Fermi — Blitz Index 31 (D)

Contracted but unbuilt — the stock is down 85% and the founder is gone.

Score breakdown

Radar axes (0–100): Moat 38 · Speed 48 · Creative 60.

Watch item

Gates: G1 conditional pass (2 conditions) · G2 pass (estimated, mid confidence). First scored. IPO 2025-10-01 at $21. The gate's conditions explain the whole score.

Strengths

• 222MW TensorWave anchor lease (~$6.5B/15yr potential) — if the terms hold
• Second anchor tenant would validate the model
• Lease-receivable securitization plan (10-Q) — the funding flywheel
• 45V/45Q/48C/45J tax-credit monetization

Concerns

• Anchor lease filed without the agreement text — termination clauses unconfirmed
• First anchor terminated Dec 2025 — binding force itself is being litigated
• Stock $21 → $3.25 (−85%); REIT structure constrains funding
• Construction needs billions Fermi must guarantee

Research notes

1. Current Position

Fermi is **a company at the contract stage of blitz Act 1**. No revenue, no buildings, no deliveries. What exists is 5,236 acres of land (long-term ground lease with Texas Tech, SEC confirmed), a $6.5B contract, and a ~$2.08B market cap (press-reported). As the score file puts it — a "signed blitz": a company that finances by selling blueprints.

2. Blitzscaling Lens

**Rule 4. Launch an Embarrassing Product — it sold the embarrassing blueprint first.** Fermi is an extreme variant of Rule 4. Not an embarrassing "product" but an embarrassing "blueprint" — it sold a 15-year lease before building anything. The same species as CIFR's "pre-sale" pattern and NBIS's "build with customer money" pattern. The difference from NBIS: no customer prepayments yet — what TensorWave paid was a signature, not money. Rule 4's premise is "launch, then patch," but data centers cannot be patched (the CRWV research's "heavy blitz" grammar), and Fermi hasn't even launched. **The risk of the "signed blitz" is that the signature is everything** — the first anchor's departure is the evidence.

3. Competitive Landscape

**Within the power-bottleneck cohort**, its position is "preempted but unproven." Matador's 5,236 acres + up-to-11 GW vision + a 222 MW first anchor — the same game of "land and contracts first" as SEI·TLN·LBRT. The difference is delivery track record: TLN holds cash-generating assets plus a 17-year contract, SEI is under construction, and LBRT has its first delivery scheduled for Q4 2027. FRMI is at 0. **The earliest stage of the same game** — hence 31/D.

4. Inflection Points

**Inflection Point 1. The first delivery phase begins + first rental revenue is recognized — the moment blueprints become buildings.** What to watch: post-Q3 2026 earnings (scheduled 2026-11-10) construction-progress disclosures, the announcement of the first delivery phase's start, and the first rental revenue appearing on the income statement. The moment recognized revenue moves from $0 to $1 is the resolution of G1's conditionality. Which of the 222 MW blocks goes live first, and whether take-or-pay's start timing follows the contract, are the observation points. Link: A1 3 points ("zero proven growth") and B4 2 points ("no measured revenue or margins") both hang on this inflection point. When delivery begins, both move at once — the "signed blitz" becomes the "under-construction blitz."

5. One-Line Risk

If the anchor leaves, the contract is paper — the first anchor already left (−34%), the second anchor's guarantor is a quarter-scale startup against the contract, and the 15-year take-or-pay's binding force hangs on a termination clause whose one line has not been disclosed. Every point of this company stands on that one line.

1. Current Position

Fermi is **a company at the contract stage of blitz Act 1**. No revenue, no buildings, no deliveries. What exists is 5,236 acres of land (long-term ground lease with Texas Tech, SEC confirmed), a $6.5B contract, and a ~$2.08B market cap (press-reported). As the score file puts it — a "signed blitz": a company that finances by selling blueprints.

