop paying in cash, it often means they can’t.

Hercules’ PIK income has more than doubled in two years, from roughly $24.7 million in 2023 to $55.9 million in 2025. It now accounts for 11% of total interest and dividend income, up from approximately 5.7% just two years ago.

More important than the annual run rate is what’s piling up on the balance sheet: The outstanding PIK receivable — accumulated interest sitting on Hercules’ books — reached $109.1 million as of Dec. 31, 2025. That’s nearly triple the $38 million balance at the end of 2023.

And the 10-K’s own rollforward table reveals a telling detail: Hercules collected just $4.9 million in actual cash payments on PIK loans during 2025, a 73% collapse from the $18.3 million collected in 2024.

In the firm’s disclosure, “Payments on PIK loans are normally received only in the event of payoffs.”

Translation: This cash comes in only when a borrower gets acquired, IPOs, or refinances — events that are scarce in a venture downturn. Until then, the receivable just keeps growing.

This means Hercules is now owed more than $109 million in deferred interest payments subject to the same credit risk of the underlying loans, and the rate at which it’s actually collecting that cash has cratered. This phantom income inflates dividend coverage ratios used to justify the valuation.

Management attempted to contextualize the PIK exposure on a recent earnings call, noting that approximately 86% of Q4 PIK income was attributable to original deal underwriting rather than credit-related amendments, and that 91% came from loans rated highly on the company’s internal scale.

In its written response to Hunterbrook, Hercules said its PIK exposure relative to revenue was “approximately the median of similarly-sized BDCs and is below several of our direct competitors.” The company attributed approximately 86% of its 4Q25 PIK income to PIK that was “part of the original underwriting and not a result of any credit or performance-related amendment,” adding that nearly 91% of PIK income in the quarter came from loans it had rated as less risky. Hercules said lower cash collections on PIK loans in 2025 versus 2024 reflected “a higher volume of prepayments of loans with a PIK component in 2024 versus 2025,” noting that PIK payments are collected at contractual maturity or upon prepayment.

In its 10-K, Hercules noted that PIK receivable represented approximately 3% of total debt investments as of year-end, up from 2% a year earlier.

But this framing obscures the core issue: Regardless of whether PIK was baked into the original deal or added later, or whether 2024 saw higher prepayments than 2025, it remains noncash income owed by cash-burning borrowers. The fact that PIK was part of the original underwriting simply means Hercules is intentionally structuring more loans with deferred cash payments — which is itself a competitive concession that increases credit risk.

The Portfolio Can’t Cycle Out as Fast as Hercules Claims

When asked about software concentration risk, management often points to turnover — the idea that the portfolio cycles quickly enough that exposures can simply be rotated out. CEO Scott Bluestein has even described the portfolio as turning about every 18-to-24 months.

The implication: If software goes bad, Hercules can simply cycle out.

But that framing appears to count loan amendments and extensions as completed cycles and obscures the fact that most borrower relationships last far longer than 18 months.

A quarter-by-quarter read of Hercules’ own earnings-call disclosures shows why the distinction matters. On call after call, Hercules shares that the majority of its new fundings were to existing borrowers.

Hercules disputed these figures in aggregate, stating that “since 2022, 63% of all new fundings have been to new portfolio companies” according to its public financial statements.

Since 2022, the company has repeatedly told investors how many companies it funded in the quarter and how many of those were new borrower relationships. Across the last 16 quarters (1Q22 through 4Q25), we estimate Hercules funded 408 companies — but only 147 of them (about 36%) were new relationships. In 14 of those 16 quarters, new relationships were the minority of deals, meaning most capital deployment went to companies already in their portfolio.

The pattern tracks the cycle. In 2023, as venture funding tightened, only 30 out of 105 funded-company events (29%) were new relationships — and in 4Q23, just five out of 27 (19%) were new. In 2024, the skew persisted (3Q24: four new relationships out of 21 fundings). Even in 2025, when originations re-accelerated, three of the four quarters still had a majority of fundings going to existing borrowers (e.g., 3Q25: 24 funded companies, only seven new; 4Q25: 33 funded, only 12 new).

In other words, the capital deployment that investors see in quarterly disclosures is largely repeat lending inside the same borrower set, not a steady refresh.

The result is a credit ecosystem where “no default” may be able to coexist with slow-motion impairment. Maturities get pushed and cash interest is replaced with PIK payments — all while relationship concentration persists because the same borrower names keep receiving incremental capital.

This pattern — where borrowers who would have repaid instead amend, extend, and pile on PIK — is playing out across the software credit universe, not just at Hercules.

“The concentration of software loans maturing in 2026-2027, the largest sector cohort in the near-term maturity bracket, is overwhelmingly going to get resolved through amend-and-extend transactions, not through bankruptcy filings,” wrote JunkBondInvestor. “When a borrower can’t refinance at par but isn’t in operational distress, the playbook is familiar: the sponsor and the existing lender group negotiate an extension. Lenders get a fee, a spread bump, maybe some incremental covenant protection. The company gets two or three more years of runway. No default. No restructuring. Everyone claims victory.”

“But this has implications for loan investors that aren’t being processed,” he explains. “And here’s the part that should keep loan investors up at night: the amend-and-extend process is where creditors get coerced, and in this market, they will be. Sponsors know that a fragmented lender base facing a maturity wall has limited options. You can either extend on the sponsor’s terms, which often means PIK toggles, covenant stripping, and priming risk from new money tranches, or you can watch the loan trade down as the maturity approaches without a resolution. Each extension that adds PIK, strips covenants, or pushes maturities without meaningful deleveraging is a slow-motion impairment that never shows up in the default statistics. The risk isn’t default but extension at terms that are incrementally worse for lenders while the [terminal value] question remains unresolved.”

The Dividend Is Thinner Than It Looks

Hercules frames its payout as conservative. In 4Q25, the company reported net investment income (NII) of $0.48 per share, which it says provided 120% coverage of the base distribution of $0.40. That sounds like a healthy cushion — until you look at what shareholders actually receive.

The total cash distribution for Q4 was $0.47 per share: the $0.40 base plus a $0.07 supplemental. Against $0.48 of NII, that’s roughly 102% coverage — a cushion of a penny. For the full year, the math is equally tight: $1.91 of NII per share against $1.88 in distributions paid. Three cents of breathing room for the entire year. That’s tight.

Hercules disagreed with this perspective in a written response to Hunterbrook Media, calling it “incorrect” to measure dividend coverage against the combined distribution. The company noted that the supplemental distribution is paid from its undistributed earnings spillover of $149.9 million ($0.82 per share as of December 31, 2025 — not from NII). Even excluding all PIK and additional noncash transactions, Hercules said cash NII covered the base distribution at 105.7%. The company added that it “maintains near the high end of base dividend coverage across the entire BDC universe.”

Hercules provided the following figures:

Hercules’ distinction between the base and supplemental distribution is technically correct: The supplemental is paid from a spillover account, not directly from current-quarter NII. To maintain its status as a regulated investment company (RIC), Hercules, like every BDC, must distribute at least 90% of its taxable income to shareholders annually. The IRS allows RICs to maintain an undistributed earnings spillover — a reserve of prior taxable income not yet distributed — from which supplemental payments are drawn.

But this framing, while accurate as a matter of accounting, does not change the economic reality facing shareholders. The total cash leaving the building each quarter is $0.47 per share. The total NII coming in is $0.48. A penny of coverage. The distinction between “base” and “supplemental” is a labeling exercise: Shareholders receiving $1.88 per year in distributions are relying on a combination of current earnings and a finite savings account.

And the savings account, while substantial at $149.9 million ($0.82 per share), is not self-replenishing. If NII declines — whether from falling rates, rising defaults, or slowing deployment — the spillover gets drawn down rather than built up. At the current supplemental rate of $0.28 per year, the reserve covers roughly three years of supplemental payments before it is depleted, assuming no further accumulation.

More importantly, a meaningful slice of the $1.91 in NII is phantom. With PIK accounting for approximately 11% of investment income, the dividend is being supported in part by interest that has never been paid in cash. Even by Hercules’ own figures, cash NII (excluding PIK and other non-cash items) covers the base distribution at 105.7% — a tighter cushion than the headline 119% suggests, and one that leaves virtually no room for credit deterioration before the base distribution itself comes under pressure.

Hercules’ cash balance allows management to keep mailing extra checks even when quarter-to-quarter coverage is razor-thin. But it is not evidence that the portfolio’s current earnings power is expanding. It’s a savings account.

Then there’s dilution. Hercules has been steadily increasing its share count through its at-the-market equity program, selling stock at a premium to NAV and using the proceeds to, among other activities, fund new loans. In 4Q25, the company’s basic weighted-average shares were 180.8 million, up from 165.1 million in 4Q24. Shares outstanding rose to 182.7 million on December 31, 2025, from 170.6 million a year earlier.

That dilution shows up directly in per-share earnings: The full-year 2025 NII per share of $1.91 is lower than 2024’s $2.00 — even though the total NII dollars increased. The company acknowledges that NAV per share benefited from at-the-market issuance above book value, a tailwind that exists only as long as the stock trades at a premium. If the premium disappears, the accretive math flips to destructive. Hercules ended the year with 18.8 million shares still available under the program, having sold stock in 4Q25 and again in January 2026.

The result is a wobbly architecture: a dividend with a penny of coverage, inflated by noncash income, and sustained by drawing down a legacy savings account — all while ongoing share issuance dilutes a story that only works as long as the market pays a premium.

The Competitive Moat That Isn’t

The competitive landscape reinforces the problem. When asked what differentiates Hercules, CEO Bluestein said in a 2025 earnings call, “We have the best team in the business. The team is incredibly experienced. The team has been together for a long time. The team knows how to pick the right companies.”

That may be true. But “we have good people” is a convenient claim for a lender to make when they have no structural moat in a largely commoditized sector. The venture lending market has swollen from a handful of specialists to dozens of publicly-traded BDCs, and SVB — reborn under First Citizens Bank — still commands roughly a third of the venture banking market. Spreads across the sector have fallen significantly, quarter after quarter, for two years running.

If Hercules were truly cornering the market on the best risk-adjusted loans, you’d think competing capital might be swarming those same deals. If it’s winning deals because other lenders don’t want them — extending credit to software companies at terms that more conservative shops walk away from — then the portfolio suffers from adverse selection.

And beyond the competitive dynamics, the former finance team member described a culture with significant “time pressure,” where decisions can end up feeling rushed under tight deadlines. They also noted a “very big culture of wealth and opulence and power and ego and status,” a description that aligns with the Hercules founder’s role in the college admissions scam.

But the story of Hercules is less about culture than it is about the numbers.

Return to the Infinite Money Glitch. When Hercules trades above book value, every dollar of new equity is accretive — cheap capital that funds new loans, generates income, supports the dividend, and sustains the premium. Strategy ran the same machine with bitcoin. TriplePoint ran it with venture loans. In each case, the cycle was elegant on the way up and devastating on the way down. When the premium vanished, equity issuance turned destructive, loan growth stalled, and the dividend followed.

Hercules has long been the exception to the rule — but the exception is narrowing. The premium has been cut in half in a matter of months. The software sector that backs 70 cents of every dollar of NAV is in its worst downturn in years. The income is increasingly phantom. The full dividend cushion is a rounding error. The warrant portfolio that once absorbed losses has shrunk to a fraction of what it was. And the markets still have not meaningfully moved — at least not yet.

Authors

Matthew Termine is a lawyer with nearly five years of experience leading the legal team at a mortgage technology company. In 2017, Matt was credited by the Wall Street Journal, among others, for identifying suspicious mortgage loan transactions that led to several successful criminal prosecutions, including that of a prominent political operative and the chief executive officer of a federally chartered bank. He is a graduate of Trinity College and Fordham University School of Law.

Michelle Cera trained as a sociologist specializing in digital ethnography and pedagogy. She completed her PhD in Sociology at New York University, building on her Bachelor of Arts degree with Highest Honors from the University of California, Berkeley. She has also served as a Workshop Coordinator at NYU’s Anthropology and Sociology Departments, fostering interdisciplinary collaboration and innovative research methodologies.

Sam is a New York Times best-selling author who has written books with former United States Attorney General Eric Holder and former United States Acting Solicitor General Neal Katyal. Sam has published in the New York Times, Washington Post, Boston Globe, Time Magazine, and other outlets — and occasionally volunteers on a fire speech for a good cause. He has a BA in Government from Harvard, where he was named a John Harvard Scholar and wrote op-eds like “Shut Down Harvard Football,” which he tells us were great for his social life.

Editor

Jim Impoco is the award-winning former editor-in-chief of Newsweek who returned the publication to print in 2014. Before that, he was executive editor at Thomson Reuters Digital, Sunday Business Editor at The New York Times, and Assistant Managing Editor at Fortune. Jim, who started his journalism career as a Tokyo-based reporter for The Associated Press and U.S. News & World Report, has a Master’s in Chinese and Japanese History from the University of California at Berkeley.

Graphic

Dan DeLorenzo is a creative director with 25 years reporting news through visuals. Since first joining a newsroom graphics department in 2001, he has built teams at Bloomberg News, Bridgewater Associates, and the United Nations, and published groundbreaking visual journalism at The Wall Street Journal, Associated Press, The New York Times, and Business Insider. A passion for the craft has landed him at the helm of newsroom teams, on the ground in humanitarian emergencies, and at the epicenter of the world’s largest hedge fund. He runs DGFX Studio, a creative agency serving top organizations in media, finance, and civil society with data visualization, cartography, and strategic visual intelligence. He moonlights as a professional sailor working toward a USCG captain’s license and is a certified Pilates instructor.

LEGAL DISCLAIMER

© 2026 by Hunterbrook Media LLC. When using this website, you acknowledge and accept that such usage is solely at your own discretion and risk. Hunterbrook Media LLC, along with any associated entities, shall not be held responsible for any direct or indirect damages resulting from the use of information provided in any Hunterbrook publications. It is crucial for you to conduct your own research and seek advice from qualified financial, legal, and tax professionals before making any investment decisions based on information obtained from Hunterbrook Media LLC. The content provided by Hunterbrook Media LLC does not constitute an offer to sell, nor a solicitation of an offer to purchase any securities. Furthermore, no securities shall be offered or sold in any jurisdiction where such activities would be contrary to the local securities laws.

Hunterbrook Media LLC is not a registered investment advisor in the United States or any other jurisdiction. We strive to ensure the accuracy and reliability of the information provided, drawing on sources believed to be trustworthy. Nevertheless, this information is provided “as is” without any guarantee of accuracy, timeliness, completeness, or usefulness for any particular purpose. Hunterbrook Media LLC does not guarantee the results obtained from the use of this information. All information presented are opinions based on our analyses and are subject to change without notice, and there is no commitment from Hunterbrook Media LLC to revise or update any information or opinions contained in any report or publication contained on this website. The above content, including all information and opinions presented, is intended solely for educational and information purposes only. Hunterbrook Media LLC authorizes the redistribution of these materials, in whole or in part, provided that such redistribution is for non-commercial, informational purposes only. Redistribution must include this notice and must not alter the materials. Any commercial use, alteration, or other forms of misuse of these materials are strictly prohibited without the express written approval of Hunterbrook Media LLC. Unauthorized use, alteration, or misuse of these materials may result in legal action to enforce our rights, including but not limited to seeking injunctive relief, damages, and any other remedies available under the law.

By: Sam Koppelman, Till Daldrup, Andrew Ford, Blake Spendley

Editor: Jim Impoco

Based on Hunterbrook Media’s reporting, Hunterbrook Capital is long $GNRC and short a basket of comparable securities at the time of publication. Positions may change at any time. See full disclosures on our website.

You can get a backup generator for your house in a week. But if you’re building a billion-dollar data center, you might wait over two years — unless you buy from Generac ($GNRC).

And now, states — whose grids are being tested by an AI buildout with a seemingly insatiable appetite for power — are starting to require hyperscalers to install them.

Generator lead times from major players like Caterpillar, Cummins, and Rolls-Royce can stretch as much as 107 weeks. The companies constructing the most expensive AI infrastructure in history are scrambling for alternatives.

Enter: Baudouin — a French engine brand owned by Chinese conglomerate Weichai Group. Baudouin makes the engines that are the key component of multi-megawatt backup generators like those used in major data centers, and in the United States, they are distributed most prominently as part of the new integrated generator systems put together by Generac.

One of Baudouin’s customers?

The Stargate data center in Abilene, the crown jewel of the $500 billion AI buildout from SoftBank, OpenAI, Crusoe, and Oracle. This is according to an obscure section of a Texas air permit reviewed by Hunterbrook. Buried on pages 42 and 43: 28 emergency backup generators powered by Baudouin.

The Stargate permit doesn’t name the company set to package the Baudouin engines into finished generators — but there aren’t many options. Baudouin has a sparse distribution network in the United States, consisting largely of marine engine shops: a small LLC in Washington State, a boat repair company in Florida, you get the idea.

The only two Baudouin distributors to data centers identified by Hunterbrook are IGSA Power — a relatively small Mexico-based manufacturer with an office in Laredo, Texas — and Generac.1

Based on reporting by Hunterbrook Media, it appears that IGSA may be supplying at least some of the 28 Baudouin units at the Abilene site, potentially through an intermediary.2

But the bigger story is what may come next.

The Stargate buildout alone envisions nearly 10 gigawatts of capacity, even with recent reporting from The Information indicating OpenAI won’t lead the build out itself. The backup generation required to support that — thousands of large-megawatt units — is well beyond what a small company like IGSA has demonstrated it can deliver.

Which raises the question: Who scales Baudouin in America?

The answer, almost certainly, is Generac — the $13 billion Wisconsin company known to Wall Street as the stock that dips and rips around hurricanes. The almost 70-year-old company is the “Kleenex” of backup power, a name synonymous with those gray, suitcase-shaped boxes sitting in suburban backyards that hum to life during a storm.

Whereas IGSA manufactures around 2,000 units per year, across various generator sizes and engine brands, Generac sells an estimated 800,000 generators annually.3

And as the AI boom turns electrical power into a scarce commodity, Generac’s suburban and rural identity is undergoing a metamorphosis. Generac has struck a major deal with Baudouin to partner on data center generators — launching a new product line last April; building a $400 million backlog in under a year; and racing to double its manufacturing capacity. The company has redesigned its plant in Oshkosh, Wisconsin, specifically to build its data center generator lineup and moved production of some smaller generators to a new facility in Beaver Dam.

For all these reasons, Generac seems to be on its way to becoming by far the largest U.S. distributor of the Baudouin brand — while other generator manufacturers are already booked out for years.

In a podcast, Ricardo Navarro, Generac’s senior VP & general manager of global telecom and data centers, said the company is moving much faster than competitors. Generac delivers certain offerings within around 30 weeks. “If you compare that to the current supply of existing OEMs, it’s one third of that lead time,” he said. “Everyone is interested in that.”

Trade data supports Navarro’s description of demand. In recent months, Weichai’s shipments to the United States have soared — amid what Andrew A. Chien, a computer science professor at The University of Chicago who studies data center infrastructure, described as a “power rush” in an interview with Hunterbrook.

Of the hyperscalers, Chien said: “They’re probably trying to get everything they can get.”

Investing in backup generators, in particular, may soon become a legal necessity for data center operators across the United States.

Look at Texas, which passed a law that has transformed backup power into a cost of doing business.

