Hunterbrook alleges deliberate misrepresentation: CEO Sridhar stated 'no China supply chain' at least five times since February 2025, on earnings calls, to Semafor, on a podcast, and onstage with the Wall Street Journal, while Bloom's own 3Q25 filing quietly disclosed China supplies '70% of rare earth metals used in electronic and electromechanical components' in its tier 2 and tier 3 suppliers. Bloom's $20 billion marketed backlog is more than 40 times its audited $492.6 million RPO; 74% of 4Q25 revenue came from Brookfield joint ventures Bloom part-owns; and the 5 GW ramp is mathematically constrained by global scandium supply. Governance concerns include four CFOs since 2024, including nearly a year with no permanent CFO during which the Brookfield structure was built.

Hunterbrook constructed an independent scandium supply-demand model from government filings, Bloom patents, peer-reviewed studies, and USGS data showing global supply of ~240 tons versus demand of ~310 tons by 2030. Documentary evidence includes Chinese corporate filings (Chunhui's Shenzhen Stock Exchange disclosure, Three Circle's Hong Kong IPO prospectus, DKK analyst reports), ImportYeti trade data showing Hunan Oriental's direct shipments to Bloom's Delaware plant in 2023–2024, and Bloom's own SEC filings showing evolving China disclosures. Primary evidence includes direct messages with a Hunan Oriental Scandium sales representative confirming 'We are also BE's largest supplier of scandium' and 'Not exported directly,' interviews with New Mexico officials and environmental attorneys, and satellite imagery of CoreWeave and Nebius deployment sites.

A specific datable event is the automatic snapback of China's sweeping '0.1% rule' rare earth export controls on November 10, 2026, unless the Trump-Xi trade truce is extended, which would require Chinese export licenses for any product made anywhere containing Chinese-origin rare earths, directly reaching Bloom's Thailand, Japan, and Korea waypoints. A pending catalyst without a fixed date is the New Mexico Environment Department's decision to hold a public hearing on Oracle's air permit for Project Jupiter (with NMED stating it 'may be pursuing civil or criminal prosecution' over fabricated public comments), which Hunterbrook's interviewees estimate could delay or cancel the project into the early-to-mid 2030s.

Not Present: Thesis-Only

CEO Made False 'No China' Supply Chain Claims

Hunterbrook alleges Bloom CEO KR Sridhar stated at least five times since February 2025, on earnings calls, in a Semafor interview, on an energy podcast, and onstage with the Wall Street Journal on June 10, 2026, that Bloom has 'no China supply chain' and is 'not dependent on China for scandium,' with Sridhar claiming the company made this decision as early as 2004 or 2005 and 'never coupled' itself to China. Bloom's own 3Q25 quarterly filing, added after the April 2025 earnings call, disclosed that China supplies 'multiple components including 70% of rare earth metals used in electronic and electromechanical components that are part of our tier 2 and tier 3 sub-assembly suppliers,' though both that filing and the 10-K simultaneously maintained 'Our supply chain is not dependent on China.' A representative of Hunan Oriental Scandium, which claims over 50% of global market share for fuel-cell-grade scandium oxide, told Hunterbrook directly: 'We are also BE's largest supplier of scandium,' and when asked how material reaches U.S. customers under Beijing's controls, replied: 'Not exported directly.' Hunterbrook identified at least four separate China-linked supply routes: direct shipments from Hunan Oriental to Bloom's Delaware plant (at least four times between August 2023 and May 2024 per ImportYeti trade data); ceramic electrolyte membranes via Three Circle Group's Thai subsidiary Glory Winner (154+ metric tons shipped July 2024–November 2025); scandia-stabilized zirconia powder via Japanese supplier DKK, whose Chinese subsidiary DKK (Shanghai) Materials Trading specializes in zirconium compounds; and Korean electrolyte substrates via Amosense, whose raw material supplier KV Materials is owned by Chinese scandium-products maker Vital Group, which spent 127 billion won ($83 million) purchasing from its Hong Kong parent in 2025.

Global Scandium Supply Cannot Support 5 GW Ramp

Hunterbrook's supply-demand model, built from government filings, Bloom's patents, USGS data, industry reports, and peer-reviewed studies, estimates Bloom alone would need roughly 220 tons of scandium oxide annually to achieve 5 GW production plus field replacement units by 2030, against total projected global supply of only about 240 tons and total global demand of about 310 tons, implying a market deficit exceeding 65 tons. Using two independent methods that converge, a bottom-up atomic calculation from Bloom's patents (which specify ~10 mol% scandium in the electrolyte, yielding ~25 tons of scandium oxide per GW sealed into sellable products, rising to ~37 tons per GW after accounting for yield losses) and a top-down market share approach (Bloom at ~75% of SOFC demand, SOFCs at ~two-thirds of global consumption), Hunterbrook estimates Bloom consumed about 30 tons in 2025 while deploying 0.7–0.8 GW. Every published Wall Street deployment trajectory reviewed by Hunterbrook, including Morgan Stanley's path to 4.7 GW and Jefferies' to 5 GW by 2030, pushes the global scandium market into deficit by 2028 per Hunterbrook's model. Bloom's own July 7, 2026 blog post claimed its supply chain 'can support up to 25 GW per year' without providing supporting calculations; per Hunterbrook's model, 25 GW would require roughly 620 tons of scandium annually, approximately 10 times total global consumption in 2025 per USGS figures. John Mavrogenes, a professor of economic geology at the Australian National University, told Hunterbrook: 'Right now in the Western world, there's basically no scandium processing capability to speak of, or just very, very little. It would be a long road to say we're gonna get serious about scandia.'

