Stock Analysis · Bulls vs Bears

Fermi Inc. FRMI

🐂 Bulls vs Bears 🐻Report #19 Oct 2026
★ Willem owns this stock · 1.5% of the portfolioOlder version — see the latest →
77
very high risk
🐂 7vs8 🐻
Bears lead
35%
▲ Chance of the biggest tailwind the forecasts miss
60%
▼ Chance of the biggest headwind the forecasts miss
The verdict

Fermi owns a rare pile of AI-power ingredients (land, permit, turbines, gas, water) but has not yet turned any of it into a contract that pays, and its balance sheet cannot carry the build without a creditworthy tenant and project debt. Owning it today is a bet that the next ~3 months bring three things: the TensorWave guarantor, project financing and the Texas Tech notice-to-proceed. The evidence so far leans…

The Bulls vs Bears podcast is being made
Two AI hosts debate this report, NotebookLM-style. It appears here automatically within a few hours.

Written by Claude (Anthropic’s AI) from public sources · Willem owns FRMI · AI can make mistakes — numbers in brackets point to the sources · Not financial advice

🐂 Bull arguments 7

  1. 1Scarce, already-bought "speed-to-power" assetsStrong
  2. 2A first binding anchor lease with a pass-through structureMedium
  3. 3Management reset to buildersMedium
  4. 4Capital-light growth via HillcoreWeak
  5. 5A strong demand backdropMedium
  6. 6Cheap West Texas gas, with costs passed throughWeak
  7. 7Optionality: nuclear, land and asset valueWeak

🐻 Bear arguments 8

  1. 1The anchor lease may not close, and the tenant is weak creditHigh
  2. 2A large funding gap, expensive debt and likely dilutionHigh
  3. 3A record of missed promises, and nothing energisedHigh
  4. 4A governance war and legal overhangMedium
  5. 5Concentration and hard deadlinesHigh
  6. 6Rivals are ahead with better tenantsMedium
  7. 7Exposure to the neocloud and AI-capex cycleMedium
  8. 8The business model is still in fluxMedium

Bull arguments show how well they survive the devil’s advocate (strong / medium / weak); bear arguments show the damage they could do after the counter-argument (high / medium / low).

Bull case — all positive points (most important first)

🐂 Bull #1Scarce, already-bought "speed-to-power" assets

  • The point: in AI infrastructure, the binding constraint is power that can be delivered within ~2 years. Fermi already holds most of the slow-to-get pieces.
  • Evidence:
    • About 1.5 GW of turbines landed, on site or leased [3,4].
    • A ~6 GW final air permit (2026-02-25) [41,2].
    • Firm gas of 220,000 MMBtu/d [4], water rights, and a 99-year site [1].
    • New-build gas turbines are scarce. GE Vernova's gas backlog plus slot reservations rose to 116 GW in Q2 2026 [26,27]. Siemens Energy reported a firm gas-turbine backlog of ~69 GW and lead times of 3+ years [28].
  • Why it matters: a tenant choosing Fermi could in principle get power in 2027–28, while a new site could take 2030+. That is what supports rent and makes the equipment saleable even if the project stalls.
  • 😈 Devil's advocate:
    • "Landed" is not "installed". EPC contracts for both turbine types are still unsigned [4], the leased TM2500s are not delivered until mid-2027 [1], and nothing has been energised [1].
    • Rivals with similar assets are already delivering, e.g. Galaxy's Helios Phase I (133 MW to CoreWeave, July 2026) [30].
    • Most of the turbine fleet is pledged to lenders. MUFG may start marketing the F-class collateral if there is no 400 MW lease by 2026-11-10 [1].
  • Verdict: Strong as an asset fact, Medium as a business advantage. The assets are real, but the time advantage erodes every quarter they sit idle.

🐂 Bull #2A first binding anchor lease with a pass-through structure

  • The point: TensorWave proves a tenant will sign at Matador. The contract design keeps commodity risk off Fermi.
  • Evidence:
    • 222 MW, 15 years, ~$6.5B contracted revenue, two 5-year renewals, and an expansion option to 650 MW [1,16].
    • The variable power charge equals actual electricity cost [1]. Management says this template will be reused [6].
    • The stock rose ~21% on the announcement (2026-08-10) [48].
  • Why it matters: at full term, $433M a year of rent would be large relative to a ~$2.3B market cap (641M shares at ~$3.5–3.8) [12,36,38]. Expansion to 650 MW would roughly triple it.
  • 😈 Devil's advocate:
    • The lease is not effective.
    • Closing needs a guaranty, project financing, board approvals, work letters and a service-level agreement. Either side may walk away if the conditions fail [1].
    • The guarantor is still unnamed after two months and one extension [17].
    • The founder alleges the originally announced guarantee "was rescinded within hours". This is an adversary's claim and is uncorroborated [13].
    • Fermi has given a completion guaranty and owes escalating rent credits if late. The tenant can terminate past an outside date and recover its prepayments [1].
    • Only 222 MW is committed: below the 400 MW the MUFG covenant wants, and only ~2% of the 11 GW vision.
  • Verdict: Medium Signing is real progress, but until 2026-10-31 it is an option, not a contract that pays.