Last June, the Lone Star state – now poised to be the largest data center market in the country, enacted Senate Bill 6, requiring new or expanded facilities drawing 75 megawatts or more from the grid to be prepared to curtail their loads or activate on-site backup generation during emergencies.

Facilities that connected after Dec. 31, 2025 must install remote-disconnect equipment allowing the grid operator, ERCOT, to cut them off. (Asked about Stargate, ERCOT declined to comment on any specific ongoing project.)

The law was tested almost immediately.

During Winter Storm Fern in January, the Department of Energy issued an emergency order authorizing ERCOT to direct data centers to fire up their backup generators. The practical effect is that major data centers in Texas now need to curtail load when the grid is pushed past its limits — potentially threatening active training runs — or turn on a fleet of large-megawatt generators like the ones produced by Generac. For companies that can deliver those generators faster than anyone else, the demand is structural.

And Texas is just the beginning.

U.S. data centers consumed roughly 25 gigawatts of power in 2024; by 2030, that figure is expected to reach somewhere between 50 and 100 gigawatts. As the Generac CEO put it in an interview on CNBC: “That’s like adding 20 New York Cities.”

Six states — including New York, Virginia, and Georgia — have introduced moratorium bills to pause new data center construction, and local governments in at least 14 states had enacted their own moratoriums by the end of 2025.

The political message is clear: Data centers must bring their own power and stop burdening the grid. That could be very good news for backup generator manufacturers like Generac.

The Permit

The discovery of OpenAI’s embrace of Baudouin generators comes five months after Hunterbrook Media and Citrini Research flew a drone over the Stargate data center campus in Abilene, Texas. The footage went viral, revealing a facility the size of Lower Manhattan rising from the red Texas clay: eight massive AI computing halls, backed by natural gas turbines from GE Vernova and Caterpillar.

What the flyover didn’t capture — because none of the emergency backup generators appear to have been delivered yet — was Baudouin’s presence.

The January operating permit lists 62 emergency generators at the Longhorn Data Center: 34 Caterpillar units and 28 Baudouin units, providing roughly 170 megawatts of total backup capacity. The 28 Baudouin generators, listed as GEN-35 through GEN-62, are each rated at 2.8 megawatts and classified as Tier IV emergency units. Together they represent approximately 78 megawatts of backup power — enough for 60,000 average homes or nearly half the emergency generation capacity outlined in this permit at one of the most closely watched AI infrastructure projects in the world.

This aligns with the specs of Generac’s generators — which launched back in April. The company released a full lineup of five models, ranging from 2.25 to 3.25 megawatts, all built on the Baudouin M55 engine platform.

One discrepancy: Generac’s published spec sheet for the SD2750/MD2750 lists the Baudouin engine as EPA Tier 2, rated at 2,750 kW — while the permit classifies the Baudouin units as Tier IV at 2,800 kW. But this gap may not be meaningful. Generac’s own product materials describe the lineup as “ready for use with compliant Tier 4 aftertreatment systems,” meaning the same base engine can meet different emissions tiers depending on configuration. (IGSA, for its part, is also labeled as Tier 2, indicating that isn’t a barrier to entry.) And based on Baudouin documents, each of their engines can power up from the name-plated capacity.

The Caterpillar fleet at the same site demonstrates the same pattern — 3516 engines appear in the permit in both Tier II and Tier IV variants.

All in all, it’s a modest amount of backup power, according to Chien, the computer science professor. “I heard a rumor that they were going to go with less than full backup power at that site,” he said. “Normally, if they have a gigawatt of IT equipment, they’ll have a gigawatt of backup. But … that might be because they can accept the unreliability — because it is pretty rare that the grid goes down — or they couldn’t get the generation.”

Still, Stargate could pose environmental concerns, according to Damian Pitt, associate professor of urban and regional studies and planning at Virginia Commonwealth University.

“Whenever they are running, they will be producing air pollution in the form of nitrous oxide, particulate matter, carbon monoxide, sulfur dioxide, and so on that would diffuse into nearby neighborhoods,” Pitt said. “And if there are people living in those areas, then they could be exposed to air pollution impacts.”

And the existing permits may only be for a fraction of what ends up being the total number of generators needed at Stargate. So far, the designated area for generators — according to public plans — remain empty, according to a review of recent satellite and aerial imagery.

Scale

TD Cowen has described Generac as having “exclusive U.S. access” to Baudouin engines for data center applications. But that characterization may overstate the arrangement — IGSA Power also sells Baudouin-powered generators in the U.S. and appears to have placed units at least one Texas data center. Late last month, IGSA shared a pair of images mentioning a major installation at a new data center project in Texas.

Hunterbrook analyzed the images IGSA shared and geolocated them not to Stargate, but to a multi-unit generator installation at Serverfarm’s CTX2 data center in Houston. By comparing tree lines, concrete pads, and distinctive wall structures across more than 20 newly built or recently commissioned data centers in Texas, we found a clear match to footage previously released by Serverfarm in a promotional video last year.

Generac did not respond to repeated requests for comment on the scope of its Baudouin distribution rights.

While boutique vendors may fulfill individual orders, the sheer magnitude of the Stargate buildout requires an industrial muscle that only a company of Generac’s size can provide. IGSA has between 200 and 500 employees, according to LinkedIn. Generac has roughly 9,000 — and they are working on building out the capacity to produce more than a billion dollars worth of data center generators annually starting next year.

On its earnings call last week, Generac described 2026 as an “inflection point” for its data center business. CEO Aaron Jagdfeld said to investors: “We further developed partnerships in the quarter with multiple hyperscalers, including progressing to the pilot phases of our relationships with two specific customers as we prepare for potential significant volumes in 2027 and 2028.”

Jagdfeld told analysts the $400 million backlog was built almost entirely from orders placed by colocators and developers — meaning these estimates exclude any potential revenue after the hyperscaler pilots.

If those convert to long-term supply agreements, the implications are substantial. Jagdfeld has estimated the total addressable market for large-megawatt generators at $15 billion annually and has set a target of capturing 10%-15% of that market — implying at least $1.5 billion in annual data center revenue alone, which would roughly double the company’s current commercial and industrial business.

To support this, Generac is scaling manufacturing capacity aggressively. Jagdfeld acknowledged on the Q4 call that the company took a bit of a “risk” expanding ahead of signed hyperscaler contracts — but said projected volumes for 2027 and 2028 could take them “easily above” $1 billion in capacity.

Jagdfeld projected that order intake would “accelerate over the next several quarters,” adding: “We are making progress with other data center colocators and developers.”

What Generac didn’t say on that earnings call: The most prominent data center buildout in the world has already permitted for the exact generator engines they sell.

Authors

Sam Koppelman is a New York Times best-selling author who has written books with former United States Attorney General Eric Holder and former United States Acting Solicitor General Neal Katyal. Sam has published in the New York Times, Washington Post, Boston Globe, Time Magazine, and other outlets — and occasionally volunteers on a fire speech for a good cause. He has a BA in Government from Harvard, where he was named a John Harvard Scholar and wrote op-eds like “Shut Down Harvard Football,” which he tells us were great for his social life.

Till Daldrup joined Hunterbrook from The Wall Street Journal, where he focused on open-source investigations and content verification. In 2023, he was part of a team of reporters who won a Gerald Loeb Award for an investigation that revealed how Russia is stealing grain from occupied parts of Ukraine. He has an M.A. in Journalism from New York University and a B.S. in Social Sciences from University of Cologne. He’s also an alum of the Cologne School of Journalism (Kölner Journalistenschule).

Andrew Ford is an investigative journalist who exposed systemic flaws and prompted reforms in healthcare, business, policing, and state government. His reporting was published by ProPublica, USA Today, The Arizona Republic, Asbury Park Press, and Florida Today. He holds a journalism bachelor’s from the University of Florida and is based in Phoenix, AZ.

Blake Spendley joined Hunterbrook from the Center for Naval Analyses (CNA), where he led investigations as a Research Specialist for the Marine Corps and US Navy. He built and owns the leading open-source intelligence (OSINT) account on X/Twitter, called @OSINTTechnical (over 1 million followers), which also distributes Hunterbrook Media reporting. His OSINT research has been published in Bloomberg, the Wall Street Journal, and The Economist, among other top business outlets. He has a B.A. in Political Science from USC.

Editors

Jim Impoco is the award-winning former editor-in-chief of Newsweek who returned the publication to print in 2014. Before that, he was executive editor at Thomson Reuters Digital, Sunday Business Editor at The New York Times, and Assistant Managing Editor at Fortune. Jim, who started his journalism career as a Tokyo-based reporter for The Associated Press and U.S. News & World Report, has a master’s in Chinese and Japanese History from the University of California at Berkeley.

Hunterbrook Media publishes investigative and global reporting — with no ads or paywalls. When articles do not include Material Non-Public Information (MNPI), or “insider info,” they may be provided to our affiliate Hunterbrook Capital, an investment firm which may take financial positions based on our reporting. Subscribe here. Learn more here.

Please email ideas@hntrbrk.com to share ideas, talent@hntrbrk.com for work, or press@hntrbrk.com for media inquiries.

LEGAL DISCLAIMER

© 2026 by Hunterbrook Media LLC. When using this website, you acknowledge and accept that such usage is solely at your own discretion and risk. Hunterbrook Media LLC, along with any associated entities, shall not be held responsible for any direct or indirect damages resulting from the use of information provided in any Hunterbrook publications. It is crucial for you to conduct your own research and seek advice from qualified financial, legal, and tax professionals before making any investment decisions based on information obtained from Hunterbrook Media LLC. The content provided by Hunterbrook Media LLC does not constitute an offer to sell, nor a solicitation of an offer to purchase any securities. Furthermore, no securities shall be offered or sold in any jurisdiction where such activities would be contrary to the local securities laws.

Hunterbrook Media LLC is not a registered investment advisor in the United States or any other jurisdiction. We strive to ensure the accuracy and reliability of the information provided, drawing on sources believed to be trustworthy. Nevertheless, this information is provided “as is” without any guarantee of accuracy, timeliness, completeness, or usefulness for any particular purpose. Hunterbrook Media LLC does not guarantee the results obtained from the use of this information. All information presented are opinions based on our analyses and are subject to change without notice, and there is no commitment from Hunterbrook Media LLC to revise or update any information or opinions contained in any report or publication contained on this website. The above content, including all information and opinions presented, is intended solely for educational and information purposes only. Hunterbrook Media LLC authorizes the redistribution of these materials, in whole or in part, provided that such redistribution is for non-commercial, informational purposes only. Redistribution must include this notice and must not alter the materials. Any commercial use, alteration, or other forms of misuse of these materials are strictly prohibited without the express written approval of Hunterbrook Media LLC. Unauthorized use, alteration, or misuse of these materials may result in legal action to enforce our rights, including but not limited to seeking injunctive relief, damages, and any other remedies available under the law.

By: Till Daldrup, Michelle Cera, Evan Comen

Editor: Sam Koppelman, Jim Impoco

Hunterbrook Media’s investment affiliate, Hunterbrook Capital, does not have any positions related to this article at the time of publication. Positions may change at any time. Full disclosures on our website. If you’ve been impacted, and have a story to tell, each out at ideas@hntrbrk.com.

Note: Shout out to our new newsletter subscribers — who found us via The Compound and Friends. We plan to publish our next financial markets focused story later this week.

TLDR:

• Indorama Ventures pledged to be a responsible neighbor. Documents obtained by Hunterbrook show the opposite: The Thai petrochemical company has exposed thousands in Texas to carcinogens, violated environmental permits, and concealed toxic leaks from regulators.

• According to Texas Commission on Environmental Quality (TCEQ) records, Indorama has emitted several thousand pounds of ethylene oxide and other hazardous air pollutants in violation of its permits. The documents show a pattern of accidents and leaks at the plant that have exposed thousands of local residents to toxic chemicals.

• The TCEQ also found that the company didn’t notify regulators and the public of several toxic releases at its facility. At least two large emissions of harmful substances were discovered more than a year after the fact.

• Jefferson County, where the Indorama plant is located, has the highest breast cancer mortality among women under the age of 50 in Texas, according to National Cancer Institute data. Long-term ethylene oxide exposure is linked to a higher rate of breast cancer in women.

See the full, interactive graphic — and information on the methodology behind it — on our website.

The air in Port Neches, Texas, is thick, even in December. A pungent smell permeates the city — schools, parks, restaurants, hospitals, and grocery stores. The odor is reminiscent of burning plastic mixed with the smell of a gas station pump. Some locals say they stopped noticing it a long time ago.

This region in southeast Texas is known as the Golden Triangle because of the wealth generated by the massive oil boom in the early 1900s. But today, some residents say the economic opportunities the petrochemical industry has offered them and their neighbors have come at a high personal cost.

“You live here, you’re gonna get cancer,” a local mother said. She prefers not to be named because of her relationships with people who work in the oil and chemical industries.

Timothy, her son, is an ICU nurse. Both of them grew up in the Golden Triangle. They have another name for the region: cancer alley, reminiscent of the nickname for an 85-mile stretch of petrochemical plants and refineries along the Mississippi River in Louisiana.

He said he has personally treated countless locals with various cancers, including numerous cases of patients with lung cancer who never smoked.

For Timothy and his mother, the flames, the smoke, the never-ending smell of chemicals coming from refineries and petrochemical plants are a fact of life. So are the sirens that go off during the day. And the eerie orange glow in the sky at night, from the lights on machinery at the plants.

Residents feel “helpless to do anything about it,” Timothy’s mother said. And the plants and refineries are the economic lifeblood of the region. “You weigh the risks of living in the area with the good living wages that you get.“

Inside Indorama’s Chemical Footprint in Texas

About 600 people in this area work for Indorama Ventures ($IVL), a $3.8-billion Thai petrochemical company. In 2020, it bought a complex in Port Neches that borders Groves and Nederland. Indorama referred to the plant as a “large flagship site on the US Gulf Coast.”

Documents obtained by Hunterbrook Media through public records requests show that the company has exposed residents of all three cities to high levels of a cancer-causing chemical.

Indorama’s plant produces ethylene oxide (EtO), a colorless, flammable gas used to make antifreeze, detergents, and textiles, and as a sterilizing agent for medical devices. It is also a ruthlessly efficient carcinogen. In 2016, the EPA in a revised toxicity assessment concluded it is 30 times more carcinogenic than previously thought. Ethylene oxide is linked to blood cancers like leukemia, non-Hodgkin lymphoma, and multiple myeloma, as well as breast cancer. In 2024, the EPA issued stricter rules that require commercial medical sterilizers and large chemical plants to reduce their ethylene oxide emissions.

A former Indorama employee who specialized in ethylene oxide at the Port Neches plant emphasized the extreme danger the gas poses. Malfunctions could “wipe out Port Neches,” they told Hunterbrook.

Given the highly hazardous nature of the chemicals it handles, Indorama has publicly stressed its responsibility to the surrounding communities.

“It’s a question of what you have to do right,” Chad Anderson, at that time Indorama’s vice president of manufacturing, told local media in 2021, “and that falls down to being responsible and abiding by our environmental permits, keeping employees and clients happy and maintaining our civil engagement.” That was shortly after the company took over the plant.

Yet, Texas Commission on Environmental Quality records show that Indorama has repeatedly failed to contain the carcinogen. The plant has emitted several thousand pounds of ethylene oxide and other toxic chemicals without TCEQ authorization. The records show a pattern of accidents and leaks at the plant that have exposed thousands of local residents to hazardous air pollutants.

During six record reviews in 2024, the TCEQ found Indorama had exceeded its permitted emission limits several times in 2022 and 2023 and hadn’t notified regulators or the public of large chemical releases at its facility.

In one previously unreported incident, Indorama in September 2023 emitted roughly 1,880 pounds of ethylene oxide during a mere three-minute leak caused by defective tubing.

This brief emission alone would have made Indorama one of the top 20 emitters of ethylene oxide in the United States in 2023, according to the EPA’s Toxics Release Inventory database.

For context, in 2023, Sterilization Services of Tennessee, a medical sterilization company, announced it would shut down its plant following months of community advocacy against its EtO emissions. It had released 2,000 pounds of ethylene oxide that year. Indorama released nearly that same amount in 180 seconds.

What’s more, the records show that Indorama did not report three releases to regulators until a TCEQ investigator discovered them during a record review. Two of these leaks were uncovered over a year after the fact. According to Texas regulations, facilities like Indorama’s are required to notify authorities of large emissions events within 24 hours.

Jefferson County — which includes Port Neches, Groves, and Nederland — has the highest breast cancer mortality among women under the age of 50 in Texas, according to the latest available National Cancer Institute data. Long-term ethylene oxide exposure is linked to a higher rate of breast cancer. The county also has the second-highest incidence of lung and bronchus cancers among people under 50 in the state, as well as one of the highest death rates from stomach cancer.

A 2021 ProPublica analysis calculated the cancer risk caused by industrial pollution across the United States based on emissions data and identified the area around Indorama’s facility as a major national hotspot. The estimated cancer risk around the facility is 1 in 53 — meaning that for every 53 residents, 1 is likely to develop cancer solely due to toxic exposure.

Indorama’s plant is not the only polluter in the area. Several nearby refineries and petrochemical facilities are also adding to the toxic cocktail that residents are exposed to daily.

But an ethylene oxide emitter of that size in such close proximity to several communities is rare, according to a Hunterbrook review of large EtO emitters. The only plant that released more ethylene oxide than Indorama’s Port Neches facility in 2023 was Dow Chemical’s ethane cracker in Plaquemine, Louisiana. But while Dow’s facility is located next to a relatively sparsely populated strip of land, Indorama’s plant is located less than 1,000 feet from Groves Middle School. Port Neches Primary School is less than one and half miles from the facility.

And it’s not just ethylene oxide. Indorama released more than 7,300 pounds of volatile organic compounds (VOCs) and 1,500 pounds of nitrogen oxides (NOx) between July 7 and July 10, 2023 — in violation of its permits, according to the TCEQ. An incident report shows this was due to an equipment malfunction.

Volatile organic compounds can cause various health issues, from eye and throat irritation and headaches to liver, kidney, and central nervous system damage. Some can cause cancer. Nitrogen oxides like nitrogen dioxide can aggravate respiratory diseases, particularly asthma, leading to symptoms like coughing, wheezing, or difficulty breathing.

Indorama did not respond to a request for comment.

“It’s very hard on my asthma,” Timothy said of the plant’s emissions. “I’ve been on the bathroom floor vomiting from coughing so hard.”

He and his mother had never heard of these accidental releases and were unaware of the thousands of pounds of carcinogenic ethylene oxide that Indorama emitted in violation of its permits in recent years. “We don’t know that,” Timothy said. “They hide that from us.”

Cindy and Don Colwell said the same — and they live directly across the street from Indorama. “We’ve never heard anything,” Cindy said. They mentioned they had never seen or heard anything about the releases during their time living in Groves — nothing in the news, no phone alerts, and no written notifications.

“It would be nice to know what we are breathing,” Cindy said.

Their wood-paneled living room is filled with photos of their family. Their rescue animals, two dogs and three cats, lay by their feet. The Colwells have gotten used to the flares, they said, and the noises from the plant that sound like airplanes.

