A pre-revenue bet on the one thing AI is short on — power — ripping ~50% in five days even as an ousted co-founder fights to retake the board. The vision is enormous. Almost nothing is built or proven.
First gate: where's the trend? Sharp uptrend off the lows — but a volatile, news-driven one, not a clean grind.
Honest trend read: FRMI has ripped ~50% in 5 days (and well off its $4.47 low) back toward ~$9.46, so the near-term momentum is up — but this is a news-driven, whipsaw uptrend (it also fell ~88% from $36.99 to get there), not the steady, order-backed grind of an OSS or ACMR. It clears the "is it moving up right now" bar, but with a giant asterisk: the moves are driven by binary headlines (OpenAI lease chatter, board fight) that can reverse the trend overnight. So it passes the gate technically, but it's the opposite of a low-volatility trend — size and stops matter more here than on any name we've covered.
The stock is going up fast right now — nearly doubled off the bottom — so it technically passes the "only ride uptrends" test. But it's a wild, headline-driven ride (it also crashed ~88% to get here), not a calm steady climb. It can flip direction on a single news item, so it's far riskier to trade than the smoother trends.
Power is the new bottleneck. Fermi's bet is owning it outright — and renting it to AI.
AI data centers need enormous amounts of electricity, and the normal power grid has a years-long waitlist. Fermi wants to build its own giant power plant in Texas and rent both the power and the buildings to AI companies. The catch: none of it is built yet.
A multi-billion IPO on a blueprint. Near-zero revenue today — with an explicit ramp guided into 2027.
| Metric | Value | Read |
|---|---|---|
| Revenue (Q1 2026) | ~$0 (missed ~$600K est) | nothing online yet |
| Q1 net loss | ~$189M (~70% non-cash SBC) | mostly stock comp, not cash burn |
| FY26 revenue guide | ~$42.8M | first subleases/power begin |
| FY27 revenue guide | ~$2.0B | the inflection — if it lands |
| Cash / capex | ~$243M cash · ~$441M capex | spending ahead of revenue |
| Gas secured / LOI | 2.3 GW for 2026 | real equipment behind it |
| Financing | $500M MUFG loan + prior commitments | major bank backing |
| Market cap | ~$6.06B | paying for the blueprint |
| IPO (Oct 2025) | $682M raised at ~$15B val | richly valued from day one |
Fermi raised $682M in its October 2025 IPO at a ~$15B valuation. The stock ran to a $36.99 high, collapsed to a $4.47 low, and has ripped ~50% in five days back to ~$9.46 on OpenAI-lease chatter. It remains pre-revenue: Q1 posted a ~$189M net loss, ~70% of which was non-cash stock-based compensation (so the true cash burn is far smaller than the headline), against ~$243M cash and ~$441M of Project Matador capex. The key change from the original blueprint pitch: management now puts explicit numbers on the ramp — ~$42.8M revenue in FY26 accelerating to ~$2.0B in FY27. That guide reframes the story from "pure pre-revenue moonshot" to "pre-revenue with a steep, near-term, management-stated ramp" — the entire bet is whether that ~$2B FY27 number is real or aspirational.
The company hasn't earned real money yet and lost ~$189M last quarter — but most of that was stock-based pay, not cash going out the door. It has ~$243M in the bank and is spending heavily to build. The big new detail: management now says revenue jumps from ~$43M this year to ~$2 billion next year. Whether that giant leap actually happens is the whole investment.