🐂 Bull #3Management reset to builders

  • The point: the promoter-founder era has given way to an engineering-and-construction CEO and a professional board. That should lower execution and governance risk.
  • Evidence:
    • McIntire's background is Bechtel, CH2M Hill and TerraPower [1].
    • New CFO, COO, Chief Commercial Officer and General Counsel by July 2026 [1].
    • Stifel suggested that friction with Neugebauer may have slowed tenant talks [19].
    • Q2 cash burn fell ~50% quarter-on-quarter (CFO) [6].
    • Q2 net loss was $25.8M, versus $188.7M in Q1, which had $134M of share-based pay [3,10].
  • Why it matters: tenants and lenders sign with people they trust. A construction-credible CEO is a precondition for project debt.
  • 😈 Devil's advocate:
    • The team is 2–6 months old with no delivery record at Fermi.
    • The founder owns ~22.5% and is litigating. A for-cause firing plus federal and SEC inquiries keep governance noisy [1,12].
    • The new GC brings related-party ties (MAD Energy, Davillier) [1].
    • The 70% supermajority bylaw and class-action waiver adopted mid-fight look defensive [13].
  • Verdict: Medium It is the right profile, but unproven, and the overhang has not gone away.

🐂 Bull #4Capital-light growth via Hillcore

  • The point: Hillcore would pay for, own and run up to ~2.6 GW, so Fermi can sell more megawatts without raising the capital for the power plant.
  • Evidence: a build-own-operate-transfer framework (2026-08-11). Fermi is anchor offtaker for ≥50% of campus power, with its guarantee capped at 12 months of capacity charges and a fair-value purchase option after ~10 years [1,4]. First ~350–360 MW comes within ~24 months of notice to proceed [4].
  • Why it matters: it shrinks Fermi's own power capex. The CFO's benchmark is $3–4M per MW [6]; at that rate 2.6 GW would cost roughly $8–10B. Fermi would focus its capital on buildings that earn rent.
  • 😈 Devil's advocate:
    • The deal is non-binding [1].
    • Hillcore's own financing capacity and turbine supply were not found.
    • A third-party plant needs paying tenants first, so it does not solve the tenant problem.
    • A take-or-pay offtake puts Fermi in the middle: obligated to Hillcore, but dependent on tenants.
  • Verdict: Weak to Medium until definitive documents are signed (due ~2026-11-09).

🐂 Bull #5A strong demand backdrop

  • The point: demand for AI power far exceeds supply that can be delivered quickly, so any site that can deliver power should eventually fill.
  • Evidence:
    • 2026 capex at the four largest hyperscalers is ~$700B+ in secondary roundups (company filings not read) [49].
    • Fermi's 10-K cites AI power demand rising from 55 GW (2023) to up to 219 GW (2030) (McKinsey, company's selection) [2].
    • Grid queues are crowded: ~474 GW of ERCOT large-load requests sparked a governor-ordered audit in Aug 2026 [32].
  • Why it matters: this would let Fermi sign the dozen prospects it claims and use the 650 MW option.
  • 😈 Devil's advocate: the market has plenty of competing sites. Peers signed investment-grade or CoreWeave leases first (Galaxy, Applied Digital, Crusoe/Abilene) [29,30]. Fermi's December 2025 investment-grade tenant walked away even in this "hot" market [2]. Demand is real; demand for this site is unproven.
  • Verdict: Medium It is a necessary tailwind, but it has not been enough so far.

🐂 Bull #6Cheap West Texas gas, with costs passed through

  • The point: Permian gas is among the cheapest in the world, so Matador's power should be cheap for tenants. Pass-through also keeps price risk off Fermi.
  • Evidence:
    • Waha (the West Texas gas hub) traded negative on a record ~99 days in 2026 [35].
    • The EIA's October outlook has Henry Hub at $3.48 (2026) and $3.16 (2027) [35].
    • Leases pass actual energy cost through to the tenant [1].
  • Why it matters: a lower all-in cost per MWh is a selling point against grid-tied sites.
  • 😈 Devil's advocate: new Permian pipelines (Blackcomb, Hugh Brinson) will narrow the Waha discount [35]. Amarillo is ~300 miles north of Waha, and the gas contract price was not found.
  • Verdict: Weak to Medium. It is a modest selling point, not a moat.

🐂 Bull #7Optionality: nuclear, land and asset value

  • The point:
    • An 8,400-acre permitted site with a COLA under NRC review could be worth more than the market credits, especially to a strategic buyer.
    • The founder argues the market values ~10 GW of rights near zero [13].
  • Evidence:
    • Fermi is the first applicant in the NRC's applicant-prepared EIS pilot. The final EIS is targeted for Dec 2026 [24,25].
    • DOE's $17.5B conditional AP1000 loan programme (June 2026) [34].
    • PP&E was $1.55B at 2026-06-30 [1].
  • Why it matters: it gives some floor value or takeout logic if the standalone plan stalls.
  • 😈 Devil's advocate:
    • Part 3 of the COLA was withdrawn, and the licensing decision is "TBD" [24]. Nuclear power is a 2033+ event at best [1].
    • Whether Fermi is in the DOE programme was not found.
    • Enterprise value (~$2.3B market cap + ~$950M debt incl. the convertibles − ~$500M cash, rough pro forma) is already above book PP&E [1,11]. Interpretation: this is not a deep discount to invested capital.
  • Verdict: Weak It is real but distant, and it does not cover near-term funding.