But they are also aware of the risk. Several of their loved ones have died from cancer. Cindy listed all of the cancers that had plagued her family and friends: breast cancer, lung cancer, stomach cancer, liver cancer, mesothelioma, and more.

While the Colwells are deeply concerned about their young grandchildren, for them, the tragedies that come with the oil and chemical industry in southeast Texas are part of the trade-off. “Well, I’m gonna die of something,” Cindy joked. “You can live as healthy as you want, it’s not gonna stop carcinogenic things from tearing into your body.”

But the health data is impossible to ignore. Jefferson County has the highest breast cancer mortality rate among women under 50 in the state. Its mortality rates for heart and long-term lung disease are respectively 38% and 46% above the state average, according to data from the National Center for Health Statistics (NCHS).

A Hunterbrook analysis using National Oceanic and Atmospheric Administration plume models shows that the single three-minute leak in September 2023 dispersed ethylene oxide as far as Louisiana.

The EPA models cancer risks for some air pollutants based on continuous exposure over a 70-year lifespan, so it’s difficult to estimate the increased cancer risk from acute exposure. For ethylene oxide, the agency considers a continuous lifetime exposure above 0.02 micrograms per cubic meter an “unacceptable” cancer risk. After the release at Indorama’s facility, ethylene oxide concentrations peaked at levels over 6,000 times higher than that on one residential block close to the plant.

New EPA rules for chemical plants would have required Indorama to set up fenceline monitoring by July of this year, with an action level for ethylene oxide set at an annual average of 0.2 micrograms per cubic meter. Exceeding this threshold would have triggered a root cause analysis and corrective actions. But Indorama received a presidential exemption for its Port Neches plant in July 2025, which granted the company a two-year extension of compliance deadlines set in the EPA rule.
Ethylene oxide concentrations peaked at more than 600 times the action level after the three-minute release at Indorama’s facility. Almost 5,000 people were exposed to concentrations five times the threshold for at least an hour.

And documents reviewed by ProPublica suggest that ethylene oxide levels around the facility have at least periodically been above the EPA threshold in recent years: When the agency asked Indorama to temporarily set up fenceline monitors in 2023, the ethylene oxide concentration was seven times higher than the action level, according to ProPublica’s investigation.

“Too Big to Fail”

For some residents, the cycle of leaks and lenient fines has bred a deep cynicism.

“Because of their contribution to the economy, they all enjoy the ‘too big to fail’ pass,” said Bill Johnson.

He and his wife Alicia live in Port Neches near the Indorama plant with their teenage daughter. They said the “deafening” sounds from the facility and the pollution are a part of their daily lives. The Johnsons were also unaware of the dangerous emissions from Indorama in recent years.

“If I am responsible for an accident that causes damage to 60,000 homes in the area,” Bill said, giving a figurative estimate, “I’m going to prison for eternity, right? I’m gonna be fiscally liable. But because of their contribution to the economy, they get a pass.”

He worked in the industry for many years. Risk has been a constant in his working life, something he learned to live alongside rather than fear.

“It’s become such a fact of life around here,” Bill said. “I don’t wanna be living across from a refinery for the rest of my life … but there isn’t really an option.”

Moving is more of a fantasy than a plan for him and his family. “It does kinda feel like the big bad wolf is the plant, and in the end it’s them saying, ‘If you don’t like it, you can move’ … but I don’t have millions of dollars. Y’all do,” Bill said.

Alicia agreed. “Who can afford to just pack up and move?”

Like nearly every other local who spoke to Hunterbrook for this story, their lives are intertwined with illness. A friend of theirs who grew up in Port Neches developed cancer and was told by doctors that it was environmental. Bill and Alicia understood the connection but also the futility of confronting it.

To the Johnsons, Indorama is not just a neighbor; it’s a company that defines everything around them. Bill described the plant as “swallowing up the town.”

“This is Port Neches, not Port Indorama,” Bill said.

The penalties the TCEQ proposed for the various violations it identified at Indorama’s plant in Texas add up to almost $75,000, according to the documents Hunterbrook obtained. For a company that recorded $15.4 billion in revenue in 2024, that’s barely a rounding error.

Two of the proposed orders with penalties totaling around $43,000 were adopted by the Commission on January 28. Another, carrying a penalty of about $10,000, was signed by the TCEQ’s executive director earlier that month. One other order is still pending approval.

The Commission is also working on enforcement cases for five other air emissions incidents at Indorama’s facility involving ethylene oxide releases between 2021 and 2024, according to a TCEQ spokesperson.

Broken Promises and Suspicious Emissions Numbers

Forty miles east of Port Neches, in Westlake, Louisiana, Indorama operates another facility with a similarly troubling record. The area surrounding the plant here is much like its Texas counterpart — a persistent toxic stench, smoky skies, and schools and homes close by.

After it bought the facility in 2015, Indorama made the same pledge to be a good neighbor in Louisiana that it had made in Texas. Then-Governor Bobby Jindal welcomed the company with a 1.5 million grant as well as a tax subsidy worth at least 73 million over 10 years, according to calculations by the nonprofit Together Louisiana. Indorama pledged to “meet or exceed all environmental regulations.” The company did not keep its promise.

The problems began as soon as Indorama started up the plant, which had been closed down in 2001.

Repeated mechanical failures led to emergency flaring at the facility in 2019. The flare belched black smoke into the community, roaring with such violence that it rattled the windows of nearby homes.

“It was like a 300-foot flame, and it went on for weeks,” said James Hiatt, director of a Lake Charles nonprofit called For a Better Bayou, at the time. “People couldn’t sleep at night.”

According to a Louisiana Department of Environmental Quality document, the vent flare of the facility burned for a total of 2,256 hours — or 94 days — in 2019.

The issues forced Indorama to shut the plant down in June 2019 before starting back up again. As it turned out, the company was also missing certain key permits for its facility.

Indorama quietly settled various environmental and compliance violations related to the bungled ethane cracker startup with the Louisiana Department of Environmental Quality this year and paid $175,000, according to a document published on the department’s website. The company was finally able to restart the ethane cracker in January 2020.

Roishetta Sibley Ozane lives less than a mile from Indorama’s Westlake facility. She is a mother of six and is concerned about the links between petrochemical plants and maternal health, including premature births and infant death. Ozane had a stillborn baby and several of her children have asthma.

“Indorama is a serial polluter,” she said. “I’m starting to realize that it’s my close proximity to these facilities and constantly breathing in this polluted air,” Ozane said about her and her children’s health issues.

She decided to purchase a special camera to show pollution that can’t be seen with the naked eye. “And it showed that Indorama was polluting way more than we even expected.”

Ozane said that she doesn’t understand why companies like Indorama only have to pay small fines when they release toxic chemicals that affect residents and the environment. “These facilities get to come here and pollute our air and our water, kill our children, cause us to have cancer and all these other different respiratory issues, and they’re not paying their fair share,” she said.

And emissions from Indorama’s plant are getting worse, a Hunterbrook data analysis shows. The company’s releases of known or probable carcinogens at the facility were four times higher in 2024 than in 2021, its first full year of operation. Emissions of 1,3-butadiene — a chemical linked to cancers of the stomach, blood, and lymphatic system — have increased tenfold.

The analysis also shows that the company kept violating its permits once the facility finally opened: Fugitive benzene releases were above emissions limits not only in 2020, but also in 2022 and 2023.

Fugitive emissions are the industry term for leaks that escape through valves, cracked seals, and loose connectors rather than being filtered through controlled smokestacks. In Louisiana, state data shows that large amounts of Indorama’s toxic releases come from these hard-to-track sources.

But a forensic review of the company’s filings suggest these numbers may be little more than fiction. For two consecutive years — 2022 and 2023 — the company submitted identical fugitive emission figures. Then, in 2024, the data became even more improbable: Indorama reported emitting exactly the same amount its permits allowed across a wide range of chemicals, down to the single pound.

A comparison with neighboring facilities underscores the anomaly. While similar plants in the area reported emissions with decimal-point precision, Indorama’s figures remained suspiciously static, despite the company enjoying permit limits significantly higher than those of competitors.

If Indorama indeed calculated its emissions inaccurately, surrounding communities would be left in the dark about the amount of toxic chemicals they are breathing in — and the company could potentially dodge fines for permit violations.

The Louisiana Department of Environmental Quality did not respond to a request for comment.

Residents in Louisiana are fighting back. Ozane decided to start organizing her community. She brought neighbors to city council meetings and police jury meetings, and even took a trip to Washington, D.C., to meet with the Federal Energy Regulatory Commission.

“We are literally fighting for our lives, trying to ensure that our children can breathe clean air and drink clean water,” she said.

“Dont Mess With Our Economy”

In Port Neches, however, speaking out against the petrochemical industry is sometimes viewed as civic betrayal. While a number of residents were happy to meet and share their stories with Hunterbrook for this article, others pushed back and defended Indorama vigorously.

“Ms New York, dont mess with our economy,” one resident wrote under a Hunterbrook reporter’s post on Facebook. “Somebody will run you out of town,” another wrote. “Go back to NY,” a third person commented under another Facebook query.

When a reporter knocked on the doors of homes near Indorama’s plant to talk about living in such close proximity to the facility, one resident threatened to call the police.

For many locals, raising concerns about Indorama isn’t simply about pollution or health risks; it’s about threatening the foundation of a city whose identity and livelihood are tied to the plant.

“It is the main economy,” Timothy’s mother said.

As an ICU nurse, Timothy’s livelihood doesn’t depend on Indorama. He dreams often of leaving Texas — of finding a place where the environment isn’t a daily hazard.

But for now, he stays for her.

He turned to his mother, the smell of the plant hanging thick outside.

“Do you know how nice it would be to wake up to clean air in the morning?”

Authors

Till Daldrup joined Hunterbrook from The Wall Street Journal, where he focused on open-source investigations and content verification. In 2023, he was part of a team of reporters who won a Gerald Loeb Award for an investigation that revealed how Russia is stealing grain from occupied parts of Ukraine. He has an M.A. in Journalism from New York University and a B.S. in Social Sciences from University of Cologne. He’s also an alum of the Cologne School of Journalism (Kölner Journalistenschule). Till is based in New York.

Michelle Cera trained as a sociologist specializing in digital ethnography and pedagogy. She completed her Ph.D. in Sociology at New York University, building on her Bachelor of Arts degree with Highest Honors from the University of California, Berkeley. She has also served as a Workshop Coordinator at NYU’s Anthropology and Sociology Departments, fostering interdisciplinary collaboration and innovative research methodologies.

Evan Comen is currently the senior data editor at U.S. News & World Report, where he focuses on government rankings and accountability reporting. He has worked as a data journalist since 2015, covering climate change, urban economics, and public policy. Evan has a B.A. in economics from the University of North Carolina at Chapel Hill and is based in New York.

Editors

Sam Koppelman is a New York Times best-selling author who has written books with former United States Attorney General Eric Holder and former United States Acting Solicitor General Neal Katyal. Sam has published in the New York Times, Washington Post, Boston Globe, Time Magazine, and other outlets — and occasionally volunteers on a fire speech for a good cause. He has a B.A. in Government from Harvard, where he was named a John Harvard Scholar and wrote op-eds like “Shut Down Harvard Football,” which he tells us were great for his social life. Sam is based in New York.

Jim Impoco is the award-winning former editor-in-chief of Newsweek who returned the publication to print in 2014. Before that, he was executive editor at Thomson Reuters Digital, Sunday Business Editor at The New York Times, and Assistant Managing Editor at Fortune. Jim, who started his journalism career as a Tokyo-based reporter for The Associated Press and U.S. News & World Report, has a master’s in Chinese and Japanese History from the University of California at Berkeley.

Graphic

Dan DeLorenzo is a creative director with 25 years reporting news through visuals. Since first joining a newsroom graphics department in 2001, he has built teams at Bloomberg News, Bridgewater Associates, and the United Nations, and published groundbreaking visual journalism at The Wall Street Journal, Associated Press, The New York Times, and Business Insider. A passion for the craft has landed him at the helm of newsroom teams, on the ground in humanitarian emergencies, and at the epicenter of the world’s largest hedge fund. He runs DGFX Studio, a creative agency serving top organizations in media, finance, and civil society with data visualization, cartography, and strategic visual intelligence. He moonlights as a professional sailor working toward a USCG captain’s license and is a certified Pilates instructor.

Hunterbrook Media publishes investigative and global reporting — with no ads or paywalls. When articles do not include Material Non-Public Information (MNPI), or “insider info,” they may be provided to our affiliate Hunterbrook Capital, an investment firm which may take financial positions based on our reporting. Subscribe here. Learn more here.

Please email ideas@hntrbrk.com to share ideas, talent@hntrbrk.com for work, or press@hntrbrk.com for media inquiries.

LEGAL DISCLAIMER

© 2026 by Hunterbrook Media LLC. When using this website, you acknowledge and accept that such usage is solely at your own discretion and risk. Hunterbrook Media LLC, along with any associated entities, shall not be held responsible for any direct or indirect damages resulting from the use of information provided in any Hunterbrook publications. It is crucial for you to conduct your own research and seek advice from qualified financial, legal, and tax professionals before making any investment decisions based on information obtained from Hunterbrook Media LLC. The content provided by Hunterbrook Media LLC does not constitute an offer to sell, nor a solicitation of an offer to purchase any securities. Furthermore, no securities shall be offered or sold in any jurisdiction where such activities would be contrary to the local securities laws.

Hunterbrook Media LLC is not a registered investment advisor in the United States or any other jurisdiction. We strive to ensure the accuracy and reliability of the information provided, drawing on sources believed to be trustworthy. Nevertheless, this information is provided “as is” without any guarantee of accuracy, timeliness, completeness, or usefulness for any particular purpose. Hunterbrook Media LLC does not guarantee the results obtained from the use of this information. All information presented are opinions based on our analyses and are subject to change without notice, and there is no commitment from Hunterbrook Media LLC to revise or update any information or opinions contained in any report or publication contained on this website. The above content, including all information and opinions presented, is intended solely for educational and information purposes only. Hunterbrook Media LLC authorizes the redistribution of these materials, in whole or in part, provided that such redistribution is for non-commercial, informational purposes only. Redistribution must include this notice and must not alter the materials. Any commercial use, alteration, or other forms of misuse of these materials are strictly prohibited without the express written approval of Hunterbrook Media LLC. Unauthorized use, alteration, or misuse of these materials may result in legal action to enforce our rights, including but not limited to seeking injunctive relief, damages, and any other remedies available under the law.

DISCLOSURES

© 2026 by Hunterbrook Media LLC. All rights are reserved. Use of this website is strictly governed by our Terms of Use. The content herein is furnished to users on an “AS IS” basis without any form of express or implied warranty. For detailed information regarding the handling of user privacy, please refer to our Privacy Policy.

This website may feature forward-looking statements reflecting Hunterbrook Media’s current perspectives concerning, inter alia, its operational and investment strategies. Such statements, identifiable by terms including but not limited to “anticipate,” “believe,” “estimate,” “expect,” “intend,” “may,” “plan,” “predict,” “should,” and “will,” are inherently speculative and subject to various risks and uncertainties. Consequently, actual results or outcomes may substantially deviate from those projected in these statements. Hunterbrook Media expressly disclaims any obligation to update or revise any forward-looking statements, irrespective of subsequent developments or new information.

We use this platform to share our opinions, research, and any related investment positions held by Hunterbrook Media’s affiliates. Positions may change at any time. Descriptions of any investment methodologies and strategies of Hunterbrook Media or its affiliates are indicative and subject to modification at Hunterbrook Media’s discretion. Use of Hunterbrook Media’s research is entirely at the user’s risk. Users shall not use this site for any illegal or prohibited purposes and comply with all relevant local, state, national, and international laws and regulations in their use of this site.

The information presented is current as of the date indicated. No representation regarding the completeness or timeliness of this information is made. This information should not be construed as investment advice or a recommendation of any kind. Some information is sourced from third parties and, while deemed reliable, has not been independently verified, and its accuracy or completeness is not guaranteed. Hunterbrook Media is not affiliated with any sites linked to this website.

This site does not constitute an offer or solicitation to purchase or sell services. Investment in securities and other financial instruments inherently involves significant risk, including the potential loss of principal. Users are advised to undertake their own due diligence and consult professional advisors before engaging in any investment activity. All users and each potential investor must independently evaluate the suitability of any investment opportunity in light of their specific objectives and circumstances, considering both the potential risks and benefits.

By accessing this site, users acknowledge that Hunterbrook Media, potentially in conjunction with Hunterbrook Media’s affiliates, may hold positions in securities mentioned herein, potentially yielding significant gains from stock price fluctuations post-publication. Hunterbrook Media reserves the right to transact in these securities post-publication, adopting positions which may be long, short, or neutral, irrespective of initial reporting on positions.

Should any provision of these Terms of Use be adjudged invalid or unenforceable, the remaining provisions shall remain in full force and effect. Third-party site link trademarks belonging to their respective owners. Hunterbrook Media retains copyright on all the text, contents, graphics and trademarks displayed on this site. All the text, graphics and trademarks displayed on this site are owned by Hunterbrook Media.Search

By: Sam Koppelman, Blake Spendley

Editor: Wendy Nardi

Based on Hunterbrook Media’s reporting, Hunterbrook Capital is short $BBAI at the time of publication. Positions may change at any time. See full disclosures on our website.

Five military branches just adopted an internal Pentagon AI product partnered with Google and OpenAI. Who lost out? BigBear, which had just spent the equivalent of over half its cash acquiring a military AI tool offering a very similar product. BigBear’s newly named CTO took to LinkedIn to call the Pentagon’s move “stupid, wasteful and moronic.”

“BREAKING NEWS! The Pentagon just announced they’re adding ChatGPT and Grok to GenAI.mil.

Sounds great right? WRONG.”

So says Nicolas M. Chaillan, who was named the Chief Technology Officer of BigBear.ai ($BBAI) after announcing he would sell the defense contractor his start-up, Ask Sage, for $250 million in November. Twenty nine days after BigBear’s acquisition was announced, the Pentagon launched a free alternative that seemingly does much of the same thing.

In a comment, Chaillan called the launch “beyond stupid, wasteful and moronic,” admonishing the Pentagon’s CTO for being “as technical as my dog.” The federal government is responsible for the majority of BigBear’s revenue.

The timing was, admittedly, tough for BigBear.

BigBear publicized the acquisition of Ask Sage on November 10 — agreeing to pay a quarter billion in cash for a generative AI platform tailored to the military. It was, apparently, BigBear’s largest-ever acquisition.

Just weeks later, on December 9, the Pentagon launched its competing platform, GenAI.mil — with Defense Secretary Pete Hegseth issuing a top-down mandate instructing all 3 million Department of Defense personnel to use it.

As of February 2, five out of six military branches have formally designated GenAI.mil as their enterprise AI platform, sidelining legacy tools that include Ask Sage, according to the military news outlet DefenseScoop.com. The sixth is the Coast Guard, which is piloting a separate product. The new GenAI.mil platform has allegedly already surpassed 1.1 million unique users.

Then, this week, the Pentagon announced it will add OpenAI’s ChatGPT to GenAI.mil, joining Google’s Gemini. xAI’s Grok is also reportedly in the mix. (Anthropic is, reportedly, hesitant to agree to the government’s “all lawful uses” clause.)

For BigBear.ai, this came at a difficult moment.