The catalysts are real — so is the disagreement. Genuine positives sitting right next to genuine red flags.
| Validator | Action | Note |
|---|---|---|
| MUFG | $500M equipment loan (Feb 2026) to accelerate the first 2.3 GW | A top-10 global bank underwriting the build |
| Siemens Energy | Turbine LOI; 2.3 GW gas secured/LOI for 2026 | Real equipment behind the build |
| OpenAI / Nvidia (reported) | OpenAI reportedly in talks for a 10 GW Nvidia-funded Ohio campus | Citizens calls it a positive read-through for Fermi |
| Analysts | Buy avg ~$20.5 (range $6–$35) | The skeptic near ~$11 — wide disagreement |
The setup is genuinely two-sided: a major MUFG financing commitment, 2.3 GW of secured gas, and the OpenAI/Nvidia Ohio data-center read-through sit right next to a skeptical analyst near ~$11 and a boardroom war (next section). The OpenAI signal is indirect — it's about hyperscaler demand for exactly the kind of large, power-ready campuses Fermi is building, not a signed Fermi lease — which is why the stock jumps on the read-through but hasn't re-rated permanently. The single milestone that converts story to substance: the first ~1 GW actually online and a signed anchor lease, not reported talks or analyst inference.
Big positives — a major bank loan, secured gas turbines, and signs that AI giants like OpenAI desperately want power-ready campuses — sit next to big negatives (a boardroom fight, a skeptical analyst). The OpenAI link is hopeful rather than signed: it shows demand for what Fermi builds, not a done deal. Watch for the first power switching on and an actual signed lease.
The risk that's unique to FRMI. An ousted co-founder, a short report, and a control battle that's currently stalling.
Beyond the ordinary execution risk of a pre-revenue build, FRMI carries a governance overhang most names don't — though as of mid-June it has cooled somewhat:
| The issue | What's happening |
|---|---|
| Board-control fight | Ousted co-founder Toby Neugebauer pushed to call a special meeting to retake control / force a sale |
| Current status | As of mid-June, Neugebauer lacks the support to call the special meeting; the board urged shareholders to reject it |
| Buyer interest | Double Eagle said it has no interest in pursuing a Fermi proposal — cooling the "forced sale" path |
| Short-seller report | Fuzzy Panda alleged fraud against prior management (unproven) |
| Litigation | At least one securities-fraud class action filed — unproven but material |
None of these allegations are proven, and the control fight has lost momentum — the dissident currently can't force a meeting and a rumored buyer walked. That reduces the near-term "forced sale" optionality some bulls hoped for, but also removes a major distraction overhang if it fully resolves. Still, stacked on a pre-revenue, leveraged, capital-hungry build, an unresolved governance dispute is the kind of thing that can resurface and derail financing or timelines. It belongs in the position-sizing math — just weighted lower than it was a month ago.
There was a fight over who controls the company, plus a short-seller fraud accusation and a lawsuit. The latest: the challenger doesn't have enough support to force a vote, and a rumored buyer said no thanks — so the drama has quieted. That's calmer than before, but a company that was recently at war with itself while raising billions is still a risk worth watching.
Powering AI, four different ways. Fermi is the most ambitious — and the least de-risked.
| Name | Price | What they are | vs FRMI |
|---|---|---|---|
| FRMI · Fermi | ~$9.46 | Pre-revenue REIT: private power + data-center campus (17 GW planned) | Subject — most ambitious, least de-risked |
| APLD · Applied Digital | $45.90 | Powered data-center developer leasing to AI/hyperscalers | Closest comp — further along, already leasing |
| IREN · IREN Ltd | $59.17 | Ex-miner running AI cloud on owned power | Operating & revenue-generating |
| TLN / VST | $436 / $163 | Established independent power producers selling to data centers | Profitable, proven — the "safe" AI-power play |
The spectrum is clear: Vistra/Talen are the profitable, proven power producers; IREN and Applied Digital are operating, revenue-generating powered-data-center plays already leasing to AI; and Fermi is the pure blueprint — the biggest vision (17 GW, owned generation) but the only one with zero revenue and the most binary outcome. You're trading proven cash flow for maximum optionality. If Matador executes, the upside dwarfs the comps; if it stalls, the comps are still standing and FRMI isn't.
There are safer ways to bet on AI's power hunger — established utilities (Vistra, Talen) and companies already renting out powered data centers (IREN, Applied Digital). Fermi is the all-or-nothing version: the grandest plan but nothing built and nothing earned yet. Bigger potential payoff, much bigger chance of failure.