Bear case — all negative points (most important first)

🐻 Bear #1The anchor lease may not close, and the tenant is weak credit

  • The point: everything depends on TensorWave becoming effective with an outside guarantor. If it fails, Fermi has no revenue, no Phase-1 project debt, and runs straight into the TTU and lender deadlines.
  • Evidence:
    • Closing conditions are a guaranty, project-level financing, board approvals, work letters and an SLA. Either party may terminate if they are not met [1].
    • Management said on 2026-08-13 that the backstop would be named "in the coming days" (summary only) [6]. On 2026-09-22 the deadline moved to 2026-10-31, with "our remaining work is to finalize the lease guaranty" [17,21].
    • TensorWave was valued at $1.55B in June 2026 [18].
  • Why it matters:
    • Failure would leave the TTU 200 MW lease condition unmet [1].
    • It would trigger 10% quarterly amortisation and collateral marketing on the ~$445M MUFG turbine loan [1].
    • It would trigger mandatory prepayment on the $120M Keystone loan if there is no approved customer agreement by 2026-12-31 [1].
    • It would be a second public tenant failure within a year.
  • 🛡️ Counter-argument: TensorWave is backed by AMD Ventures and Magnetar [18], so a guarantee from a deep-pocketed AI player is plausible. A one-month extension is common in complex leases. The guaranty may already be largely agreed.
  • Severity: High It is binary and near-term (three weeks), and it is the gateway to every other milestone.

🐻 Bear #2A large funding gap, expensive debt and likely dilution

  • The point: Fermi needs billions before rent starts in late 2027 and has hundreds of millions.
  • Evidence:
    • Phase 0+1 capex "could exceed $3B", ~$2B of it in the next 12 months. All phases are ~$70–90B [1].
    • Cash plus restricted cash was $91.7M at 2026-06-30. The convertibles added ~$416.8M net [1].
    • Debt was $520.1M at 2026-06-30, before the $431M of 5% convertibles. The Beal facility costs 12% plus a $37M exit fee [1].
    • The 10-Q: resources are "not sufficient" against forecast disbursements, with "substantial doubt" about going concern before management's plans [1]. Also: "If the TensorWave Lease does not commence… our liquidity will be materially constrained" [1].
    • Share-based pay was $118.7M in H1 2026, with $134.7M not yet recognised [1].
    • The convertible conversion price is ~$9.52, far above today's ~$3.5–3.8 [1,38].
  • Why it matters: even if everything goes right, project lenders will want sizeable equity. At ~$3.5 a share, raising $500M–1B would mean issuing ~20–40% more shares (interpretation, illustrative). If things go wrong, it is a rescue.
  • 🛡️ Counter-argument:
    • Management will defer capex until tenants and financing are in place [1].
    • Hillcore shifts power capex off balance sheet [4].
    • The convertibles have no maintenance covenants [11].
    • A closed, guaranteed lease is financeable at project level, which caps the equity need.
  • Severity: High Even the bull path needs outside money at a depressed price.

🐻 Bear #3A record of missed promises, and nothing energised

  • The point: the company has repeatedly over-promised timing.
  • Evidence:
    • Power targets kept shrinking: 1.1 GW online by end-2026 (IPO) was abandoned by 2026-03-30 [2]. The 2.0 GW by end-2027 target (10-K) became 640 MW nameplate by Q4 2027 [6].
    • Equipment slipped: TM2500 delivery moved to Q3 2027 [1], and SGT-800 deliveries moved into 2028 [4]. The first nuclear unit moved to 2033 [1].
    • Grid power is unconfirmed: 86 MW from SPS was due in H2 2026 and is not confirmed energised [2,42].
    • Contracts are unsigned: EPC contracts are not final [4], and there is no data-center contractor yet [5].
    • Statements were contradicted: "Getting the tenant is the easy part" [7], and a 12-month liquidity assurance in March was followed by going-concern language in August [1,7].
  • Why it matters: TensorWave's lease has hard delivery dates (halls on 2028-01-01 and 02-01), with rent credits for lateness and termination rights past an outside date [1]. Further slips now cost money.
  • 🛡️ Counter-argument: most misses happened under the old team. The new CEO is a construction professional, and the 640 MW plan is far more modest and matched to one tenant [6]. Physical progress is real: the turbines arrived, the power-island excavation is done, and the gas line is laid [4,5].
  • Severity: High The pattern is long and the new team has no track record yet.

🐻 Bear #4A governance war and legal overhang

  • The point: a 22.5% founder at war with the board, plus federal inquiries, can scare off tenants, lenders and guarantors.
  • Evidence [1,12,13,47]:
    • The fight:
      • Neugebauer was terminated for cause.
      • His consent solicitation ran May–Jun 2026, then was suspended.
      • Lawsuits are pending, with an amended complaint against three directors filed 2026-09-18. His temporary restraining order was denied on 2026-09-16, and a motion to dismiss is set for 2026-11-19.
      • His shareholder proposal asks for an independent strategic review.
      • The board adopted a 70% supermajority bylaw.
    • Legal and regulatory:
      • An EDNY subpoena (2026-07-30) and an SEC request (2026-08-03).
      • The class action Lupia v. Fermi (S.D.N.Y., filed 2026-01-05) over tenant-demand disclosures [40].
      • Other suits: Firebird (motion to dismiss denied 2026-08-11), Bayonne (≥$8.5M), and Peña/Summers (>$415M claimed over reclaimed founder units).
    • A material weakness in internal control persists.
  • Why it matters: a guarantor doing diligence will price this. If the federal inquiries escalate, financing could freeze.
  • 🛡️ Counter-argument: none of the inquiries alleges wrongdoing, and the scope is "former members of management" [1]. The founder's campaign failed to force a meeting, and courts have so far sided with the company [12]. Most claims look small relative to the asset base, apart from Peña/Summers, which the company says is not probable.
  • Severity: Medium High. It is mainly a drag on financing and dealmaking, with a tail risk if the inquiries lead to charges.