The company announced the Ask Sage deal alongside a third-quarter earnings report that showed revenue down 20%, margins compressing, and full-year guidance cut by as much as $40 million from its original outlook. But investors seemingly supported the Hail Mary acquisition, so the stock temporarily popped.

BigBear had essentially bet its future on Ask Sage, spending the equivalent of half the cash on its balance sheet to purchase Ask Sage at 10 times the startup’s 2025 annual recurring revenue.

“We look to our acquisition of Ask Sage to support accelerating our growth into 2026 and beyond,” said BigBear’s CEO, who highlighted Ask Sage’s over 100,000 government users.

The problem: At the time, it was not publicly known that the Pentagon itself would be releasing a competitor to Ask Sage, in partnership with some of the biggest companies in the world.

A month later, GenAI.mil reportedly already has about 11 times as many people on its platform. And Chaillan — Ask Sage founder, newly named BigBear CTO — is, apparently, Big Mad.

Chaillan has repeatedly posted about GenAI.mil on LinkedIn — lamenting that the platform did not integrate various models as effectively as Ask Sage.

“Over a million DoD personnel deserve better than a tabbed interface where they have to start from scratch every time they switch models,” he wrote.

“But hey, at least they’re moving. Just… in the wrong direction.”

One commenter replied: “LinkedIn needs to add a sad face emoji,” prompting another to lament: “and a scared astonished one.”

Chaillan’s response? “Indeed.”

A different commenter implied Chaillan — and, consequently, BigBear — was being punished for not bribing the administration.

“@Nicholas M. Chaillan you forgot to contribute to the new West Wing project is my guess.”

Chaillan didn’t reply to this comment — but he did like it.

Around this same time, Reddit observers pointed out that Chaillan appears to have removed BigBear from his LinkedIn job title. Asked for an explanation, BigBear declined to provide Hunterbrook with a comment. Nor did Chaillan himself, who ghosted a Hunterbrook reporter on LinkedIn, despite clearly being, ah, quite active (!) on the platform.

What BigBear.ai Bought — and Why

When BigBear.ai announced its agreement to acquire Ask Sage, the company framed the deal as transformational. CEO Kevin McAleenan called it an accelerant that would position BigBear.ai for “leadership in our core national security market.”

Chaillan — the former first chief software officer of the U.S. Air Force — founded Ask Sage in 2023. The startup built its brand around serving the defense AI market. The platform’s key selling points: It was branded as model-agnostic, running multiple large language models under the hood to cross-check answers. And it claimed to hold an authorization called FedRAMP High, plus an ability to handle DoD Impact Level 5, IL6, and Top Secret information.

The crown jewel was Ask Sage’s role in powering the Army Enterprise LLM Workspace, launched under a $49 million, five-year IDIQ contract. The Pentagon’s Chief Digital and Artificial Intelligence Office awarded Ask Sage an additional $10 million in June 2025 to expand the platform to combatant commands, the Joint Staff, and the Office of the Secretary of Defense.

Between the two contracts, it wasn’t much revenue — $60 million total over five years — but BigBear paid $250 million for Ask Sage anyway. And on paper, it wasn’t necessarily crazy: It looked like a floundering company making an expensive purchase to buy its way into a burgeoning defense tech growth vector.

Then, BigBear’s primary customer, the Pentagon, partnered directly with the LLM providers powering Ask Sage, including Google, to create what is essentially a competitor.

The Pentagon’s Internal Competitor

A month after BigBear.ai announced the Ask Sage deal, the Pentagon launched GenAI.mil, a centralized platform to deliver commercial AI models directly to the entire Department of War workforce.

“For the first time ever, by the end of this week, 3 million employees, warfighters, and contractors are going to have AI on their desktop, every single one,” Pentagon Chief Technology Officer Emil Michael — also known for threatening a journalist and her family as an exec at Uber — announced at the launch.

Defense Secretary Hegseth followed with a directive ordering personnel to “log in, learn it, and incorporate it into your workflows immediately.”

Google’s Gemini for Government was the first model to go live for handling sensitive but unclassified data. The Pentagon holds $200 million contracts with each of four frontier AI companies — Google, OpenAI, Anthropic, and xAI. On Monday, the Department formally announced its partnership with OpenAI to bring ChatGPT to GenAI.mil.

The adoption velocity has been impressive. Within two months, GenAI.mil hit 1.1 million unique users, according to DefenseScoop, and achieved what the Pentagon claims is 100% uptime since launch.

Five of Six Branches Using GenAI.mil

On February 2, DefenseScoop reported that five of six military branches had formally elevated GenAI.mil as their enterprise AI platform. The service-by-service breakdown of their reporting paints a stark picture of how quickly Ask Sage’s addressable market is eroding.

The Navy went furthest, with a spokesperson telling DefenseScoop that the sea service “has designated GenAI as the mandated platform for mission owners and all Department of the Navy users.” That word — mandated — is tough for any competing platform.

The Marine Corps officially designated GenAI.mil as its preferred LLM platform, authorized for all Marines, civilians, and contractors. The service’s AI lead called it “the first true enterprise solution for Marines to use generative AI.”

The Air Force sunset its homegrown NIPRGPT chatbot — which had over 700,000 users across the DoD — on December 31 in favor of GenAI.mil. Users were given just three weeks to transition. The Air Force and Space Force have both formally adopted GenAI.mil as their enterprise platform.

The Army declared itself an “AI-first” organization turning to GenAI.mil, though it has said it will not completely cut off access to preexisting AI tools — which includes the Ask Sage-powered Enterprise LLM Workspace.

The Coast Guard is apparently the sole holdout, developing its own prototype called “Ask Hamilton,” which does not have any apparent relationship with Ask Sage.

The military did not respond to Hunterbrook’s request for comment.

Ask Sage Hanging On

Ask Sage may retain some defensible territory — for now.

Its alleged authorization at Impact Level 6 and Top Secret gives it access to classified environments that GenAI.mil, limited to IL5 as of two months ago, cannot yet reach. And its “Ask Sage Edge” product purportedly serves disconnected, degraded, and intermittent environments — forward operating bases and maritime command centers where cloud-based platforms can’t operate.

The Army’s decision to maintain access to its existing AI tools, including the Ask Sage-powered Enterprise LLM Workspace, provides a near-term revenue backstop. The $49 million IDIQ contract runs for five years, and existing task orders don’t vanish overnight, though they, alone, don’t explain a $250 million valuation.

And Chaillan, for one, isn’t acting like all is well and good.

In a totally separate LinkedIn post, he accused the Air Force of retaliating against him: “A DISGRACE!”

The retaliation, he argued, was for publicly resigning from the Air Force in 2021 with a blistering public letter warning that the Pentagon was too slow to adopt new technology.

The irony: Four years later, the Pentagon moved fast enough to create an alternative to his product before the acquisition ink was dry.

Authors

Sam Koppelman is a New York Times best-selling author who has written books with former United States Attorney General Eric Holder and former United States Acting Solicitor General Neal Katyal. Sam has published in the New York Times, Washington Post, Boston Globe, Time Magazine, and other outlets — and occasionally volunteers on a fire speech for a good cause. He has a BA in Government from Harvard, where he was named a John Harvard Scholar and wrote op-eds like “Shut Down Harvard Football,” which he tells us were great for his social life.

Blake Spendley joined Hunterbrook from the Center for Naval Analyses (CNA), where he led investigations as a Research Specialist for the Marine Corps and US Navy. He built and owns the leading open-source intelligence (OSINT) account on X/Twitter, called @OSINTTechnical (over 1 million followers), which also distributes Hunterbrook Media reporting. His OSINT research has been published in Bloomberg, the Wall Street Journal, and The Economist, among other top business outlets. He has a B.A. in Political Science from USC.

Editor

Wendy Nardi joined Hunterbrook after working as a developmental and copy editor for academic publishers, government agencies, Fortune 500 companies, and international scholars. She has been a researcher and writer for documentary series and a regular contributor to The Boston Globe. Her other publications range from magazine features to fiction in literary journals. She has an MA in Philosophy from Columbia University and a B.A. in English from the University of Virginia.

Hunterbrook Media publishes investigative and global reporting — with no ads or paywalls. When articles do not include Material Non-Public Information (MNPI), or “insider info,” they may be provided to our affiliate Hunterbrook Capital, an investment firm which may take financial positions based on our reporting. Subscribe here. Learn more here.

Please contact ideas@hntrbrk.com to share ideas, talent@hntrbrk.com for work opportunities, and press@hntrbrk.com for media inquiries.

LEGAL DISCLAIMER

© 2026 by Hunterbrook Media LLC. When using this website, you acknowledge and accept that such usage is solely at your own discretion and risk. Hunterbrook Media LLC, along with any associated entities, shall not be held responsible for any direct or indirect damages resulting from the use of information provided in any Hunterbrook publications. It is crucial for you to conduct your own research and seek advice from qualified financial, legal, and tax professionals before making any investment decisions based on information obtained from Hunterbrook Media LLC. The content provided by Hunterbrook Media LLC does not constitute an offer to sell, nor a solicitation of an offer to purchase any securities. Furthermore, no securities shall be offered or sold in any jurisdiction where such activities would be contrary to the local securities laws.

Hunterbrook Media LLC is not a registered investment advisor in the United States or any other jurisdiction. We strive to ensure the accuracy and reliability of the information provided, drawing on sources believed to be trustworthy. Nevertheless, this information is provided “as is” without any guarantee of accuracy, timeliness, completeness, or usefulness for any particular purpose. Hunterbrook Media LLC does not guarantee the results obtained from the use of this information. All information presented are opinions based on our analyses and are subject to change without notice, and there is no commitment from Hunterbrook Media LLC to revise or update any information or opinions contained in any report or publication contained on this website. The above content, including all information and opinions presented, is intended solely for educational and information purposes only. Hunterbrook Media LLC authorizes the redistribution of these materials, in whole or in part, provided that such redistribution is for non-commercial, informational purposes only. Redistribution must include this notice and must not alter the materials. Any commercial use, alteration, or other forms of misuse of these materials are strictly prohibited without the express written approval of Hunterbrook Media LLC. Unauthorized use, alteration, or misuse of these materials may result in legal action to enforce our rights, including but not limited to seeking injunctive relief, damages, and any other remedies available under the law.

By: Blake Spendley, Sam Koppelman, Jenny Ahn, Andrew Ford

Editor: Jim Impoco

Based on Hunterbrook Media’s reporting, Hunterbrook Capital is short $UI and long a basket of comparable securities at the time of publication. Positions may change at any time. See full disclosures on our website. Our affiliate Hunterbrook Law is in conversations with litigation firms regarding potential private litigation.

The jury is out on whether Jeffrey Epstein was working with Russia. But they apparently shared a loyalty to the same wireless supplier: Ubiquiti.

Emails and photographs from the Epstein files reviewed by Hunterbrook Media reveal that Ubiquiti ($UI) equipment was integral to the communications and surveillance infrastructure on Epstein’s notorious private island, Little St. James, in the U.S. Virgin Islands.

Epstein appears to be part of a specific consumer class: bad actors who put a primacy on privacy and are drawn to Ubiquiti’s local control architecture. That seemingly includes the Russian military fighting in Ukraine, as Hunterbrook recently revealed in an investigation that has since reached millions.

Also apparently a Ubiquiti fan: Sean Combs, known as P. Diddy. This assumption is based on a Department of Justice image from a separate investigation depicting what Hunterbrook has identified as a Ubiquiti device photographed inches away from some of the “Mo Money Mo Problems” artist’s notorious collection of baby oil and sex toys.

Evidence of Ubiquiti’s role on Epstein’s island is starker, with photos pointing also to a surveillance system supported by Ubiquiti.

One photograph taken from the island by the FBI shows a box of what appears to be a UniFi Video G3 — Ubiquiti’s surveillance camera. Another photo of a storage area shows what appears to be the same camera — with Ubiquiti’s “U” logo on it — mounted near the ceiling. The FBI also seized a UniFi Video Recorder from the server rack, according to an FBI log of items collected from Little St. James.

The hardware visible in these photos goes beyond what would be needed for a simple island internet hookup. This is an enterprise-grade communications operation, according to Steven Burgess, a court-recognized expert on digital forensics. He reviewed six images of Ubiquiti equipment from the Epstein files.

“It’s more than you would need to bring the internet to it, but it’s probably not more than what you would need to coordinate several hundred devices,” Burgess told Hunterbrook, estimating the setup could support 800 to 1,000 devices — what one might expect for large gatherings or dignitaries traveling with security details.

As for what data may have been stored on the servers: “Anything and everything. Pictures, communications, emails, documents, whatever,” Burgess said. He said that there likely was a log of devices that interacted with Epstein’s Wi-Fi network, such that if a visitor were carrying a smartphone, that connection to the Wi-Fi would be recorded.

The documents show that the tech staff at Epstein’s island specifically sought out Ubiquiti products for Wi-Fi coverage, network switching, and security camera systems beginning in late 2017 and continuing through at least 2019, the year Epstein was arrested on federal sex trafficking charges.

The paper trail apparently began in December 2017, when Jermaine Ruan, who seemingly worked for Epstein on the island property known as “LSJ,” said he reached out to Ubiquiti for a quote.

“I have approach Ubiquiti network solutions … to quote me their recommendation for equipment for the wifi,” Ruan wrote in a December 28, 2017, email to Epstein.

The email thread reveals an island struggling with significant infrastructure challenges: “Areas that don’t have internet due to fiber damage … will be set up with a wireless bridge to bring them back online.”

By January 2018, Ruan was compiling detailed procurement lists for network equipment. In response, Epstein wrote, “order today,” referring to an email list labeled “LSJ AMAZON LIST” that included multiple Ubiquiti products.

The installation appears to have continued through March 2019. “I will be scheduling myself to be on LSJ from Wednesday to Friday … to install, configure, and setup the new incoming equipment to the island,” Liston Thomas, a tech contractor, wrote in a March 15, 2019, email. Further back in the email chain is a UPS delivery confirmation on the same day from Ubiquiti to Jeffrey Epstein.

Privacy Primacy: Why Not Google?

Perhaps unsurprisingly, the preference for Ubiquiti may have been driven by a need for privacy and local control. When discussing home Wi-Fi options for a separate property in Connecticut in July 2018, a technology consultant named James Ce explicitly warned against Google’s mesh Wi-Fi products.

“Software wise, it’s even worse,” Ce wrote of Google Wi-Fi in a July 20, 2018, email. “Google manages the configuration of the routers and the traffic on their servers remotely. Besides the privacy nightmare scenario there — they literally can use every site you visit to build profiles of you — if their servers go down, so does your internet.”

Ce instead recommended Ubiquiti’s consumer mesh product: “The Ubiquiti Labs Amplifi HD wireless mesh system should be perfect for your home.”

Unanswered Questions

Hunterbrook found no evidence Ubiquiti knew about the end use of its equipment on Epstein’s island. The company’s products are widely used in commercial and residential deployments. Other than the UPS confirmation on a delivery from Ubiquiti, there is no indication the company had knowledge of its recording devices being used by a sex offender. (Which, you know, was probably kind of the point!)

The documents don’t reveal what specific footage was captured by the camera systems, but they indicate that the storage devices themselves have been seized by the DOJ as potential evidence.

Ubiquiti did not respond to multiple requests for comment.

Authors

Blake Spendley joined Hunterbrook from the Center for Naval Analyses (CNA), where he led investigations as a Research Specialist for the Marine Corps and US Navy. He built and owns the leading open-source intelligence (OSINT) account on X/Twitter, called @OSINTTechnical (over 1 million followers), which also distributes Hunterbrook Media reporting. His OSINT research has been published in Bloomberg, the Wall Street Journal, and The Economist, among other top business outlets. He has a B.A. in Political Science from USC.

Sam Koppelman is a New York Times best-selling author who has written books with former United States Attorney General Eric Holder and former United States Acting Solicitor General Neal Katyal. Sam has published in the New York Times, Washington Post, Boston Globe, Time Magazine, and other outlets — and occasionally volunteers on a fire speech for a good cause. He has a B.A. in Government from Harvard, where he was named a John Harvard Scholar and wrote op-eds like “Shut Down Harvard Football,” which he tells us were great for his social life.

Jenny Ahn joined Hunterbrook after serving many years as a senior analyst in the US government. She is a seasoned geopolitical expert with a particular focus on the Asia-Pacific and has diverse overseas experience. She has an M.A. in International Affairs from Yale and a B.S. in International Relations from Stanford. Jenny is based in Virginia.

Andrew Ford is an investigative journalist who exposed systemic flaws and prompted reforms in healthcare, business, policing, and state government. His reporting was published by ProPublica, USA Today, The Arizona Republic, Asbury Park Press, and Florida Today. He holds a journalism bachelor’s from the University of Florida and is based in Phoenix, Arizona.

Editor

Jim Impoco is the award-winning former editor-in-chief of Newsweek who returned the publication to print in 2014. Before that, he was executive editor at Thomson Reuters Digital, Sunday Business Editor at The New York Times, and Assistant Managing Editor at Fortune. Jim, who started his journalism career as a Tokyo-based reporter for The Associated Press and U.S. News & World Report, has a master’s in Chinese and Japanese History from the University of California at Berkeley.

Graphic

Dan DeLorenzo is a creative director with 25 years reporting news through visuals. Since first joining a newsroom graphics department in 2001, he has built teams at Bloomberg News, Bridgewater Associates, and the United Nations, and published groundbreaking visual journalism at The Wall Street Journal, Associated Press, The New York Times, and Business Insider. A passion for the craft has landed him at the helm of newsroom teams, on the ground in humanitarian emergencies, and at the epicenter of the world’s largest hedge fund. He runs DGFX Studio, a creative agency serving top organizations in media, finance, and civil society with data visualization, cartography, and strategic visual intelligence. He moonlights as a professional sailor working toward a USCG captain’s license and is a certified Pilates instructor.

Hunterbrook Media publishes investigative and global reporting — with no ads or paywalls. When articles do not include Material Non-Public Information (MNPI), or “insider info,” they may be provided to our affiliate Hunterbrook Capital, an investment firm which may take financial positions based on our reporting. Subscribe here. Learn more here.

Please contact ideas@hntrbrk.com to share ideas, talent@hntrbrk.com for work opportunities, and press@hntrbrk.com for media inquiries.

LEGAL DISCLAIMER

© 2026 by Hunterbrook Media LLC. When using this website, you acknowledge and accept that such usage is solely at your own discretion and risk. Hunterbrook Media LLC, along with any associated entities, shall not be held responsible for any direct or indirect damages resulting from the use of information provided in any Hunterbrook publications. It is crucial for you to conduct your own research and seek advice from qualified financial, legal, and tax professionals before making any investment decisions based on information obtained from Hunterbrook Media LLC. The content provided by Hunterbrook Media LLC does not constitute an offer to sell, nor a solicitation of an offer to purchase any securities. Furthermore, no securities shall be offered or sold in any jurisdiction where such activities would be contrary to the local securities laws.