Record Revenue Driven by Circular Brookfield Financing, Not End Customers

Hunterbrook alleges that Bloom's recent record-setting quarters are substantially attributable to financing vehicles Bloom itself part-owns rather than genuine end-customer adoption: in 4Q25, $574 million of $778 million in revenue (74%) came from related parties; in 1Q26, $373 million of $751 million came from related parties. For full-year 2025, $892 million of $2.02 billion in revenue (44%) came from related parties, of which $862.1 million came from Brookfield 'Fund JVs,' entities that did not exist until August 2025, yet within roughly five months became the source of approximately 43% of Bloom's annual revenue. Bloom holds only 9.9% in one JV and 15% in another, with Brookfield holding the rest. Bloom's own filing footnotes acknowledge the counterparty 'may be a project-finance affiliate rather than the ultimate end user of the products.' A new balance sheet line item of $62.3 million in 'non-current' contract assets appeared in 2025, revenue recognized where 'billing milestones have not been reached' and invoicing sits more than a year out, with 78% related-party. The Brookfield partnership's initial $5 billion announcement specifically promised a European AI factory site 'that will be announced before the end of the year'; it never was, and Brookfield's June 30, 2026 $25 billion expansion announcement made no mention of the European site. The designated anchor tenant, Radiant, is Brookfield's own captive cloud company formed in February 2026, with no publicly named CEO and no disclosed customers, but is an 'NVIDIA Cloud Partner' backed by Nvidia, which also anchors the Brookfield AI Infrastructure Fund. Deloitte flagged the JV accounting as a critical audit matter.

Flagship Projects Jupiter and AEP Face Years of Delays

Oracle's Project Jupiter in New Mexico, intended to house up to 2.45 GW of Bloom fuel cells for an OpenAI data center campus, has no approved air permit for fuel cell emissions and no gas pipeline, with Oracle telling regulators 'Without a reliable source of natural gas, the power plant that will generate electricity for the data center campus cannot function.' Research firm SemiAnalysis already pushed its base case for first power from 2027 to 2029 due to pipeline regulatory delays. The New Mexico Environment Department decided Oracle must face a public hearing after canvassers allegedly submitted fabricated supportive comments using residents' identities without consent; NMED spokesperson Drew Goretzka called it 'unprecedented' and said NMED 'may be pursuing civil or criminal prosecution.' NMELC attorney Kacey Hovden told Hunterbrook Oracle's application is 'incredibly incomplete' and 'wildly underestimated' emissions, estimating first power is now likely 'early-to-mid 2030s' or the project may not happen at all. AEP's $2.65 billion fuel cell deal, representing nearly 40% of Bloom's backlog per Jefferies analysts, slipped from 'no later than 12/31/2028' in AEP's February 2025 investor deck to 'no later than 2030' in its May 2025 deck, a potential delay of up to two years in under 90 days. The data center campus the AEP deal would power reportedly lost its developer Crusoe in June 2026, with the prospective anchor customer, allegedly Google, which has reportedly signed nothing, raising concerns about costs and timelines; if the hyperscaler walks away, it covers AEP's costs incurred plus roughly 10%, per AEP's disclosed terms. Both delays followed Bloom raising guidance on its last earnings call.

Bloom's Pattern: Financiers, Not Customers, Underwrite the Numbers

Hunterbrook traces a recurring structural pattern across Bloom's history in which a financial intermediary, not an end customer, guarantees the volumes that make Bloom's numbers. In Delaware, state lawmakers in 2011 authorized a surcharge on roughly 300,000 Delmarva Power ratepayers legally obligated to purchase Bloom's output through 2033; initially pitched as ~67 cents per month, the fee rose to $5–$6, and in a single year ending May 2019 ratepayers paid $34 million for power Delmarva resold for only $9 million, a $25 million annual subsidy, while Bloom delivered 600 fewer jobs than promised and returned over $1.5 million under the clawback provision. SK ecoplant committed to purchase 'a minimum of 500 megawatts' through 2024 (estimated $4.5 billion), propped up 2022–2023 results, then quietly received three extra years to comply when Korean demand stalled; a December 2023 re-announcement of a '500 MW sales agreement' included by its own arithmetic roughly half undelivered volume from the original 2021 commitment. SK ecoplant ceased to be a related party on July 10, 2025 and sold the last of its stake on April 13, 2026; Brookfield's financing framework was formed within weeks of SK's exit. Bloom has had four CFOs since the start of 2024, including nearly a full year with no permanent CFO, which coincided with the construction of the Brookfield structure during which related-party revenue reached 74% of a quarter, and five since 2020; the most recent hire, Simon Edwards, arrived April 13, 2026, after serving as Groq's CFO for three months and then CEO for three months per Bloom's own 8-K. The CEO has never changed in 25 years.