🐻 Bear #5Concentration and hard deadlines

  • The point: a few counterparties and dates control the project's fate.
  • Evidence:
    • One tenant (100% of contracted revenue) [1].
    • One landlord (TTU): termination right if no notice to proceed by 2026-12-31, plus $9M escrow and $2M prepaid rent due [1].
    • One utility agreement (SPS) underpins the "power as incident of tenancy" model. The 10-K flags that power delivery could be limited if it is challenged [2].
    • One main turbine lender (MUFG): the 2026-11-10 trigger [1].
    • TM2500 lease rent runs to 2045 whether or not the units are needed [1].
  • Why it matters: several of these fall in the same ~10-week window (Oct 31, Nov 10, Dec 31). One failure can cascade into the others.
  • 🛡️ Counter-argument: TTU, the lenders and the lessor are all better off renegotiating than terminating a project with $1.5B sunk. The 10-Q says management expects to "defer, scale, or renegotiate" obligations [1].
  • Severity: High in the next 3 months, lower if TensorWave closes.

🐻 Bear #6Rivals are ahead with better tenants

  • The point: Fermi's competitive window is closing as other power-first campuses deliver to investment-grade or proven tenants.
  • Evidence:
    • Galaxy Helios delivered 133 MW to CoreWeave in July 2026 under a 15-year lease; ~800 MW is approved and contracted [30].
    • Crusoe/Abilene (Stargate) runs behind-the-meter gas and is targeting 1.2 GW by end-2026 [29].
    • Applied Digital has CoreWeave and investment-grade hyperscaler leases (tracker data, not independently confirmed).
    • Pacifico GW Ranch is pursuing 5 GW off-grid in Pecos County (unconfirmed permit) [29].
  • Why it matters: hyperscalers sign with the developer that can deliver soonest with the least counterparty risk. Fermi's governance and balance sheet make it the riskier landlord.
  • 🛡️ Counter-argument: demand is large enough for many sites [2]. Fermi's permitted 6–11 GW scale is larger than most rivals. Being in SPP sidesteps ERCOT's new large-load rules [31,32].
  • Severity: Medium

🐻 Bear #7Exposure to the neocloud and AI-capex cycle

  • The point: the first tenant is a GPU neocloud. These businesses are highly leveraged and depend on continued AI demand and chip pricing.
  • Evidence: TensorWave has raised ~$495M in total [18]. Nvidia introduced a backstop model for partners in July 2026 (secondary). Industry commentary describes a "neocloud refinancing wall" in 2026–28 (secondary, low quality).
  • Why it matters: a 15-year lease is only as good as the tenant and guarantor across an AI cycle.
  • 🛡️ Counter-argument: an investment-grade guarantor, if obtained, transfers most of this risk. AMD has a strategic reason to keep a large AMD-only cloud alive.
  • Severity: Medium It depends on who the guarantor is.

🐻 Bear #8The business model is still in flux

  • The point: core economics have not been disclosed and keep shifting: REIT vs C-corp, shell vs turnkey, own power vs Hillcore.
  • Evidence:
    • The REIT election was deferred to at least after 2026 [1].
    • Management declined to give margins [6].
    • The first deal is turnkey. At $10–12M/MW for the data center plus $3–4M/MW for power (CFO benchmarks) [6], 222 MW implies ~$3B or more of capital, consistent with the 10-Q's ">$3B" [1].
  • Why it matters: turnkey builds put the full construction-cost risk on Fermi. Returns on that capital are unknown.
  • 🛡️ Counter-argument: a fixed return on invested capital is built into the rent stack [2], which partly protects returns against cost overruns, depending on how the lease is written (not found).
  • Severity: Medium

Tailwinds NOT in the forecasts

  • TensorWave closes with a named investment-grade guarantor and Phase-1 project debt — 35% chance by mid-2027 · Impact if it happens: High

    • What it is: a guarantor such as AMD or a hyperscaler signs a 15-year rent guaranty. That would make the lease bankable and unlock project financing, the TTU notice to proceed and lender relief.
    • Why it's not priced in: the stock fell from ~$7.60 (2026-08-12) to ~$3.5 (2026-10-08) even with the lease signed [36,37,38]. The market is treating the lease as likely to fail or be heavily diluted.
    • What would trigger it / the sign to watch: an 8-K naming the guarantor and confirming the lease is effective by 2026-10-31. Then a project-debt term sheet.
    • Why this probability: the guarantor is still missing two months on, and one extension has already been used. Closing and project debt together need several parties to agree, so the odds are below even, though the parties clearly want a deal.