Hunterbrook Media LLC is not a registered investment advisor in the United States or any other jurisdiction. We strive to ensure the accuracy and reliability of the information provided, drawing on sources believed to be trustworthy. Nevertheless, this information is provided “as is” without any guarantee of accuracy, timeliness, completeness, or usefulness for any particular purpose. Hunterbrook Media LLC does not guarantee the results obtained from the use of this information. All information presented are opinions based on our analyses and are subject to change without notice, and there is no commitment from Hunterbrook Media LLC to revise or update any information or opinions contained in any report or publication contained on this website. The above content, including all information and opinions presented, is intended solely for educational and information purposes only. Hunterbrook Media LLC authorizes the redistribution of these materials, in whole or in part, provided that such redistribution is for non-commercial, informational purposes only. Redistribution must include this notice and must not alter the materials. Any commercial use, alteration, or other forms of misuse of these materials are strictly prohibited without the express written approval of Hunterbrook Media LLC. Unauthorized use, alteration, or misuse of these materials may result in legal action to enforce our rights, including but not limited to seeking injunctive relief, damages, and any other remedies available under the law.

By: Sam Koppelman, Till Daldrup

Editor: Wendy Nardi

Based on Hunterbrook Media’s reporting, Hunterbrook Capital is short $SOC at the time of publication. Positions may change at any time. See full disclosures on our website.

This morning, Sable Offshore ($SOC) disclosed that it had received subpoenas from both the U.S. Attorney’s Office for the Southern District of New York (SDNY) and the Securities and Exchange Commission (SEC).

The disclosure — buried in a prospectus dated February 2 for a $250 million stock offering — confirms the Exxon spinout is now facing federal scrutiny over “issues raised in an October 31, 2025 report by Hunterbrook Media.”

The Hunterbrook investigation revealed apparent incidents of potential selective disclosure of material information to certain investors including the legendary lefty golfer Phil Mickelson, among other issues at Sable.

Sable announced in November that members of its board had formed a special committee to independently investigate the claims made in the Hunterbrook investigation — and now says it is “providing documents and cooperating with the Government Requests.”

“If the SEC or SDNY were to conclude that an enforcement action is appropriate, the SEC could impose civil penalties and fines, and other sanctions against us or against our current and former officers and directors,” the company wrote. “The SDNY could impose criminal penalties.”

Sable has been seeking to relaunch an idled pipeline that runs through Santa Barbara, California. The relaunch has become the focus of civil and criminal charges, as well as a flashpoint between the Trump administration and the California attorney general. You can read Hunterbrook’s update on Sable from January here.

Sable’s prospectus revealed the company had about $98 million in cash as of the end of 2025, after conducting a separate offering in November.

Sable could not be reached for comment by time of publication.

Authors

Sam Koppelman is a New York Times best-selling author who has written books with former United States Attorney General Eric Holder and former United States Acting Solicitor General Neal Katyal. Sam has published in the New York Times, Washington Post, Boston Globe, Time Magazine, and other outlets — and occasionally volunteers on a fire speech for a good cause. He has a BA in Government from Harvard, where he was named a John Harvard Scholar and wrote op-eds like “Shut Down Harvard Football,” which he tells us were great for his social life.

Till Daldrup joined Hunterbrook from The Wall Street Journal, where he focused on open-source investigations and content verification. In 2023, he was part of a team of reporters who won a Gerald Loeb Award for an investigation that revealed how Russia is stealing grain from occupied parts of Ukraine. He has an M.A. in Journalism from New York University and a B.S. in Social Sciences from University of Cologne. He’s also an alum of the Cologne School of Journalism (Kölner Journalistenschule).

Editor

Wendy Nardi joined Hunterbrook after working as a developmental and copy editor for academic publishers, government agencies, Fortune 500 companies, and international scholars. She has been a researcher and writer for documentary series and a regular contributor to The Boston Globe. Her other publications range from magazine features to fiction in literary journals. She has an MA in Philosophy from Columbia University and a BA in English from the University of Virginia.

Hunterbrook Media publishes investigative and global reporting — with no ads or paywalls. When articles do not include Material Non-Public Information (MNPI), or “insider info,” they may be provided to our affiliate Hunterbrook Capital, an investment firm which may take financial positions based on our reporting. Subscribe here. Learn more here.

Please email ideas@hntrbrk.com to share ideas, talent@hntrbrk.com for work, or press@hntrbrk.com for media inquiries.

LEGAL DISCLAIMER

© 2026 by Hunterbrook Media LLC. When using this website, you acknowledge and accept that such usage is solely at your own discretion and risk. Hunterbrook Media LLC, along with any associated entities, shall not be held responsible for any direct or indirect damages resulting from the use of information provided in any Hunterbrook publications. It is crucial for you to conduct your own research and seek advice from qualified financial, legal, and tax professionals before making any investment decisions based on information obtained from Hunterbrook Media LLC. The content provided by Hunterbrook Media LLC does not constitute an offer to sell, nor a solicitation of an offer to purchase any securities. Furthermore, no securities shall be offered or sold in any jurisdiction where such activities would be contrary to the local securities laws.

Hunterbrook Media LLC is not a registered investment advisor in the United States or any other jurisdiction. We strive to ensure the accuracy and reliability of the information provided, drawing on sources believed to be trustworthy. Nevertheless, this information is provided “as is” without any guarantee of accuracy, timeliness, completeness, or usefulness for any particular purpose. Hunterbrook Media LLC does not guarantee the results obtained from the use of this information. All information presented are opinions based on our analyses and are subject to change without notice, and there is no commitment from Hunterbrook Media LLC to revise or update any information or opinions contained in any report or publication contained on this website. The above content, including all information and opinions presented, is intended solely for educational and information purposes only. Hunterbrook Media LLC authorizes the redistribution of these materials, in whole or in part, provided that such redistribution is for non-commercial, informational purposes only. Redistribution must include this notice and must not alter the materials. Any commercial use, alteration, or other forms of misuse of these materials are strictly prohibited without the express written approval of Hunterbrook Media LLC. Unauthorized use, alteration, or misuse of these materials may result in legal action to enforce our rights, including but not limited to seeking injunctive relief, damages, and any other remedies available under the law.

By: Sam Koppelman

Editor: Wendy Nardi

Based on Hunterbrook Media’s reporting, Hunterbrook Capital is short $RR and long a basket of comparable securities at the time of publication. Positions may change at any time. See full disclosures on our website.

After Richtech ($RR) stock added more than $370 million in market cap on the announcement of a “collaboration” Tuesday, the company announced a dilutive fundraise the next morning. Microsoft tells Hunterbrook Media the engagement was a “standard” customer program with “no commercial element.” This comes after Richtech missed its 10-K deadline, hampering its ability to raise money through at-the-market offerings.

Beep, boop, fraud?

Shares of Richtech Robotics ($RR) surged more than 40% on Tuesday after the company announced a “collaboration” with Microsoft through the tech giant’s AI Co-Innovation Labs. On the back of that pump, the following morning, Richtech announced a $38 million dilutive fundraise.

But Microsoft tells Hunterbrook Media that Richtech is just a customer.

“Richtech participated in an AI Co-Innovation Lab engagement, which is a standard customer engagement focused on exploring and prototyping AI solutions using Microsoft technologies,” a Microsoft representative told Hunterbrook. “There is no commercial element in this lab engagement.”

Specifically, Richtech appears to have participated in a program Microsoft describes on its website as being available to “any customers/partners looking to implement Microsoft’s AI tools,” including “startups.” (You can apply right here — and who knows, maybe you, too, can add $370 million in market cap to your name! Expect a response in “3-5 business days.”)

This strikes a very different tone from Richtech’s press release, which cited a “close collaboration” between Richtech and Microsoft.

“By working closely with the Microsoft AI Co-Innovation Labs, our teams were able to jointly develop and deploy intelligent capabilities that strengthen reliability, enhance customer interactions, and support scalable automation across physical environments,” said Wayne Huang, founder and CEO of Richtech Robotics.

The market, at least, seemed to think this implied a meaningful relationship between the two companies — the day of the announcement, $RR saw its steepest daily percentage increase in market cap in over a year. But based on Microsoft’s statement to Hunterbrook, no material partnership appears to exist.

And to the extent Richtech actually achieved anything with this press release, its signature accomplishment appears to have been convincing the market that an ordinary, complimentary customer training session amounted to some kind of partnership. (Sherwood News, owned by Robinhood: “Richtech Robotics soars after announcing partnership with Microsoft.” Benzinga: “Richtech Robotics Stock Surges On Microsoft AI Partnership.” And so on.)

In reality, Microsoft’s AI Co-Innovation Labs are described on the company’s website as offering “one-week long complimentary, personalized development sprints for engineering teams.” In other words, the “collaboration” Richtech announced appears to be participation in a free prototyping program available to Microsoft customers — not a commercial partnership.

In response to a request for comment, Richtech directed Hunterbrook to its delayed 10-K, filed over a week prior to this press release, in which the company disclosed that it had entered a “non-commercial technology collaboration” with Microsoft. When Hunterbrook followed up to ask why Richtech didn’t specify the relationship was “non-commercial” in its press release, the company did not reply.

Richtech has been under scrutiny for months.

In September 2025, short seller Capybara Research published a report labeling Richtech a “China Hustle” and opining that it was “riddled with fraud” and “uninvestable.” The report alleged a range of misconduct including insider self-dealing, ties to individuals under active DOJ prosecution and SEC investigation, fabricated enterprise partnerships, and claims that the company simply rebrands off-the-shelf Chinese robots and resells them at massive markups. Capybara also cited a lawsuit from a former partner alleging Richtech’s robots were “defective and could not operate as promised.”

Richtech did not publicly respond to the report.

Missed SEC Deadline Raises Questions

The Microsoft announcement came just one week after Richtech finally filed its delayed 10-K with the SEC on January 20, 2026, even missing its extended deadline.

Richtech’s fiscal year ended September 30, 2025. As a smaller reporting company, it had until December 29 to file its annual report. Instead, the company filed an NT 10-K (notification of late filing), claiming “additional time is needed to finalize the financial statements.” That extension seemingly granted Richtech until January 13 to file. The deadline came and went with no filing.

Failing to file timely reports, as required by Nasdaq Listing Rule 5250(c)(1), triggers an automatic deficiency notice. Once Nasdaq’s Listings Qualifications Department issues that notice — which may already have happened behind closed doors — Richtech has just four business days to publicly disclose its deficiency notice via an 8-K filing.

A review of Richtech’s SEC filings shows no such 8-K was filed between the January 13 deadline and the January 20 10-K filing. Did Richtech receive a deficiency notice and fail to disclose it? Richtech did not answer this question in its reply to Hunterbrook’s request for comment.

The Morning After: Richtech Raises $38.7 Million in Private Offering — Taking Advantage of Stock Pump for Liquidity

Even without the fraud allegations, Richtech’s numbers are bleak.

The 10-K filing included Richtech’s financial figures, which reveal a cash-burning machine. Fiscal year 2025 revenue totaled $5.045 million, with a net loss of $15.754 million — nearly double the prior year’s $8.14 million loss. After the Microsoft announcement, the company now has a market capitalization of more than $1 billion.

Without substantial revenue, to fund its operations, Richtech has repeatedly diluted shareholders by raising money through the issuance of new shares. But missing its 10-K filing deadline limited its options to secure more cash.

To use the Form S-3 process — the streamlined registration that enables fundraising approaches like shelf offerings and ATM programs — companies must have filed all required SEC reports on time for the preceding 12 months. SEC Rule 12b-25 allows a 15-day grace period for late filers, during which the filing is “deemed” timely. Richtech missed that deadline. The 10-K officially became delinquent, and barring some departure from standard policy, the company should have lost S-3 eligibility.

That means Richtech cannot tap its shelf registration or ATM program until it has 12 consecutive months of timely filings — so the earliest it could regain eligibility is January 2027. For a company burning over $20 million a year on $5 million in revenue, losing easy access to capital markets may be an existential problem.

So Richtech found another way to raise money.

On Wednesday morning — one day after the Microsoft announcement sent shares soaring — Richtech announced a $38.7 million private placement, selling millions of shares to an institutional investor “priced at the market under Nasdaq rules.” Unlike ATM offerings, private placements under Regulation D don’t require S-3 eligibility.

The timing could be problematic. Under Nasdaq rules, the “Minimum Price” for a private placement is the lower of the closing price or trailing five-day average immediately preceding the signing of the binding agreement. We don’t yet know when the agreement was signed — that will be disclosed in the 8-K filing.

If the deal was signed after Monday’s 44% surge, part of the pump would be baked into the Minimum Price calculation, allowing the company to raise capital at a higher price than would have been possible days earlier, while the investor would still be able to buy Richtech stock meaningfully below market price.

But if the deal was signed before the Microsoft announcement, it’s even more striking: The institutional investor locked in shares at a price based on the pre-pump stock, and then the very next day, a “collaboration” announcement sent the stock soaring — meaning the institution is now sitting on immediate paper gains while retail investors bought in at the top.

The playbook, in other words, seems to be: Miss your 10-K deadline, limiting your ability to fundraise; announce a “close collaboration” that in reality appears to be a one-week program open to just about anybody; watch the stock surge 44%; then disclose a private placement the next morning, allowing institutional investors to buy shares at a cheaper price than retail.

We may never know what Richtech learned about software engineering at the Microsoft Innovation Lab. But what seems clear is these guys could teach a course of their own on financial engineering.

Author

Sam Koppelman is a New York Times best-selling author who has written books with former United States Attorney General Eric Holder and former United States Acting Solicitor General Neal Katyal. Sam has published in the New York Times, Washington Post, Boston Globe, Time Magazine, and other outlets — and occasionally volunteers on a fire speech for a good cause. He has a BA in Government from Harvard, where he was named a John Harvard Scholar and wrote op-eds like “Shut Down Harvard Football,” which he tells us were great for his social life. Sam is based in New York City.

Editor

Wendy Nardi joined Hunterbrook after working as a developmental and copy editor for academic publishers, government agencies, Fortune 500 companies, and international scholars. She has been a researcher and writer for documentary series and a regular contributor to The Boston Globe. Her other publications range from magazine features to fiction in literary journals. She has an MA in Philosophy from Columbia University and a BA in English from the University of Virginia.

Hunterbrook Media publishes investigative and global reporting — with no ads or paywalls. When articles do not include Material Non-Public Information (MNPI), or “insider info,” they may be provided to our affiliate Hunterbrook Capital, an investment firm which may take financial positions based on our reporting. Subscribe here. Learn more here.

Please email ideas@hntrbrk.com to share ideas, talent@hntrbrk.com for work, or press@hntrbrk.com for media inquiries.

LEGAL DISCLAIMER

© 2026 by Hunterbrook Media LLC. When using this website, you acknowledge and accept that such usage is solely at your own discretion and risk. Hunterbrook Media LLC, along with any associated entities, shall not be held responsible for any direct or indirect damages resulting from the use of information provided in any Hunterbrook publications. It is crucial for you to conduct your own research and seek advice from qualified financial, legal, and tax professionals before making any investment decisions based on information obtained from Hunterbrook Media LLC. The content provided by Hunterbrook Media LLC does not constitute an offer to sell, nor a solicitation of an offer to purchase any securities. Furthermore, no securities shall be offered or sold in any jurisdiction where such activities would be contrary to the local securities laws.

Hunterbrook Media LLC is not a registered investment advisor in the United States or any other jurisdiction. We strive to ensure the accuracy and reliability of the information provided, drawing on sources believed to be trustworthy. Nevertheless, this information is provided “as is” without any guarantee of accuracy, timeliness, completeness, or usefulness for any particular purpose. Hunterbrook Media LLC does not guarantee the results obtained from the use of this information. All information presented are opinions based on our analyses and are subject to change without notice, and there is no commitment from Hunterbrook Media LLC to revise or update any information or opinions contained in any report or publication contained on this website. The above content, including all information and opinions presented, is intended solely for educational and information purposes only. Hunterbrook Media LLC authorizes the redistribution of these materials, in whole or in part, provided that such redistribution is for non-commercial, informational purposes only. Redistribution must include this notice and must not alter the materials. Any commercial use, alteration, or other forms of misuse of these materials are strictly prohibited without the express written approval of Hunterbrook Media LLC. Unauthorized use, alteration, or misuse of these materials may result in legal action to enforce our rights, including but not limited to seeking injunctive relief, damages, and any other remedies available under the law.

By: Jenny Ahn, Blake Spendley, Andrew Ford, Till Daldrup, Michelle Cera

Editor: Jim Impoco, Sam Koppelman, Wendy Nardi

Based on Hunterbrook Media’s reporting, Hunterbrook Capital is short $UI and long a basket of comparable securities at the time of publication. Positions may change at any time. See full disclosures on our website. Our affiliate Hunterbrook Law is in conversations with litigation firms regarding potential private litigation on behalf of Ukrainians impacted.

• Ubiquiti radio bridge antennae serve a critical communications need for the Russian military in Ukraine, including for drone operations, according to our investigation. A Ukrainian communications officer estimated 80% of Russian radio bridges they’d observed on the battlefield were made by Ubiquiti. “Ubiquiti is made for regular people — basically plug-and-play. Tons of tutorials on YouTube,” the officer explained. “There is no alternative,” a Russian vendor told Hunterbrook. The U.N. has called Russia’s drone attacks against civilians crimes against humanity.

• Hunterbrook identified at least nine Russian military units accused of war crimes, or individuals associated with those units, that use Ubiquiti equipment in Ukraine, based on a review of Telegram posts and open-source materials. One fundraising group that said it had donated Ubiquiti antennae to Russian forces is run by a terrorist convicted of the 2015 bombing of a Ukrainian government building.

• Hunterbrook tested how easily export-banned Ubiquiti equipment could reach Russian forces. Posing as a Russian military procurement officer, a reporter contacted Russian vendors and multiple official Ubiquiti distributors worldwide. Nearly a dozen agreed to sell export-banned equipment. One vendor even shared thank-you letters they said were for providing Ubiquiti equipment to the Russian military. Official distributors, including U.S.-based Multilink Solutions, agreed to ship to third countries like Turkey for pickup even after the customer identified as being based in Russia — a known sanctions evasion tactic flagged by U.S. authorities.

• The total value of Ubiquiti shipments crossing the border into Russia surged 66% after the invasion despite U.S. and EU sanctions and export controls, according to Hunterbrook’s analysis of trade records. Shipments include the latest Ubiquiti models released after the ban, suggesting ongoing access to the company’s supply chain.

• Official Ubiquiti distributors appear to have continued supplying Russia after the invasion, sometimes rerouting shipments through intermediaries in high-diversion-risk countries like Turkey and Kazakhstan, trade records show. Some appear to have used intermediaries later sanctioned by the U.S. for export control evasions.

• Russian vendors use Ubiquiti’s trademarked name and logos despite Russian intellectual property laws that would allow quick takedowns and potential criminal prosecution for infringers. In 2025, Ubiquiti filed to protect its stylized “U” logo in Russia with the Russian federal intellectual property agency — and won — but took no apparent action against these Russian vendors.

• Ubiquiti’s global distribution network is riddled with compliance red flags beyond Russia. In Paraguay, the founder and president of Flytec Computers — once responsible for 13% of Ubiquiti’s revenues — has been accused of smuggling, massive tax evasion, and money laundering in Brazil, and of involvement with the illegal housing of a large-scale bitcoin mining operation. In Mexico, Ubiquiti distributor SYSCOM was quietly acquired in 2021 by Hikvision, a Chinese state-owned company the U.S. designated that same year as part of China’s military-industrial complex.