Two scores. The vision: 8/10 · the stock as a risk-adjusted bet: 4/10.
| Dimension | Grade | Why |
|---|---|---|
| Vision / TAM | A | AI-power is the bottleneck; owned 17 GW generation is a huge prize |
| Near-term momentum | B | +~50% in 5 days, but whipsaw & headline-driven |
| Catalyst optionality | B+ | OpenAI lease + Siemens turbines + financing are real |
| Execution proof | D− | $0 revenue, nothing online, build not started at scale |
| Governance | D− | Board-control war, short report, securities litigation |
| Balance sheet | D+ | Leveraged, negative cash flow, no revenue cushion |
The gap between 8 and 4 is the trade: a genuinely enormous vision wrapped in an almost-entirely-unproven, leveraged, governance-troubled shell. This is a lottery ticket on AI power — the upside is real and large if Matador hits its ~$2B FY27 guide and an anchor tenant signs, but the base rate for pre-revenue, capital-intensive builds fighting an internal board war is unforgiving. It passes the uptrend gate, but it's the highest-variance name we've covered: position-size it like the binary option it is, key it to hard milestones (first GW online, signed anchor lease, board fight resolved), and respect that a single headline moves it 30%.
The dream is a 9; the reality of buying it today is a 4. It's a lottery ticket — potentially huge if everything goes right, but a real chance of going to zero. If you play it, treat it like a small bet you can afford to lose, and watch for the real milestones (power switched on, a signed lease, the board fight settled).
John's read. A real moonshot — trade it as the binary option it is, not as an investment.
- The thesis is the cleanest "AI needs power" pure-play there is. Owning 17 GW of generation behind-the-meter and renting it to AI is a genuinely massive prize, and the bottleneck is real. The vision earns the 8.
- But it's a blueprint, not a business. $0 revenue, leveraged, nothing online — you're paying ~$6B for what management says it will build. That's fine as a small speculative bet, not as a core position.
- The board war is the under-discussed risk. OpenAI headlines get the attention, but an ousted co-founder fighting for control + a short-seller fraud report + litigation, all during a capital-hungry build, is exactly how financing and timelines slip. Don't let the exciting catalyst hide the governance mess.
- On the trend rule: it passes — it's up ~50% in five days — but it's the whipsaw kind, not the steady kind. It fell ~88% from its high to get here and moves double-digits on a single headline. That's a trade-it-with-stops name, not a hold-and-compound one.
- How I'd play it: a small, sized-to-lose lottery ticket on AI power, keyed strictly to hard milestones — first 1.1 GW online, a signed anchor lease, and the board fight resolved. Stay only while the uptrend holds; the moment momentum breaks or financing wobbles, there's no revenue floor to catch it. Huge upside optionality, real zero risk — size accordingly.
Want the milestone checklist that confirms — or kills — the Matador bet?
Join the Discord to find out! →Fermi Inc. IPO & Oct 2025 filings ($682M raised, ~$15B valuation) · Q1 2026 results (revenue ~$0 vs ~$600K est, ~$189M net loss — ~70% non-cash SBC, ~$243M cash, ~$441M capex; FY26 guide ~$42.8M, FY27 ~$2.0B) · MUFG $500M equipment-loan 8-K (Feb 10, 2026, accelerating first 2.3 GW) · Siemens turbine LOI & 2.3 GW gas secured · reported OpenAI/Nvidia 10 GW Ohio data-center talks + Citizens read-through (press reports, not a Fermi lease) · Evercore-type skeptic ~$11; Buy avg ~$20.5 (range $6–$35) · Fuzzy Panda short report & securities-fraud class action (unproven) · Neugebauer special-meeting proxy fight (lacks support as of mid-June); Double Eagle declined to pursue · price/range via Robinhood/StockAnalysis (~$9.46; 52W $4.47–$36.99; ~$6.06B cap). Figures from filings & press reports; not live IBKR pricing.