  • TensorWave expansion or a second tenant (≥400 MW in total) — 25% chance within 36 months · Impact: High

    • What it is: TensorWave exercises its options toward 650 MW, or a hyperscaler signs separately [1,6].
    • Why it's not priced in: consensus treats even the first 222 MW as uncertain. No second tenant is named.
    • What would trigger it / the sign to watch: a named second lease, or option exercise before the 2028-12-31 energisation deadline.
    • Why this probability: management claims a dozen prospects, but none has closed in a year. Any expansion largely depends on the first lease working.

  • Strategic sale or takeover of the site or company — 18% chance within 36 months · Impact: High

    • What it is: a hyperscaler, utility, infrastructure fund or energy company buys Fermi or the Matador assets. The founder's proposal explicitly seeks a strategic review [13].
    • Why it's not priced in: the price reflects standalone distress, not a takeout.
    • What would trigger it / the sign to watch: the board hiring a bank, the founder's proposal gaining votes on 2026-10-30, or approaches reported in the press.
    • Why this probability: a permitted multi-GW site with landed turbines is a rare asset. But litigation, a 22.5% hostile holder and federal inquiries complicate any deal.

  • Hillcore becomes definitive and funded — 35% chance within 12 months · Impact: Medium

    • What it is: binding BOOT agreements and Hillcore financing for the first ~360 MW block [4].
    • Why it's not priced in: it is a non-binding framework with little public detail.
    • What would trigger it / the sign to watch: definitive documents around 2026-11-09, or a Hillcore turbine order.
    • Why this probability: the take-or-pay requires tenant demand Fermi does not yet have, so it likely moves only with TensorWave.

Headwinds NOT in the forecasts

  • Highly dilutive equity raise, rescue financing or restructuring — 60% chance within 36 months · Impact: High

    • What it is: issuing stock or preferred at depressed prices, or restructuring debt, to bridge to rent in late 2027.
    • Why it's not priced in: the average analyst target is $17–21 [36], which implies little further dilution. The share count (641M) does not include future project equity.
    • What would trigger it / the sign to watch: an ATM programme filing, a PIPE, the MUFG amortisation step-up after 2026-11-10, or Q3 cash below ~$300M.
    • Why this probability: ~$2B of 12-month capex against ~$0.5B of resources makes outside equity likely even in success, and certain in failure. Some deferral could reduce the size.

  • TensorWave lease collapses (terminated or not closed by a second deadline) — 35% chance within 6 months · Impact: High

    • What it is: a second tenant failure, triggering the TTU, MUFG and Keystone deadlines.
    • Why it's not priced in: consensus models still show revenue from 2027 [36]. The market price partly discounts this, but the analyst models do not.
    • What would trigger it / the sign to watch: another extension without a named guarantor, or an 8-K reporting termination.
    • Why this probability: the parties have used one extension and the guarantor is still unnamed. But both sides have invested heavily and want a deal, so failure is less likely than closing or re-extension.

  • Federal inquiry escalates (charges, Wells notice or enforcement involving the company or former management) — 15% chance within 36 months · Impact: Medium-High

    • What it is: the EDNY subpoena or SEC request leads to formal action [1,22].
    • Why it's not priced in: the risk is disclosed but not quantified, and no wrongdoing is alleged.
    • What would trigger it / the sign to watch: new 10-Q legal disclosures, indictments of former executives, or SEC Wells notices.
    • Why this probability: a criminal subpoena plus a for-cause firing is a meaningful signal, but most inquiries end without company-level charges.

  • TTU lease renegotiated on worse terms, or terminated — 20% chance of a material adverse renegotiation or termination by 2027 · Impact: High

    • What it is: if no notice to proceed by 2026-12-31, TTU "may terminate" [1].
    • Why it's not priced in: the deadline is little discussed publicly.
    • What would trigger it / the sign to watch: an 8-K amending the ground lease, or an unfunded $9M escrow at year-end.
    • Why this probability: TTU benefits financially from the project and is more likely to extend than terminate, but it may extract better terms.

Investment Risk Score: 77/100

Risk area Weight Score (0–100) Why
Business & competition 25% 80 Pre-revenue and pre-power. One conditional lease. Rivals are delivering to stronger tenants first. The edge (permits and landed turbines) is real but fades with time.
Financial health & dilution 20% 85 ~$2B of 12-month capex against ~$0.5B of resources. Going-concern language. 12% equipment debt with tenant-linked triggers. $119M of H1 share-based pay. Equity raise likely.
Execution (delivering the plan) 20% 85 1.1 GW by 2026 abandoned. Nothing energised. EPC contracts unsigned. A new team with no track record faces hard 2028 delivery dates and rent-credit penalties.
Dependency & concentration 15% 85 One tenant, one ground landlord with a 2026-12-31 termination right, one utility agreement, and one main turbine lender with a 2026-11-10 trigger.
Regulation, policy & geopolitics 10% 55 SPP location avoids ERCOT's new rules. But it relies on a Texas landlord-power exception, faces federal inquiries and a growing Texas backlash over data-center water and power, and nuclear licensing is "TBD".
Sentiment & expectations 10% 45 The stock is −83% from the $21 IPO and at an all-time low ($3.42 on 2026-10-08), so the market expects little. But analyst targets ($17–21) still assume success, and short interest is ~10%.