• More than a decade after Ubiquiti was fined for “reckless disregard” for sanctions obligations when its products ended up in Iran, Hunterbrook found Ubiquiti products may be still flowing there. A current official distributor listed on Ubiquiti’s website, Yemen-based Alfa Tech, may be operating branches in Shiraz and Tehran, according to its own Persian-language advertisements. In 2014, U.S. regulators fined Ubiquiti after at least $589,000 worth of prohibited equipment was diverted to sanctioned Iranian entities. CEO Robert Pera told Forbes at the time, “It can’t happen again. If it does, I’ll be in a lot of trouble.”

• Ubiquiti openly admits “we do not have any visibility” over purchases from its distributors, but legal experts told Hunterbrook that’s not a viable defense. U.S. export controls and sanctions operate on a strict liability basis — meaning, even unwitting violations are still violations. “Ignorance is not really a practical excuse, or rather, a legal excuse,” a former senior State Department sanctions official told Hunterbrook. A sanctions compliance lawyer added, “You, doing very little effort, were able to determine that it’s available for purchase by the Russian armed forces. … The company’s compliance team should be taking additional steps to prevent that.”

• Industry experts told Hunterbrook that Ubiquiti likely has the technical means to trace products appearing inside Russia — and may already be doing so. As of September 2025, Ubiquiti appeared to have placed an IP-based ban on firmware updates for Russian users, based on complaints on a Russian forum, suggesting the company can identify device locations. Experts said Ubiquiti could even trace specific hardware back to the original distributor through serial numbers and MAC addresses. Ubiquiti shut down a user forum discussion raising concerns about its Russia business with the message, “Your post violates community guidelines.”

• Ubiquiti did not respond to repeated requests for comment.

Read the full version of our investigation on our website, where you will find eye-opening photos and graphics, as well as footnotes outlining our methodology.

A Russian soldier climbs a tower, carrying an antenna, a recent Russian newsclip shows. On the antenna is the logo of Ubiquiti ($UI), a $34 billion American tech company.

In the segment, a Russian signalman describes the importance of their job: “A break in communications means total loss of control over combat operations.”

The reporter agrees, “No communication, no front.”

Around the world, Ubiquiti’s products are so popular, they made its CEO the youngest owner of an NBA team, the Memphis Grizzlies, in 2012.

And Hunterbrook has thank-you letters supplied by a Russian Ubiquiti vendor in which members of the Russian military express their gratitude for the technology.

A monthslong Hunterbrook Media investigation found that Ubiquiti radio bridge antennae — devices used to extend Wi-Fi signals across long distances — serve a critical communications need of the Russian military, including in drone operations. The U.N. has called Russia’s drone attacks against civilians crimes against humanity. There is no adequate alternative to Ubiquiti, multiple sources, including soldiers on active duty in the war zone, told Hunterbrook.

Despite U.S. sanctions and export controls designed to keep them out of Russian hands, Ubiquiti products appear readily available. A Hunterbrook reporter, posing as a Russian army procurement officer, found nearly a dozen Russian stores standing ready to supply Russian combat troops with Ubiquiti equipment. Hunterbrook also found numerous social media posts by Russian fundraising groups and military officials seeking or reporting on Ubiquiti donations to various military units fighting in Ukraine, including some units accused of war crimes.

Enabling the supply chain is Ubiquiti’s dense network of distributors, despite layers of sanctions and export controls designed to prevent those flows.

Ubiquiti did not respond to Hunterbrook’s repeated requests for comment — but the company has said in the past that it doesn’t know where its products end up. Experts told Hunterbrook that’s no excuse.

There Is No Alternative

The newsclip was shared with a Hunterbrook reporter — posing as a Russian officer seeking to purchase Ubiquiti equipment for the Russian military — by Nina Kuznetsova. Kuznetsova has multiple Russian storefronts with domain names like ubiquiti.ru that openly sell Ubiquiti products banned from export to Russia since the war.

A representative at Kuznetsova’s store also shared thank-you letters they said were for providing Ubiquiti equipment to the Russian military.

“Radio bridges are necessary to provide wireless connection over long distances and to provide high-quality communication in hard-to-reach areas,” one letter said.

Another thanked Kuznetsova for the radio bridges “sent to the combat zone.”

Ubiquiti was founded in 2005 with a focus on “democratizing network technology on a global scale.” Its revolutionary idea — offering affordable, easy-to-use, high-quality networking equipment with local control and privacy — helped Ubiquiti spread internet connectivity to remote areas around the globe.

But that same model appears also to have made Ubiquiti the equipment of choice for the Russian military’s frontline communication needs, Hunterbrook Media found.

“There are no stable, inexpensive alternatives to this manufacturer,” said a message forwarded from Kuznetsova by her staff. “95 percent of the shipments for the SVO are Ubiquiti gear,” one of her representatives chimed in. “SVO” is an acronym for “special military operation,” the Russian euphemism for the war against Ukraine.

Ukrainian sources echoed those claims.

“They simply have no alternative,” a Ukrainian communications officer who uses the call-name Django told Hunterbrook in an interview from the war zone.

Django estimated about 80% of Russian radio bridges observed on the front line are Ubiquiti, adding he has yet to spot rival products like Mikrotik’s.1 Radio bridges are antenna-based devices that wirelessly connect two distant locations so they can share the same internet or data network.

“Ubiquiti is made for regular people — basically plug-and-play. Tons of tutorials on YouTube,” he explained. Competitor brands, like Starlink, can be disabled remotely or, like Mikrotik, are difficult to use, another Ukrainian signalman told Hunterbrook.

Hunterbrook discovered numerous Telegram posts showing aerial footage of Russian bridge antennae identified by Ukrainian experts as Ubiquiti’s on the battlefield. In one Ukrainian drone video obtained by Hunterbrook, a church steeple is seen housing over a dozen antennae with features consistent with Ubiquiti radio bridges.2

A post by Serhii “Flash” on June 19, 2024, on his Telegram channel showing what appears to be drone footage of an antenna, next to a photo of a Ubiquiti PowerBeam. “We are recording the appearance of such antennas in all directions,” the post says (translation by Hunterbrook). Source: Telegram

The Russian war effort is heavily crowdfunded, according to multiple reports, and troops frequently post on social media requesting or expressing thanks for donated gear like Ubiquiti’s intended for Russian troops on the war front — including specifically for drone operations.

“We thank the people’s militia assistance group for providing communications equipment in the form of Ubiquiti antennae,” a soldier said in a video posted on Telegram. “Reliable communications are the key to success. The enemy will be defeated — victory will be ours.”

A September 15, 2024 Telegram post by the LPR People’s Militia Assistance Group showing a photo of a drone and communication devices, including Ubiquiti bridge radios, that the group called “a parcel for the team’s UAV operators.” Source: Telegram

Nine of the military units that received Ubiquiti gear, or individuals associated with those units, have been accused in various news outlets of having committed war crimes, Hunterbrook found.

In a recent Telegram post, “Ghost_of_Novorossia” — a fundraising group run by a terrorist convicted for organizing the 2015 bombing of a Ukrainian government building — claimed to have donated Ubiquiti antennae to the 76th Guards Air Assault Division. This division was accused of looting villages in Kherson Oblast and massacring civilians in Bucha in 2022.

Among the Russian communication needs they serve, Ubiquiti’s radio bridges provide communication links to pilot drones, transmit live video, and find targets, Django said. Hunterbrook also found Russian social media posts, such as this one, indicating Ubiquiti antennae serve as a communications backbone for drones.

The cost of this connectivity can be measured in blood.

Russian drones, including “kamikaze” drones that carry explosives and are designed to crash into a target, now account for 60%-70% of casualties on the battlefield, The New York Times reported. In some areas, the drones intentionally hunt and kill Ukrainian citizens to spread terror, according to Human Rights Watch. The U.N. Independent International Commission of Inquiry on Ukraine recently called these acts crimes against humanity and war crimes.

When asked, Django agreed that there would be fewer casualties if it weren’t for Ubiquiti antennae.

“If their supply is disrupted, I believe that coordination between intelligence and FPV operators will be disrupted, and thus there will be fewer sorties and fewer casualties, both among the military and the civilian population,” he said. FPV, or first-person view, refers to drones that allow pilots to see through the drone’s camera in real time.

In Kherson, a Ukrainian city bordering occupied Crimea, drone attacks on residents have been so frequent, they’ve earned the gruesome moniker “human safari,” according to the Ukrainian monitoring group Tochnyi. “5364 civilians hit” by Russian drones since December 2023, the group said.3

And one of the largest Russian units that has been deployed to Kherson — Russia’s 18th Combined Arms Army — had, as of October 2025, received Ubiquiti bridge antennae from a Russian fundraising group, Hunterbrook found. The Telegram group going by the name “Brotherly Aid” posted in October that it had sent two Ubiquiti antennae to the “Kherson communications team” — a reference to the 18th CAA. In a separate post, the group claimed it had “repeatedly assisted” the 18th CAA in the past.

“They just mercilessly shell this poor city and poor people,” Oleksandra Ishchenko, a former resident of Kherson, told Hunterbrook. Ishchenko said she hasn’t been back to her hometown in years but hears of the drone attacks from family and friends who remain there.

“No one is protected from this.”

Ubiquiti’s Tolerance of Its Impersonators

In October 2025, Hunterbrook went undercover to test how hard it would be to obtain Ubiquiti equipment on the Russian market. Posing as a Russian military officer procuring goods for the Russian troops fighting in Ukraine, a Hunterbrook reporter contacted over a dozen Russian e-commerce sites like Kuznetsova’s openly selling Ubiquiti items in Russia.

Many of the Russian Ubiquiti vendors — including ubiquiti.ru, ubnt.ru, ubiquiti-russia.com, ru-ubiquiti.ru — have “Ubiquiti” in their domain name. Their sites look like official Ubiquiti outlets, and some brandish the Ubiquiti name, logos, product images, and styling. Ubiquiti technical resources are sometimes virtually identical (see here and here). Some of these vendors even claim to be Ubiquiti’s “official” distributor in Russia.

Certain vendors, Hunterbrook found, are suppliers to sanctioned entities. According to Russian government procurement records, UbiRussia.ru, for example, is a supplier to the Moscow Power Engineering Institute, subject to U.S. sanctions for its reported ties to the Russian military’s missile program.

At least one vendor is itself sanctioned. According to a privacy disclosure on its website, ubiquiti-Russia.com belongs to a company named Elecom, which was sanctioned by the U.S. in 2024 for conveying restricted electronics into Russia.

Hunterbrook gave each shop the same proposition: Our military unit needs Ubiquiti equipment for operations in occupied Ukrainian territories.

The reporter asked for popular Ubiquiti wireless bridges — items like Ubiquiti’s airMax PowerBeam and airMax NanoStation.

The items are classified by U.S. export authorities as sensitive dual-use goods requiring a license for export to most countries under normal circumstances.4 Immediately following Russia’s invasion, however, the U.S. effectively placed a blanket ban on their export to Russia, labeling them as high-priority tech items that Russia needs for its “brutal attack on Ukraine.5” Moreover, U.S. sanctions laws prohibit U.S. companies from sales of any kind to Russian-occupied territories in Crimea, Donetsk, and Luhansk.

Nearly a dozen Russian dealers confirmed our shopping list and promised to deliver the items to military units in the occupied territories.

One vendor, Comptek, had a condition, however: a minimum order of at least $100,000.

Why?

“You want sanctioned equipment. We carry orders starting at $100,000,” a company representative said.

When asked, two of the vendors told Hunterbrook that they no longer have an official relationship with Ubiquiti, and that they receive their products through “parallel import” — a term to describe sourcing goods outside authorized distribution channels, often to bypass Western export controls.

But the exact nature of their ongoing relationship is murkier. Unlike its competitors Cisco, Fortinet, Siemens, and HPE/Aruba, Ubiquiti did not announce it would leave Russia after the 2022 invasion. Ubiquiti’s website no longer contains a list of its official distributors and resellers in Russia and Belarus, but an archived snapshot of the site taken in 2014 shows a long list of historic Russian distributors, many of which still operate and sell Ubiquiti products in Russia.6

And aside from tweaking its website, Ubiquiti appears to have done little to distance itself from its former Russian partners that continue to claim ties to the company.

Ubiquiti’s intellectual property policy explicitly bans its distributors from using its trademarked names and logo without written authorization. Russia’s IP protection laws should allow Ubiquiti to take down sites bearing its domain name, with infringers facing potential civil, administrative, and criminal prosecution. In fact, in 2025, Ubiquiti moved to further secure its stylized “U” logo in Russia, filing a patent application with Russian federal intellectual property agency Rospatent, and won.

And yet, the Russian websites containing imitation Ubiquiti logos were up and running as of publication. That includes ubiquiti.ru — whose proprietor is not exactly trying to hide what she is doing. Kuznetsova resides in an apartment on the iconic Leninsky Avenue in Moscow, according to a Hunterbrook review of Russian business records. Her business partner — and based on their names, presumably her son — Andrey Kuznetsov has spent the last few years frequently traveling across Europe and Asia, even posting about about it on social media.

Charting Ubiquiti Supply Routes to Russia

Amid the layers of sanctions and export controls, how is Russia getting its hands on Ubiquiti gear?

To find out, Hunterbrook also reached out to dozens of official Ubiquiti distributors — third-party global partners contracted to market and distribute Ubiquiti goods on its behalf. Ubiquiti relies heavily on nearly 600 third-party distributors and resellers worldwide rather than internal sales teams for marketing and distribution.7

Posing as a Russian customer, a Hunterbrook reporter asked them for Ubiquiti bridge antennae.

One company, ITsupport, based in Armenia, agreed to ship the U.S.-export-controlled items directly to Moscow. Two companies refused to sell at all, consistent with U.S. export policies on Russia.

Most of the others that responded refused to ship the goods directly to Russia, but agreed to deliver them to an address in their own country or to a third country for pickup — sometimes after having heard the end user was the Russian military fighting against Ukraine.

That practice — known as “transshipment” — struck close to home when a U.S. company, Multilink Solutions, agreed to ship 450 Ubiquiti devices after learning they were ultimately destined for Russia.

Initially, a Multilink representative said they could deliver the items directly to Russia.

Soon after, however, they followed up with a message that they couldn’t ship Ubiquiti devices to Russia, after all.

But they were able to work out a solution: a delivery address outside of Russia. After the reporter provided an address in Turkey, Multilink sent an invoice confirming the delivery was possible.

Multilink did not respond to Hunterbrook’s request for comment on these interactions, its compliance policies, and its relationship with Ubiquiti.

Routing products through intermediaries such as companies in third countries is a known diversion tactic, according to U.S. authorities.8

Yet, trade records suggest this is indeed how a lot of Ubiquiti products are ending up in Russia.9

After Ubiquiti halted direct shipments to Russia following the invasion, Ubiquiti products continued to flow into the country from dozens of intermediaries, according to Hunterbrook’s analysis of trade data.10

In fact, the total value of Ubiquiti shipments to Russia surged 66% after the invasion, even accounting for any spike in shipments immediately before Russia’s invasion, possibly as distributors rushed their shipments to Russia in anticipation of export bans.11

And that number may not capture the true scale of the transfer of Ubiquiti products, as some companies falsify or omit identifying information like the country of origin, brand, and volume or value of the shipments to hide their tracks. For example, nearly $4 million of Ubiquiti shipments recorded after the U.S. export ban lacked the shipper’s name.

The items shipped include the latest models of Ubiquiti devices that were released for the first time after the ban, suggesting ongoing access to Ubiquiti’s central warehouses by these networks.

Among the exporters that supplied Russia with Ubiquiti goods after the invasion are about half a dozen official Ubiquiti distributors, with little apparent effort to obfuscate their activity.

Others appear to have halted direct shipments to Russia after the invasion but rerouted their shipments to Russia through intermediaries in third countries like Turkey and Kazakhstan, Hunterbrook’s review of trade data suggests.

Sometimes, their intermediaries were caught by Western governments for sanctions and export controls evasions, confirming their transshipment tactics. At least 10 intermediaries in countries like Turkey, the UAE, and China that were identified in the trade data as exporting Ubiquiti products to Russia have since been sanctioned by the U.S., Hunterbrook found.

Discomp: Distributor Playbook for Keeping Ubiquiti Flowing Into Russia

A closer look at Discomp S.R.O., a longstanding Ubiquiti distributor based in the Czech Republic, illustrates these tactics, according to Hunterbrook’s analysis of trade data.

Discomp shipped $1.8 million worth of Ubiquiti goods to Ivan Ionov Vladimirovich, a Russian client and owner of IT equipment wholesaler WiFiMag, for more than two years after Russia’s invasion — and after Discomp disclosed a notice from Ubiquiti warning its distributors against selling its products to Russia.

WiFiMag was an official Ubiquiti reseller as of 2014 and still claims to be one on its website. When contacted by a Hunterbrook reporter posing as a Russian military end user, a WiFiMag sales representative confirmed the availability of Ubiquiti equipment.

By mid-2023, Discomp appeared finally to have stopped its direct shipments to the Russian client. But it found another way.

Later that year, Discomp appears to have begun rerouting its shipments through a Turkish company called Olimpik Gama. Russian customs records show, however, that those shipments were destined for the same Russian client, with Olimpik Gama merely serving as a freight forwarder.12

The shipments stopped in March 2024, shortly after Olimpik Gama was sanctioned by the U.S. Treasury Department for evading sanctions. Discomp then began using another Turkish company, Unico Tekstil, to continue its shipments to Russia.

Discomp’s apparent routing of shipments destined for its Russian customer through Turkish entities contrasts sharply with its disclosures to Czech regulators. In its 2024 financial statement, the company wrote that it had “coped with the consequences of geopolitical events in Russia and Ukraine” by managing to replace the losses in those markets with customers in other markets.

Discomp did not respond to Hunterbrook’s request for comment as of the time of publication.

Simple Solutions, Kazakhstan: Another Window into Distributors’ Transshipment Tactics

A closer look at Simple Solutions, an obscure IT company in Kazakhstan, provides another glimpse into how re-exporters in third countries may be obtaining Ubiquiti equipment from official distributors to supply the Russian market after the export ban.

Multiple clues point to Simple Solutions being a shell company with the sole job of facilitating transshipments to Russia.

Simple Solutions was founded on March 11, 2022, according to Kazakh corporate intelligence site kompra.kz, in a small town in northwest Kazakhstan 80 miles from the Russian border. Hunterbrook couldn’t find an active website or online customer-facing presence associated with Simple Solutions that would suggest a commercial operation in Kazakhstan; the contact email listed in Kazakhstan’s business registry uses mail.ru, a Russian-owned email service. The company appears to be registered at the same address as Wireless Communications13, a Kazakh IT company that describes itself as an authorized Ubiquiti reseller, according to its Instagram page.

Shortly after its founding, Simple Solutions began importing Ubiquiti products from three official Ubiquiti distributors — Latvia-based Getic, Poland-based Batna, and China-based Joint Harvest, trade records show.

During the same period, 100% of Simple Solutions’ outward shipments went to a Russian IT company called Wireless Networks.

Wireless Networks owns an e-commerce site, WMD.ru, which was also once Ubiquiti’s official distributor. A WMD representative agreed to sell Ubiquiti equipment for military units fighting in Ukraine, telling Hunterbrook that their “resellers work with foundations and volunteer organizations” — a likely reference to volunteer groups that organize donation drives for the military.