Overall = 77/100 (very high). The score reflects a company whose survival depends on a few binary events in the next ~12 weeks: lease closing, guarantor, TTU notice to proceed and lender triggers. It has no revenue cushion and needs heavy outside financing even in the good case. This is the first deep dive, so there is no change to report.

Conclusion

  • What kind of company this is: a single-site, pre-revenue development company, effectively a land-and-power option on AI demand. Owning it is a bet that a governance-scarred developer can convert a pile of scarce permits and turbines into bankable, investment-grade-backed leases before its cash and deadlines run out.
  • The 2–3 things that will decide the outcome:
    1. Whether TensorWave closes with a credible guarantor (by 2026-10-31).
    2. Whether Phase-1 project debt is raised without crushing dilution (by ~2026-12-31, the TTU deadline).
    3. Whether the new team actually energises ~210 MW by mid-2027 and hits the January–February 2028 hall deliveries.
  • Signposts confirming the bull case:
    • an 8-K naming an investment-grade guarantor and declaring the lease effective;
    • a project-finance term sheet or closing, and the TTU notice to proceed issued;
    • signed EPC contracts and first SPS grid power energised;
    • a second named tenant, or TensorWave expansion.
  • Signposts breaking it:
    • a second extension without a named guarantor, or lease termination;
    • MUFG collateral marketing or a covenant waiver/amendment after 2026-11-10;
    • a large equity raise below ~$3, or an unfunded TTU escrow at year-end;
    • escalation of the EDNY/SEC inquiries, or a successful founder-led board challenge.
  • Balanced verdict:
    • Strongest bull case: Fermi has spent ~$1.5B assembling what most AI developers cannot buy quickly: a giant air permit, landed turbines, gas, water and a 99-year site. A credible operator now runs it, and a 15-year, $6.5B lease is signed. If a guarantor appears, the project becomes financeable and the market cap looks small against contracted rent.
    • Strongest bear case: the lease is not real until it closes. The tenant is a startup, the guarantor is a ghost, and cash covers months, not the ~$2B of near-term capex. The company's history is one of missed targets, and its founder is suing it while federal prosecutors ask for documents.
    • Where the evidence leans: toward the bear side on near-term financing and execution, with a genuine but uncertain asset-value floor. The next three weeks are unusually decisive.

Open questions & data gaps

  • The guarantor's identity and terms: whether it covers the full 15 years, and whether it is investment grade. Not found. This is the most important unknown.
  • Any grid power energised: whether SPS has energised any of its 86 MW. Not confirmed by any primary source.
  • Lease unit economics: base rent vs power charge split, the construction-cost overrun protection in the "fixed return on invested capital", and the outside date. Not found.
  • TensorWave's own finances: revenue, debt and other lease commitments. Not found.
  • Earnings-call sources: a verbatim company Q2 call transcript was not read. Q&A came from Motley Fool's transcript and third-party summaries. The Q1 call was available only as a summary.
  • The first tenant: its identity and why it left in Dec 2025. Not found. This matters for the class action and for whether demand is site-specific.
  • Consensus revenue and EPS: aggregators conflict, so they were treated as not reliably found.
  • Hillcore: its balance sheet and turbine access. Not found.
  • DOE AP1000 loans: whether Fermi is among the seven AP1000 partners with DOE letters of intent. Not found.
  • Competitor figures: Applied Digital and Pacifico details come from trackers only and are not independently confirmed.
Sources
  1. Fermi Inc. Form 10-Q for the quarter ended 2026-06-30 — SEC EDGAR — 2026-08-14 — https://www.sec.gov/Archives/edgar/data/2071778/000207177826000051/frmi-20260630.htm
  2. Fermi Inc. Form 10-K for FY2025 — SEC EDGAR — 2026-03-30 — https://www.sec.gov/Archives/edgar/data/2071778/000207177826000010/frmi-20251231.htm
  3. Fermi Q2 2026 earnings release (8-K Ex. 99.1) — SEC EDGAR — 2026-08-13 — https://www.sec.gov/Archives/edgar/data/2071778/000207177826000046/fermiq22026erex_991.htm
  4. Fermi Q2 2026 investor presentation (8-K Ex. 99.2) — SEC EDGAR — 2026-08-13 — https://www.sec.gov/Archives/edgar/data/2071778/000207177826000046/fermi2q2026investorprese.htm
  5. Fermi Q2 2026 earnings call prepared remarks (PDF) — Fermi IR — 2026-08-13 — https://irp.cdn-website.com/4ee5afac/files/uploaded/Final+2Q26+Prepared+Remarks+8-13-26.pdf
  6. Fermi (FRMI) Q2 2026 earnings call transcript — Motley Fool — 2026-08-20 (call 2026-08-13) — https://www.fool.com/earnings/call-transcripts/2026/08/20/fermi-frmi-q2-2026-earnings-call-transcript/
  7. Fermi (FRMI) Q4 2025 earnings call transcript — Motley Fool — call 2026-03-30 — https://www.fool.com/earnings/call-transcripts/2026/05/14/fermi-frmi-q4-2025-earnings-call-transcript/
  8. Earnings call transcript: Fermi America Q1 2026 (summary only) — Investing.com — 2026-05-14 — https://www.investing.com/news/transcripts/earnings-call-transcript-fermi-america-q1-2026-reveals-189m-loss-93CH-4701692
  9. Fermi FY2025 shareholder letter (8-K Ex. 99.1) — SEC EDGAR — 2026-03-30 — https://www.sec.gov/Archives/edgar/data/2071778/000207177826000007/frmiex991fy2025shletter.htm
  10. Fermi Q1 2026 earnings release (8-K Ex. 99.1) — SEC EDGAR — 2026-05-14 — https://www.sec.gov/Archives/edgar/data/2071778/000207177826000027/fermi1q26erex_991.htm
  11. Convertible notes offering memorandum excerpts (8-K Ex. 99.1) — SEC EDGAR — 2026-07-09 — https://www.sec.gov/Archives/edgar/data/2071778/000121390026076788/ea029753501ex99-1.htm
  12. Fermi Inc. DEF 14A proxy statement (annual meeting 2026-10-30) — SEC EDGAR — 2026-10-02 — https://www.sec.gov/Archives/edgar/data/2071778/000121390026106378/ea0307100-02.htm
  13. PX14A6G exempt solicitation (Neugebauer parties) — SEC EDGAR — 2026-09-11 — https://www.sec.gov/Archives/edgar/data/0002071778/000121390026099255/ea0305276-px14a6g_toby.htm
  14. Schedule 13D/A (Caddis Holdings / Griffin Perry) — SEC EDGAR — 2026-07-08 — https://www.sec.gov/Archives/edgar/data/2071778/000212347026000004/primary_doc.xml
  15. Form 4, J. R. Perry (founder option delivery sale, 863,637 sh at $7.31) — SEC EDGAR — 2026-07-02 — https://www.sec.gov/Archives/edgar/data/1633632/000163363226000001/primary_doc.xml
  16. Fermi Announces Binding Lease Agreement with TensorWave — Fermi IR — 2026-08-10 — https://investor.fermiamerica.com/fermi-announces-binding-lease-agreement-with-tensorwave
  17. Fermi and TensorWave Agree to Extend Closing Date to October 31, 2026 — ACCESS Newswire (via StockTitan) — 2026-09-22 — https://www.stocktitan.net/news/FRMI/fermi-and-tensor-wave-agree-to-extend-closing-date-to-october-31-psrwibc17htu.html
  18. TensorWave raises $350M Series B at $1.55B valuation — Pulse 2 / Finsmes — 2026-06-10 — https://pulse2.com/tensorwave-raises-350-million-series-b-at-1-55-billion-valuation-to-expand-amd-powered-ai-infrastructure/
  19. Financially struggling AI power startup Fermi loses CEO — Fortune — 2026-04-20 — https://www.fortune.com/2026/04/20/financially-struggling-ai-power-startup-fermi-loses-ceo-reset/
  20. Fermi removes ex-CEO Neugebauer from board due to contract violations — AJ Bell — 2026-05-01 — https://www.ajbell.co.uk/news/articles/fermi-removes-ex-ceo-neugebauer-board-due-contract-violations
  21. Exclusive: Backstop for Fermi America's 222MW deal with TensorWave still being finalised (lede only, paywalled) — The Tech Capital — 2026-08-11 — https://thetechcapital.com/exclusive-backstop-for-fermi-americas-222mw-deal-with-tensorwave-still-being-finalised/
  22. FRMI plunges 4.3% amid federal subpoena — parameter.io — 2026-08-17 — https://parameter.io/fermi-frmi-stock-plunges-4-3-amid-federal-subpoena-over-texas-data-center-development/
  23. Fermi shares plunge after major tenant cancels $150 million funding deal — Alphaspread — 2025-12-12 — https://www.alphaspread.com/market-news/stock-movements/fermi-shares-plunge-after-major-tenant-cancels-150-million-funding-deal