Wireless Networks is also a supplier to ru-ubiquiti.com, another Russian Ubiquiti reseller that confirmed with Hunterbrook that it has provided Ubiquiti equipment to Russian military units, including a strategic missile force academy.

Before the invasion, Wireless Networks had received the majority of its Ubiquiti shipments directly from Ubiquiti, with Euro DK, Getic’s former name, making up the rest — or about 25%.

Unlike Discomp, however, Getic immediately halted direct sales to Russia in 2022, according to trade records. A Getic sales representative contacted by an undercover Hunterbrook reporter also flatly refused to sell to a Hunterbrook reporter posing as a Russian customer, explaining that “in accordance with existing restrictions, our products cannot be sold or transferred — either directly or through intermediaries — to the Russian Federation.”

In email responses to Hunterbrook’s request for comment, Getic CEO Igor Dekteryov said the company complies with “all legal and regulatory requirements applicable to our professional activities” with “the utmost care and diligence,” and that the company considers the military use of its products “unacceptable.” Dekteryov told Hunterbrook that it conducted due diligence on Simple Solutions and found nothing alarming.14

Batna responded to Hunterbrook’s request for comment, saying it complies with applicable sanctions regimes with “utmost diligence,” including by making their customers aware of the obligation to comply with export bans. Joint Harvest did not respond to Hunterbrook’s request for comments as of the time of publication.

Under EU sanctions law, EU-distributors like Getic and Discomp are not allowed to ship even to third countries like Turkey and Kazakhstan if they have reason to believe diversion is likely. In 2023, the EU passed a series of anti-circumvention mechanisms to hold EU companies legally liable for such exports that end up in Russia.

Simple Solutions’ new role in the Ubiquiti network is just one example. Trade records show, for example, just one Turkish exporter shipping a total of just $195,416 worth of Ubiquiti goods into Russia in the 26 months before the U.S. export ban. In the same time period after the ban, those numbers jumped to 18 Turkish exporters shipping about $3 million worth of Ubiquiti product — over a 1,000% increase.

Similarly, there are no records of the export of Ubiquiti goods from Kazakhstan, the UAE, Hong Kong, or Kyrgyzstan before the ban in the trade data Hunterbrook reviewed — those shipments appear for the first time only after the ban (see Case Study below for another example of how Ubiquiti distributors are rerouting shipments to Russia).

Ignorance Is No Excuse

The permissive nature of Ubiquiti’s distribution ecosystem is no accident. Ubiquiti’s reliance on third-party distributors is part of the lean cost model that’s been credited as key to its impressive margins — and which helped fund founder and CEO Robert Pera’s reported $377 million purchase of the Memphis Grizzlies in 2012, allowing him to become the youngest owner of an NBA team at the time.

It’s also a model that limits Ubiquiti’s control over its distribution. The company openly told its investors in its latest annual SEC filing, “We do not have any visibility on the location or extent of purchases of our products by individual network operators and service providers from our distributors.”

But that’s not a defensible excuse for violating export control and sanctions laws against providing sensitive goods to prohibited end users, legal experts told Hunterbrook.

“Ignorance is not really a practical excuse, or rather, a legal excuse, in most circumstances,” Richard Nephew, senior research scholar at Columbia University, told Hunterbrook. Nephew has served as the principal deputy coordinator for sanctions policy at the State Department.

That’s because U.S. export controls and sanctions regimes apply “strict liability,” Nephew explained, echoing other experts who also emphasized this point.

“Even if you have no knowledge that you’ve engaged in a violation … it’s still a violation,” Trevor Schmitt, a sanctions compliance lawyer at Arnold and Porter, told Hunterbrook. Strict liability is a legal principle that holds a party responsible even when there is no clear indication of negligence or intent to cause harm.

Even just having had a reason to know, or willfully avoiding knowledge, could be grounds to hold a company liable, experts explained. U.S. export laws define “knowledge” broadly to include even an awareness of a high probability and willful avoidance of knowledge.15

That definition may be difficult for a $34 billion global IT company to argue its way around.

How Much Does — Or Should — Ubiquiti Know?

Ubiquiti claims “we do not have any visibility” over the distribution network. But in a 2014 earnings call, the CEO admitted it was able to monitor its channels, specifically to Russia.

“We know that there’s a couple other guys that more than likely move some products into Russia,” CEO Robert Pera said, referring to Ubiquiti’s distributors. Ubiquiti hasn’t held an earnings call since 2018.

And a company like Ubiquiti could have the technical means to trace its products appearing inside Russia — possibly even back to the specific distributors or resellers involved — according to three industry experts Hunterbrook interviewed.

As of September 2025, Ubiquiti appeared to have placed an IP-based ban on firmware updates, based on complaints on a Russian user forum, suggesting Ubiquiti is able to identify a product’s location and deny service if the product user seeks an update.16

IP-based blocking is an inexact science and unreliable as an enforcement tool, experts told Hunterbrook. Russian users posted messages on the forum identifying simple workarounds like using a VPN to mask their location when connecting to Ubiquiti’s servers.

But IP-based detection is possible, and virtually impossible for a company like Ubiquiti to miss, industry experts told Hunterbrook.17

Large companies monitor IP flows on their cloud backends and can identify when something happens in a restricted region, especially if the users don’t turn on a VPN, Jim Salter, a veteran network-infrastructure analyst and an IT magazine contributor, told Hunterbrook. “Yeah, that’s instantly going to light up,” he said. “It’s going to look like a flare going off.”

And even when a user is trying to mask it, companies have ways to approximate the true location, according to the CEO of a U.S. smart-home integration firm that extensively deploys Ubiquiti products.

“If they chose to go digging, you can usually see, hop-to-hop-to-hop, this isn’t really in London — this is probably originating closer to Russia,” the CEO said.

Ubiquiti could even identify the specific hardware pinging Ubiquiti’s servers and trace it back to the original shipper. “From the serial number or the MAC address, you’d be able to see what unit that is and who it was originally sold to,” Lennert Buytenhek, a Lithuania-based networking engineer and author, told Hunterbrook.

But instead of prodding its distributors, applying its channel analytics for compliance, and exercising its technical controls to identify prohibited end users in Russia, Ubiquiti seems to have looked the other way.

In a post criticizing Ubiquiti’s alleged “doing business in Russia” on its online user community, an administrator shut down the discussion with a line:

“Your post violates community guidelines.18”

Ubiquiti’s Famed Efficiency: At the Expense of Compliance, Lives Lost

If a company could, should, or does know of potential diversions, what should it do to take responsibility and demonstrate compliance?

Hunterbrook asked 10 legal and sanctions experts.

The expectation isn’t for a company to prevent every violation along its supply chain, but rather for it to demonstrate that it took reasonable measures to prevent such violations, experts told Hunterbrook. Companies “can’t possibly know where every one of their devices ends up,” Doug Jacobson, trade attorney and an associate professor at American University, told Hunterbrook.

That doesn’t mean companies are off the hook. “There is an affirmative obligation for the company to conduct due diligence on end users,” Schmitt told Hunterbrook.

Similarly, Nephew said: ”The U.S. has set a pretty high bar on you have to know your customer.”

While there are no hard and fast rules on what those due diligence standards and processes must look like, experts offered concrete examples of typical measures that regulators expect from large companies involved in high-risk areas, such as using AI-enhanced screening software for anything Russia related; requiring distributors to submit end-user statements; and conducting field visits to countries like Turkey.

Detailed guidance from the U.S. Treasury Department on effective sanctions compliance includes conducting due diligence on customers, the supply chain, and any intermediaries through a compliance team led by a dedicated compliance officer; written policies and procedures; routine training; and auditing.

An effective compliance system should detect — as Hunterbrook did — when distributors openly export prohibited products directly to Russia or through identifiable transshipment routes.

“The conclusion is,” Schmitt told Hunterbrook, “that you, doing very little effort, were able to determine that it’s available for purchase by the Russian armed forces. … The company’s compliance team should be taking additional steps to prevent that, and should be identifying distributors that are the bad actors here.”

It’s unclear if Ubiquiti has implemented any of these practices. Ubiquiti did not respond to multiple Hunterbrook requests for comment on detailed questions regarding its compliance practices.

Ubiquiti’s Questionable Compliance Culture

Ubiquiti says it’s committed to compliance. But none of the policy documents Hunterbrook found — not the company’s Code of Business Conduct, a 2022 warning to a distributor, a 2019 information sheet, “U.S. Export Controls on Ubiquiti Products,” or a 2015 guide to U.S. export controls — described concrete mechanisms to enforce that commitment.19

A former Ubiquiti employee who oversaw product distribution provided a concerning glimpse of the company’s compliance culture.

The source, requesting anonymity to protect his professional reputation, described the company’s compliance department as a one-man shop, without an automated system to monitor compliance and identify red flags.

“All those processes that you would see at a large scale business don’t really solve the problem,” the source said. “The CEO of Ubiquiti recognizes that and just doesn’t care about all of the extra garbage.”

Ubiquiti’s operating structure seems to reflect a lack of emphasis on compliance. With just 1,667 total employees, 71% of the workforce is dedicated to R&D, leaving just 480 employees to staff all other departments running Ubiquiti’s global sales and operations, including compliance. That’s for overseeing $2.6 billion annual sales and nearly 600 distributors and resellers worldwide.

Fast-Forward, Repeat: Ubiquiti’s Ongoing Compliance Problems Globally

Ubiquiti’s lean, R&D-focused model had no compliance program, U.S. authorities announced in 2014, leading to prohibited Ubiquiti equipment worth at least $589,000 to be diverted to Iranian entities from 2008 to 2011.

U.S. authorities slapped a relatively modest $504,225 fine on Ubiquiti for the violations, taking into account the fact that the company had no history of violations and “took remedial action.”

“If violations should occur in the future, the response of regulators may be more severe in light of prior compliance concerns,” Ubiquiti told investors in their 2014 10-K filing.

“It can’t happen again,” the company’s CEO, Robert J. Pera, said in a 2011 Forbes article. “If it does, I’ll be in a lot of trouble.”

Over 10 years later, Ubiquiti’s compliance posture may not have improved much: Hunterbrook’s investigation suggests a Ubiquiti distributor may once again be supplying Iran.

UAE-based Alfa Tech, which Ubiquiti lists as an official distributor on its website, may be running branches in Iran, according to its Persian-language advertisements archived by Hunterbrook in May last year. The ads list Alfa Tech’s website and emails associated with Alfa Tech for contact, including the same email address associated with Alfa Tech’s domain name.

On Alfa Tech’s Telegram channel, which Hunterbrook archived in April last year, the first contact number listed is for the company’s Shiraz office, in Iran. The channel no longer contains references to an Iran branch. Alfa Tech did not respond to Hunterbrook’s request for confirmation whether the company maintains offices in Iran.

And Ubiquiti’s dealings with suspect partners don’t stop there, Hunterbrook found.

In South America, Osni Arruda, founder and president of Flytec Computers, a Paraguay-based Ubiquiti distributxor once responsible for 13% of Ubiquiti’s revenues, has been accused of smuggling, massive tax evasion, and money laundering in Brazil, and of involvement with the illegal housing of a large-scale bitcoin mining operation in Paraguay. Hunterbrook also found that, as of September 2025, Arruda also appears to own a building now occupied by IT distribution company Tecnotropolis, according to the Florida state corporate registry and Google Maps.20 Tecnotropolis has been accused of abetting a more than $1 billion money laundering operation tied to narcotrafficking, court records show.

And in Mexico, Ubiquiti distributor SYSCOM, which has a subsidiary that operates in the U.S. that is also an official Ubiquiti distributor, was quietly acquired by a Chinese state-owned IT company Hikvision in 2021, according to Hikvision’s annual report. That same year, the U.S. designated Hikvision as part of China’s military-industrial complex and banned the importing of communications equipment from Hikvision into the U.S.

The Cost of Noncompliance in Dollars … and Blood

The civil — and potentially criminal — liability a company faces for these types of compliance failures can be significant. The U.S. government warns that companies face administrative penalties including the “denial of export privileges, and placement of individuals and entities on lists that restrict or prohibit their involvement in export and reexport transactions,” as well as monetary penalties up to “twice the value of the transaction.”

Legal experts told Hunterbrook, however, that a criminal case usually requires evidence of the company’s explicit knowledge of diversion and willful violation.

That evidence may be hard to obtain without a formal government investigation with subpoena authority.

In the EU, enforcement can be uneven across jurisdictions, with some countries applying stricter definitions of EU laws to prosecute violations, Alex Prezanti, a UK-based lawyer specializing in international law and sanctions enforcement, told Hunterbrook. “Different member states have completely different standards for how they enforce EU legislation,” he said.

Beyond potential penalties for violating export controls and sanctions, Ubiquiti’s actions — or inactions — could expose the company to human-rights litigation in U.S. courts, according to Martin Flaherty, a law professor at Columbia University.

Flaherty raised the possibility of an Alien Tort Statute lawsuit in U.S. federal court by Ukrainian civilians who were harmed in Russian drone attacks. Such a case would rest on a theory that Ubiquiti — a U.S. corporation — aided and abetted Russian war crimes by providing connectivity for the Russian military to operate drones to kill civilians.

But the real cost of Ubiquiti’s compliance failures can’t be measured in regulatory fines or litigation damages. It’s the risk to Ukrainian lives — soldiers and civilians alike — killed in Russia’s brutal campaign in Ukraine, made possible in part by Ubiquiti technology.

Oleksandr Yasenchuk, an activist helping arm Ukraine’s defense, had a plea for the American tech giant: “I would ask them, if possible, to stop the sale of their technology to Russia.”

“They must understand,” he said, that the company’s money carries “the blood of Ukrainians.”

Oleksandra Ishchenko, a former Kherson resident, also urged Ubiquiti to just “look at all the materials” publicly available, to see “how much harm this brings, how many human lives it has taken.” Not doing so, she said, “equals supporting war and the murder of Ukrainians.”

“We are not stupid people, and we understand that this whole process is possibly not difficult for a large company to trace where their products are.”

“I think it’s a choice,” she added.

Appendix / Case Study: A Deep Dive into the Network Supplying Ubiquiti Goods to NAG, a Major Russian Distributor

A deep dive into the import routes of one Russian IT equipment seller, NAG, helps show how the Russian market may be obtaining prohibited Ubiquiti goods through various trade routes that evade export controls and sanctions — and how Ubiquiti official distributors may be abetting those tactics.

NAG was once Ubiquiti’s official distributor, according to the latter’s archived webpage. Among its government customers were PJSC Rostelecom and Russia’s Federal Security Service, according to clearspending.ru, an open-source portal for Russian government data. Rostelcom is Russia’s largest internet service provider and a state-owned company sanctioned by the U.S. The FSS, the U.S.-sanctioned successor to the KGB, is Russia’s main domestic intelligence service. NAG also holds a license with the FSS to develop and supply cryptographic equipment for the service, according to Russian corporate registry records retrieved from the Federal Tax Service of Russia.

NAG also owns a 49% stake in a company called Nagtech, which was founded in 2021 expressly to support the Russian government’s policy of import substitution for key telecommunications equipment, according to its own website.

When contacted by a Hunterbrook reporter posing as a Russian customer seeking to purchase Ubiquiti devices for the Russian military in the occupied territories in November 2022, NAG confirmed that the items were in stock and promised to deliver them via local postal services. According to U.S. export control laws, the requested items are prohibited from export to Russia.

Hunterbrook’s analysis of trade records suggests that Discomp, NAG’s historic supplier of Ubiquiti goods, may have continued to supply NAG after the invasion, but obscured the shipment routes by removing its name from customs declarations or using intermediaries in third countries.

In the year prior to February 2022, NAG imported all of its Ubiquiti products from Discomp, valued at nearly $1.8 million, Russian customs records show. From February 2022 to April 2023, NAG imported nearly $4 million worth of Ubiquiti products, but the supplier’s name was removed from the shipment records. On some of the records, however, the location of loading was marked as Pilsen — a city in the Czech Republic where Discomp is headquartered.

Discomp may also have used a Turkish intermediary to continue supplying NAG.

From July 2022 to April 2023, Discomp exported about $2.2 million worth of Ubiquiti products to a Turkish locomotive parts company called Ervacan Makina; Ervacan was added to the Commerce Department’s Entity List in February 2024 for violating U.S. export controls on Russia. Discomp also shipped at least $1.9 million worth of Ubiquiti goods to Ervacan through a Lithuanian logistics company called Customs Warehouse UAB Didneriai.21 In many of the shipment records, the destination country was marked as Russia rather than Turkey — another sign of Ervacan’s likely role as a freight forwarder.

During nearly the same period, Ervacan supplied NAG with $6.1 million worth of largely tech equipment and parts, suggesting an active pipeline between Ervacan and NAG, although none of the shipments were labeled specifically as “Ubiquiti” brand.

NAG also used its subsidiary based in Kazakhstan to receive shipments from another Ubiquiti official distributor based in Poland, Batna.22 After the invasion, Batna exported over $530,000 dollars worth of Ubiquiti goods to NAG Kazakhstan, according to Kazakhstan trade data. During the same time frame, records show NAG Kazakhstan was exporting $1.3 million worth of networking equipment back to NAG and Nagtech, though the items were not labeled by their brand.

NAG is affiliated with Russian IT equipment sellers that also relied on transshipment routes — until the intermediaries were caught.

Beginning in July 2023, a Turkish logistics company called MBK Lojistik shipped a total of $960,000 worth of Ubiquiti goods to two customers in Russia, customs records show: about $60,000 was to NAG, and the rest was to another Russian company called Vostochny Put, a NAG partner company incorporated in June 2023.23

That shipment route ended around February 2024, when MBK Lojistik was sanctioned by the U.S. for sanctions evasions. It’s unclear who was supplying MBK Lojistik with Ubiquiti products.

NAG is also partnered with the New IT Project LLC, owner of Russian IT distributor 3Logic Group, a New IT Project representative told an undercover Hunterbrook reporter. Trade data shows that after the 2022 invasion, New IT Project imported over $5 million worth of Ubiquiti goods from Hong Kong-based AFox Corporation — until AFox was designated by the U.S. for sanctions evasions in support of Russia’s military industrial base in August 2023. It’s unclear who AFox’s suppliers were.

Banta told Hunterbrook in an email that it complies with applicable sanctions regimes with “utmost diligence,” but did not respond to Hunterbrook’s specific questions on our findings regarding NAG. Discomp did not respond to Hunterbrook’s request for comments as of the time of publication.

Authors

Jenny Ahn joined Hunterbrook after serving many years as a senior analyst in the US government. She is a seasoned geopolitical expert with a particular focus on the Asia-Pacific and has diverse overseas experience. She has an M.A. in International Affairs from Yale and a B.S. in International Relations from Stanford. Jenny is based in Virginia.

Blake Spendley joined Hunterbrook from the Center for Naval Analyses (CNA), where he led investigations as a Research Specialist for the Marine Corps and US Navy. He built and owns the leading open-source intelligence (OSINT) account on X/Twitter, called @OSINTTechnical (over 1 million followers), which also distributes Hunterbrook Media reporting. His OSINT research has been published in Bloomberg, the Wall Street Journal, and The Economist, among other top business outlets. He has a BA in Political Science from USC.

Till Daldrup joined Hunterbrook from The Wall Street Journal, where he focused on open-source investigations and content verification. In 2023, he was part of a team of reporters who won a Gerald Loeb Award for an investigation that revealed how Russia is stealing grain from occupied parts of Ukraine. He has an M.A. in Journalism from New York University and a B.S. in Social Sciences from University of Cologne. He’s also an alum of the Cologne School of Journalism (Kölner Journalistenschule).

Andrew Ford is an investigative journalist who exposed systemic flaws and prompted reforms in healthcare, business, policing, and state government. His reporting was published by ProPublica, USA Today, The Arizona Republic, Asbury Park Press, and Florida Today. He holds a journalism bachelor’s from the University of Florida and is based in Phoenix, AZ.

Michelle Cera trained as a sociologist specializing in digital ethnography and pedagogy. She completed her PhD in Sociology at New York University, building on her Bachelor of Arts degree with Highest Honors from the University of California, Berkeley. She has also served as a Workshop Coordinator at NYU’s Anthropology and Sociology Departments, fostering interdisciplinary collaboration and innovative research methodologies.

Editors

Jim Impoco is the award-winning former editor-in-chief of Newsweek who returned the publication to print in 2014. Before that, he was executive editor at Thomson Reuters Digital, Sunday Business Editor at The New York Times, and Assistant Managing Editor at Fortune. Jim, who started his journalism career as a Tokyo-based reporter for The Associated Press and U.S. News & World Report, has a Master’s in Chinese and Japanese History from the University of California at Berkeley.

Sam Koppelman is a New York Times best-selling author who has written books with former United States Attorney General Eric Holder and former United States Acting Solicitor General Neal Katyal. Sam has published in the New York Times, Washington Post, Boston Globe, Time Magazine, and other outlets — and occasionally volunteers on a fire speech for a good cause. He has a BA in Government from Harvard, where he was named a John Harvard Scholar and wrote op-eds like “Shut Down Harvard Football,” which he tells us were great for his social life.

Wendy Nardi joined Hunterbrook after working as a developmental and copy editor for academic publishers, government agencies, Fortune 500 companies, and international scholars. She has been a researcher and writer for documentary series and a regular contributor to The Boston Globe. Her other publications range from magazine features to fiction in literary journals. She has an MA in Philosophy from Columbia University and a BA in English from the University of Virginia.

Graphics

Dan DeLorenzo is a creative director with 25 years reporting news through visuals. Since first joining a newsroom graphics department in 2001, he has built teams at Bloomberg News, Bridgewater Associates, and the United Nations, and published groundbreaking visual journalism at The Wall Street Journal, Associated Press, The New York Times, and Business Insider. A passion for the craft has landed him at the helm of newsroom teams, on the ground in humanitarian emergencies, and at the epicenter of the world’s largest hedge fund. He runs DGFX Studio, a creative agency serving top organizations in media, finance, and civil society with data visualization, cartography, and strategic visual intelligence. He moonlights as a professional sailor working toward a USCG captain’s license and is a certified Pilates instructor.

Hunterbrook Media publishes investigative and global reporting — with no ads or paywalls. When articles do not include Material Non-Public Information (MNPI), or “insider info,” they may be provided to our affiliate Hunterbrook Capital, an investment firm which may take financial positions based on our reporting. Subscribe here. Learn more here.

Please email ideas@hntrbrk.com to share ideas, talent@hntrbrk.com for work, or press@hntrbrk.com for media inquiries.

LEGAL DISCLAIMER

© 2026 by Hunterbrook Media LLC. When using this website, you acknowledge and accept that such usage is solely at your own discretion and risk. Hunterbrook Media LLC, along with any associated entities, shall not be held responsible for any direct or indirect damages resulting from the use of information provided in any Hunterbrook publications. It is crucial for you to conduct your own research and seek advice from qualified financial, legal, and tax professionals before making any investment decisions based on information obtained from Hunterbrook Media LLC. The content provided by Hunterbrook Media LLC does not constitute an offer to sell, nor a solicitation of an offer to purchase any securities. Furthermore, no securities shall be offered or sold in any jurisdiction where such activities would be contrary to the local securities laws.

Hunterbrook Media LLC is not a registered investment advisor in the United States or any other jurisdiction. We strive to ensure the accuracy and reliability of the information provided, drawing on sources believed to be trustworthy. Nevertheless, this information is provided “as is” without any guarantee of accuracy, timeliness, completeness, or usefulness for any particular purpose. Hunterbrook Media LLC does not guarantee the results obtained from the use of this information. All information presented are opinions based on our analyses and are subject to change without notice, and there is no commitment from Hunterbrook Media LLC to revise or update any information or opinions contained in any report or publication contained on this website. The above content, including all information and opinions presented, is intended solely for educational and information purposes only. Hunterbrook Media LLC authorizes the redistribution of these materials, in whole or in part, provided that such redistribution is for non-commercial, informational purposes only. Redistribution must include this notice and must not alter the materials. Any commercial use, alteration, or other forms of misuse of these materials are strictly prohibited without the express written approval of Hunterbrook Media LLC. Unauthorized use, alteration, or misuse of these materials may result in legal action to enforce our rights, including but not limited to seeking injunctive relief, damages, and any other remedies available under the law.

By: Laura Wadsten, Andrew Ford

Editor: Sam Koppelman

Hunterbrook Media’s investment affiliate, Hunterbrook Capital, does not have any positions related to this article at the time of publication. Positions may change at any time. Hunterbrook Media and its affiliate Hunterbrook Law are working with a litigation firm that may bring potential legal claims based in part on our reporting. See full disclosures on our website.

Just weeks after Hunterbrook published a yearlong investigation into PBM GPOs, members of Congress grilled the CEOs of Cigna Group, UnitedHealth Group and CVS Health about our findings that their companies diverted billions of dollars in revenue away from patients and health plans.

UnitedHealth’s Stephen Hemsley, Cigna’s David Cordani and CVS’ David Joyner were called to Washington Thursday to appear before the House Committees on Energy and Commerce and Ways and Means.

The hearings covered healthcare affordability, ranging from enhanced Affordable Care Act subsidies to the focus of Hunterbrook’s reporting: so-called Group Purchasing Organizations (GPOs) established under the conglomerates’ Pharmacy Benefit Managers (PBMs).

During the Energy and Commerce hearing, legislators repeatedly cited Hunterbrook’s findings on the shell subsidiaries created by healthcare giants to get around a crackdown on PBMs pocketing rebates from pharmaceutical companies.

Cigna created Ascent in Switzerland, CVS created Zinc in Minnesota, and UnitedHealth created Emisar in Ireland. The companies dubbed these entities “GPOs” — group purchasing organizations — but our investigation suggested the new label was just that: a label.

“The big three health insurance corporations, knowing that PBM reform is around the corner, needed a different way to retain profits while still – on paper – complying with new requirements and the coming requirements of PBM reform,” said Rep. Jake Auchincloss, D-Mass.

Auchincloss asked UnitedHealth CEO Hemsley about his company’s subsidiary Emisar, which Hunterbrook reported to be largely a shell. In response, Hemsley claimed Emisar had “several thousand” employees.

Did Helmsley lie under oath?

Our reporter visited the Dublin office of Emisar and found it empty. Only about a dozen cubicles were in the room, tucked inside the Irish headquarters of UnitedHealth’s PBM Optum.1

Hunterbrook also scraped LinkedIn and company databases and identified only 29 possible Emisar employees.

Hunterbrook repeatedly sent inquiries to the CEOs companies seeking clarity on their PBM GPOs over several months. None of them have answered questions. If United has evidence of the “several thousand” people purportedly employed by Emisar, Hunterbrook wants to see it.

Auchincloss was one of several members of Congress to ask about the Hunterbrook report.

Representative Dr. Mariannette Miller-Meeks, a Republican from Iowa, displayed a flowchart from Hunterbrook’s January 6 article mapping how money moves between entities. She blasted the CEOs for the secretive scheme.

“So a health plan might be getting 100% of the rebates the PBM received, but the plan has no idea if the rebate got everything the GPO earned or how much money the GPO may have kept,” she said, reiterating a key finding from Hunterbrook’s reporting. “It begs the question, are PBM GPOs just to cover for the Big Three to say in a contract that they pass through 100% of the rebates they get and also a way to keep anyone else from being able to fully access brand drugs?”

Rep. Erin Houchin, R-Ind., also cited Hunterbrook by name. She pressed the executives on the astonishingly lucrative entities — whose tiny headcounts make them look like some of the most profitable companies in human history, at least on paper.

“How is it so profitable to have a GPO? How are they profiting $50 million in revenue per employee, on the backs of American patients?”

In response to her questions, the Congresswoman was met with silence. “No answers? Okay.”

Billionaire Mark Cuban, who owns an alternative to PBMs called Cost Plus Drugs, shared a clip of the exchange. “Great question. Easy to answer. No answer given”.

The hearings added to mounting calls for national leaders to address the skyrocketing costs of American healthcare.

“What we heard this week was no surprise. Patients are struggling to afford their medications even when they have insurance,” Merith Basey, CEO of the bipartisan patient advocacy group Patients for Affordable Drugs Now (P4AD Now), told Hunterbrook.

“Any serious effort to lower health care costs must tackle the root causes of high drug prices and savings must actually reach patients, not just move between other actors in the system,” she continued, referencing schemes like PBM GPOs.”

Watch our investigation:

Authors

Laura Wadsten is an investigative journalist specializing in healthcare. She began her career reporting on antitrust and health care markets as a Correspondent for The Capitol Forum, a premium subscription financial publication. Previously, she was Executive Director of the nonprofit Moving to Value Alliance, where she produced the MTVA Unscripted Podcast. Laura was a Hodson Scholar and Editor-in-Chief of The News-Letter at Johns Hopkins University, where she earned a B.A. in Medicine, Science & the Humanities.

Andrew Ford is an investigative journalist who exposed systemic flaws and prompted reforms in healthcare, business, policing and state government. His reporting was published by ProPublica, USA Today, The Arizona Republic, Asbury Park Press, and Florida Today. He holds a journalism bachelor’s from the University of Florida and is in his final semester of a data analytics master’s from Georgia Institute of Technology. He’s based in Phoenix, AZ.

Editor

Sam Koppelman is a New York Times best-selling author who has written books with former United States Attorney General Eric Holder and former United States Acting Solicitor General Neal Katyal. Sam has published in the New York Times, Washington Post, Boston Globe, Time Magazine, and other outlets — and occasionally volunteers on a fire speech for a good cause. He has a BA in Government from Harvard, where he was named a John Harvard Scholar and wrote op-eds like “Shut Down Harvard Football,” which he tells us were great for his social life.

Hunterbrook Media publishes investigative and global reporting — with no ads or paywalls. When articles do not include Material Non-Public Information (MNPI), or “insider info,” they may be provided to our affiliate Hunterbrook Capital, an investment firm which may take financial positions based on our reporting. Subscribe here. Learn more here.

Please email ideas@hntrbrk.com to share ideas, talent@hntrbrk.com for work, or press@hntrbrk.com for media inquiries.

LEGAL DISCLAIMER

© 2025 by Hunterbrook Media LLC. When using this website, you acknowledge and accept that such usage is solely at your own discretion and risk. Hunterbrook Media LLC, along with any associated entities, shall not be held responsible for any direct or indirect damages resulting from the use of information provided in any Hunterbrook publications. It is crucial for you to conduct your own research and seek advice from qualified financial, legal, and tax professionals before making any investment decisions based on information obtained from Hunterbrook Media LLC. The content provided by Hunterbrook Media LLC does not constitute an offer to sell, nor a solicitation of an offer to purchase any securities. Furthermore, no securities shall be offered or sold in any jurisdiction where such activities would be contrary to the local securities laws.

Hunterbrook Media LLC is not a registered investment advisor in the United States or any other jurisdiction. We strive to ensure the accuracy and reliability of the information provided, drawing on sources believed to be trustworthy. Nevertheless, this information is provided “as is” without any guarantee of accuracy, timeliness, completeness, or usefulness for any particular purpose. Hunterbrook Media LLC does not guarantee the results obtained from the use of this information. All information presented are opinions based on our analyses and are subject to change without notice, and there is no commitment from Hunterbrook Media LLC to revise or update any information or opinions contained in any report or publication contained on this website. The above content, including all information and opinions presented, is intended solely for educational and information purposes only. Hunterbrook Media LLC authorizes the redistribution of these materials, in whole or in part, provided that such redistribution is for non-commercial, informational purposes only. Redistribution must include this notice and must not alter the materials. Any commercial use, alteration, or other forms of misuse of these materials are strictly prohibited without the express written approval of Hunterbrook Media LLC. Unauthorized use, alteration, or misuse of these materials may result in legal action to enforce our rights, including but not limited to seeking injunctive relief, damages, and any other remedies available under the law.

By: Till Daldrup, Jacob Hutt

Editor: Sam Koppelman

Based on Hunterbrook Media’s reporting, Hunterbrook Capital is short $SOC at the time of publication. Positions may change at any time. See full disclosures on our website.

Exxon spinoff Sable Offshore faces seven barriers to restart its pipeline, idled since a major oil spill in 2015. One of those approvals needs to come from the California Coastal Commission, which Sable CEO Jim Flores criticized for its “Teflon” “eco-Nazi attitude” in a leaked call recording newly obtained by Hunterbrook. Because of these barriers — and despite Trump Administration intervention — Sable’s project, originally scheduled to go online in Jan 2024, may never sell oil. At least not under the ownership of Sable ($SOC), which is quickly running out of cash.

Hunterbrook Media has reported on Sable since early 2024. These investigations have scrutinized Sable's effort to restart a project responsible for one of the biggest oil spills in California history. The articles have been covered by Bloomberg, Politico, the Wall Street Journal, and several other publications. The findings were also cited in Santa Barbara’s decision to deny the transfer of key permits to Sable from Exxon, the pipeline's prior owner. When Hunterbrook published its first article on Sable, the company had already delayed its launch timeline to Sept 2024. That launch has still not occurred as of Jan 2026. Sable is once again teasing an imminent restart. Here's our reporting on why that remains a long-shot.

This article includes detailed analysis by a recent federal law clerk engaged to clarify the complex array of litigation involving Sable. Nothing here is legal advice. Other sections were led by Hunterbrook’s reporters. Please contact ideas@hntrbrk.com to reach the reporting team.

And check out the version of this investigation on our website, which includes an infographic that adds clarity to a confusing legal landscape.

“It’s much harder to stop us once we’re in production than the status quo being off production,” Sable Offshore CEO Jim Flores told investors in a leaked audio recording newly obtained by Hunterbrook.

“And the big key thing is we don’t need to publicize.”

He added: “We’re keeping it strictly within the investment community.”

Flores was laying out a stealth strategy to restart the pipeline responsible for one of California’s worst oil spills — pump first, deal with regulators later. Once oil was flowing, he reasoned, it’d be harder for the state to shut him down.

“Then, if the judge wants to come in and do something crazy, the status quo is production.”

There were a few problems.

Among them: Flores needed approval from a range of California government authorities, including the California Coastal Commission (CCC) — which Flores criticized on the call for what he called its “eco-Nazi attitude.”

That call took place on April 15, last year. At the time, Flores said: “Our target is June 1 to start putting oil into 324, 325,” referring to Sable’s onshore Las Flores pipelines, formerly known as Plains Lines 901 and 903.

As of this January, the quiet restart Flores had promised investors still hasn’t happened — nine months after that audio recording and two years after the company initially said its project would be online.

In the meantime, with the pipeline still dormant, California has done what Flores seemed to fear: launching a wave of legal and regulatory attacks to keep it that way, putting a labyrinth in Sable’s path.

In other words, for now at least, it looks like the “eco-Nazis” are winning.

In a court appearance earlier this month, Sable’s lawyers confirmed the Exxon spinoff had not filled its pipelines — and conceded the company cannot restart while a preliminary injunction remains in place.

“This is costing us millions of dollars every day,” Sable’s attorney told Judge Donna Geck of the Superior Court of Santa Barbara County, pleading for an accelerated legal timeline.

Sable now pins its hopes on the Trump administration, which only has limited authority over the project.

In December, the federal Pipeline and Hazardous Materials Safety Administration (PHMSA) asserted jurisdiction over one aspect of the pipeline’s approval process: safety. PHMSA approved that aspect of the restart plan despite California’s objections. In court, the company trumpeted this as a breakthrough.

But as evidenced by Sable’s failure to fill its pipeline, federal backing isn’t enough. PHMSA’s authority does not cover regulatory areas beyond safety, such as mitigating coastal effects in California. Those aspects are still delegated to state authorities.

In recent weeks, Hunterbrook has reviewed the relevant laws, precedents, and court proceedings — speaking with experts in state and federal environmental regulations. This reporting has revealed that Sable, whose only asset is this project, faces at least seven independent legal barriers to restart, several of which could keep the pipeline offline indefinitely.

Without additional funding, Sable appears set to run out of cash before it clears them all, if it ever does. The company has an estimated less than $100 million in the bank if spending hasn’t meaningfully slowed in recent months.

“There’s still a lot of legal obstacles in the way and a lot of opportunity for opponents to try to tie things up,” said Deborah A. Sivas, the Luke W. Cole Professor of Environmental Law at Stanford Law School.

Those remaining barriers include:

• A federal consent decree put in place after the 2015 oil spill. The decree explicitly requires approval from California’s Office of the State Fire Marshal (OSFM), not the federal PHMSA, before restart can occur. The requirement remains legally binding and enforceable by the California Attorney General, whose office has previously sued Sable. Earlier this month, the AG office said it was exploring its options to defend state jurisdiction over Sable’s pipeline. Under significantly changed circumstances, a consent decree can be updated by requesting modification from the federal court that entered it. But Sable has yet to make any request for the court to do so — presumably because case law states the executive branch can’t unilaterally create a change that justifies departure from the terms of an agreement to which it bound itself.

• A court injunction requiring Sable to obtain “all necessary approvals and permits” before restarting the pipeline. That includes approval from OSFM, which Sable has conceded it does not have. At the next hearing in the case where the injunction was entered, scheduled for Feb. 27, the court will consider Sable’s argument that federal law preempts state authority to regulate the pipeline — including the injunction — now that PHMSA has claimed jurisdiction. According to our analysis, the active consent decree will make it hard for the judge to side with Sable that it no longer needs to obtain approval from OSFM, as OSFM still has authority under that agreement. The California AG office, who represents parties both to this case and to the consent decree, has stated it is exploring its options to protect OSFM’s authority, indicating that reconsideration of the injunction could be a drawn out process.

• A second court injunction surrounding coastal work. Judge Thomas Anderle of the Superior Court of Santa Barbara is barring Sable from any “development” activities on the pipeline. Excavation. Grading. Pipeline infrastructure work. Placement of materials on the seafloor. The injunction lasts until Sable obtains Coastal Act authorization from the CCC. Without the ability to maintain the pipeline, operating it would mean massive liability risk, including potentially for Exxon. Sable relies on Exxon’s permits in the absence of its own. Sable appealed this injunction, with briefing completed on Jan. 9. A hearing on a separate motion for reconsideration in the trial court is set for Feb. 18.

• A state parks easement that expired in 2016. The easement cannot be renewed until California State Parks completes environmental review under the California Environmental Quality Act, a process with no overall defined timeline. California State Parks informed Sable it would begin an “initial study” on Nov. 13, which appears to be ongoing, the first step in a process that typically takes months at a minimum.