  24. Fermi America COLA, Units 1–4 project page (summary only) — US NRC — updated 2026-09-23 — https://www.nrc.gov/facilities-safety/new-reactors/large-light-water-reactors/combined-license-applications-for-new-reactors/fermi-america-president-donald-j-trump-advanced-energy-and-intelligence-campus-units-1-4
  25. Notice of intent to prepare an EIS, Fermi America COLA — Federal Register — 2026-03-20 — https://www.govinfo.gov/content/pkg/FR-2026-03-20/html/2026-05487.htm
  26. GE Vernova Q2 2026 press release (summary only) — SEC EDGAR — 2026-07-22 — https://www.sec.gov/Archives/edgar/data/0001996810/000199681026000147/gevpressrelease2q26.htm
  27. GE Vernova gas turbine backlog climbs to 116 GW — Utility Dive — 2026-07 — https://www.utilitydive.com/news/ge-vernova-gas-turbine-backlog-climbs-to-116-gw/826039/
  28. Siemens Energy posts record backlog — Turbomachinery Magazine — 2026-08-06 — https://www.turbomachinerymag.com/view/siemens-energy-posts-record-backlog-joining-ge-vernova-and-baker-hughes-in-a-record-quarter-for-gas-turbines
  29. OpenAI Stargate: where the US sites stand — Epoch AI — 2026-04 — https://epoch.ai/publications/openai-stargate-where-the-us-sites-stand
  30. Galaxy completes Phase I of Helios data center campus — Galaxy release via StockTitan — 2026-07-07 — https://www.stocktitan.net/news/GLXY/galaxy-completes-phase-i-of-its-helios-data-center-campus-delivering-6867b3mwe17c.html
  31. SB 6 implementation: shaping the data center future in Texas — Perkins Coie — 2025-12-23 — https://www.ashurstperkinscoie.com/en/insights/sb-6-implementation-shaping-data-center-future-in-texas/
  32. What Governor Abbott's data center audit directive means for ERCOT — Gibson Dunn — 2026-08 — https://www.gibsondunn.com/what-governor-abbotts-data-center-audit-directive-means-for-ercot-and-the-batch-zero-study-process/
  33. FERC approves SPP HILL/HILLGA — Lexology/LexBlog — 2026-01-21 — https://www.lexblog.com/2026/01/21/ferc-continues-shaping-the-future-of-data-center-use-of-the-transmission-grid-with-approval-of-spps-hill-and-hillga/
  34. DOE announces $17.5B in conditional loans for AP1000 builds — American Nuclear Society — 2026-06-23 — https://www.ans.org/news/2026-06-23/article-8142/doe-announces-175b-in-conditional-loans-for-ap1000-builds/
  35. Short-Term Energy Outlook, October 2026 (and Waha price coverage via Reuters feed) — US EIA — 2026-10 — https://www.eia.gov/outlooks/steo/archives/oct26.pdf
  36. FRMI analyst targets, short interest and price alerts — MarketBeat — 2026-07-28 and 2026-10-09 — https://www.marketbeat.com/instant-alerts/price-fermi-nasdaq-frmi-stock-price-down-53-heres-why-2026-10-09/
  37. Fermi Inc. stock hits 52-week low at $4.14 — Investing.com — 2026-10 — https://www.investing.com/news/company-news/fermi-inc-stock-hits-52week-low-at-414-usd-93CH-4927604
  38. FRMI quote pages (all-time low $3.42 on 2026-10-08; $3.97 close 2026-10-07) — TradingView / Morningstar — 2026-10-07/08 — https://www.tradingview.com/symbols/NASDAQ-FRMI/ ; https://www.morningstar.com/stocks/xnas/frmi/quote
  39. UBS cautious on Fermi outlook — Yahoo Finance — 2026-04 — https://finance.yahoo.com/markets/stocks/articles/ubs-cautious-fermi-frmi-outlook-214431530.html
  40. Lupia v. Fermi class action notice (allegations only) — Berger Montague via Newsfile — 2026-02/03 — https://www.newsfilecorp.com/release/284449/Fermi-Inc.-FRMI-Deadline-Approaching-Berger-Montague-Advises-Investors-of-Deadline-in-Securities-Fraud-Lawsuit
  41. TCEQ approves air permits for Fermi America's proposed data center power plant — ABC7 Amarillo — 2026-02 — https://www.abc7amarillo.com/news/local/tceq-approves-air-permits-for-fermi-americas-proposed-data-center-power-plant-texas-commission-on-environmental-quality-panhandle-first-coalition-kendra-seawright-catarina-gonzalez
  42. Behind-the-meter data centers: Fermi America profile (tracker, unsourced) — Cleanview — accessed 2026-10-09 — https://cleanview.co/reports/behind-the-meter-data-centers/full-report/fermi-america
  43. Fermi America executes Electric Service Agreement with SPS (up to 200 MW) — press release via Barchart — 2025-12 — https://www.barchart.com/story/news/36484173/fermi-america-executes-electric-service-agreement-with-southwestern-public-service-company-a-subsidiary-of-xcel-energy-securing-up-to-200-mw-of-power
  44. Fermi TensorWave $6.5bn lease lacks a guarantor (commentary) — FinanceFeeds — 2026-09 — https://financefeeds.com/fermi-tensorwave-lease-guaranty-october-31/
  45. Stock market today May 14: Fermi surges after "Fermi 2.0" — Yahoo Finance — 2026-05-14 — https://finance.yahoo.com/markets/stocks/articles/stock-market-today-may-14-221321891.html
  46. Fermi Q2 2026 slides: $6.5B AI deal, shares fall — Investing.com — 2026-08-13 — https://www.investing.com/news/company-news/fermi-america-q2-2026-slides-65b-ai-deal-shares-fall-86-93CH-4858454
  47. Toby Neugebauer sues three Fermi directors (his own release) — PR Newswire via Finviz — 2026-09-18 — https://finviz.com/news/393863/toby-neugebauer-sues-three-fermi-directors-alleging-a-coordinated-effort-to-take-control-of-company-and-disenfranchise-fermi-shareholders-according-to-the-complaint
  48. Fermi stock surges 21% after TensorWave lease — TIKR — 2026-08 — https://www.tikr.com/blog/fermi-stock-surges-tensorwave-data-center-lease-agreement
  49. Hyperscalers hit $700 billion in 2026 AI spending plans — Yahoo Finance — 2026 — https://finance.yahoo.com/sectors/technology/articles/hyperscalers-hit-700-billion-2026-111243744.html
All FRMI Bulls vs Bears reports
Disclaimer. AI-generated research and podcast (Claude by Anthropic, voices by Google Gemini) for educational purposes — not financial advice and not a recommendation to buy or sell. Willem (MR.INVEST) owns this stock. Check the company’s own filings before making any decision.
← All Bulls vs Bears reports