The contract's substance is confirmed. **The TensorWave lease (signed 2026-08-09, SEC 10-Q confirmed)** — Fermi Campus 1 LLC ↔ TensorWave TEX1, LLC. 222 MW · ~311,496 sqft · 15 years (build-to-suit, modified net lease), with **~$6.5B in expected contract revenue over 15 years** (excluding renewals and expansion), a take-or-pay clause, the tenant's obligations guaranteed by TensorWave Inc., and Fermi guaranteeing the landlord's obligations and completion (completion guaranty). Expansion option: two additional halls for 650 MW (subject to going live by 2028-12-31). Project Matador's first customer lease. This differs from OKLO (LOI level) and JOBY (no quantified backlog) — a signed, quantified, bilaterally guaranteed definitive contract. The basis of G1's conditional PASS.

But two conditions attach to the score standing on the contract. ① Delivery not yet begun — recognized revenue $0, the buildings are under construction. Commercial verification comes later. ② The wound in demand credibility — the 2025-12-12 termination of the first anchor ("First Tenant")'s $150M AICA (−34% in the stock), and the class action *Lupia v. Fermi Inc.* (S.D.N.Y., pending, press-reported) alleging "the IPO materials inflated Matador demand." Guarantor TensorWave Inc. is a startup with a ~$1.55B market cap and ~$493M in cumulative raises (press-reported) — **about a quarter of the $6.5B contract's scale** — so the guaranty's substantive value must be discounted. The lease agreement itself has not been filed as an exhibit, so the termination provisions are unconfirmed.

The credibility wound spread to governance. Founder Toby Neugebauer's voting influence of ~22.47% (DEF 14A confirmed) — yet **his employment was terminated and he was removed from the board on 2026-04-30** (DEF 14A confirmed). Director Everson resigned in 2026-07 over a "books access dispute" (DEF 14A confirmed). A development-stage company's founder ouster plus board turnover is an execution-risk amplifier. D3 −3 · D4 −3, summing to −6, is these two items' score.

The capital structure is "the construction money is borrowed and the incoming money is not yet." Book debt of $520.1M (net, mostly 5% converts of $431.25M) — D/E of ~0.5, mild on the books, but the real leverage sits off the books: building 222 MW requires billions in capex, and Fermi holds the completion guaranty. Cash of $62.5M (6/30, sharply down from $408.5M at end-2025) + $416.8M in net converts proceeds in July. Interest of $21.6M per year vs revenue of $0. The stock's $21 → $3.25 (−85%) is the price of dilution and lost trust.

The GATE verdict: G1 is a conditional PASS — a definitive contract (two conditions: delivery not begun, wounded demand credibility). G2 (10-year ceiling ≥ 10x current revenue) is a PASS (estimate) — with revenue at $0, the pre-revenue exception substitutes contract demand of $6.5B for the bar (the ASTS precedent), against an AI-power-infrastructure TAM in the hundreds of billions (industry estimates).

**Two-sided assessment (per the score file):** - **On the upside:** the TensorWave lease: $6.5B/15 years · 222 MW — an SEC-disclosed definitive contract, the first customer lease · the 650 MW expansion option (subject to 2028-12-31 go-live) — the anchor's additional demand path · Matador's 5,236-acre long-term ground lease + up-to-11 GW vision — preempting scarce resources (land, power) · the Hillcore BOOT structure (2.6 GW) — power infrastructure built with others' capital · proven fundraising ability: the IPO + $431M in converts + equipment financing — capital markets still leave the door open · the take-or-pay + power-cost pass-through structure — a margin-defense design. - **On the downside:** recognized revenue $0, delivery track record 0 — everything is forward · the first anchor's termination precedent + a pending class action — demand credibility under litigation · the guarantor (TensorWave) credit at a quarter of the contract's scale — the guaranty's discounted substantive value · the founder's ouster and board turnover — governance instability · cash $62.5M vs billions in construction obligations — the weight of the completion guaranty · the stock −85% (IPO $21 → $3.25) — dilution sensitivity of future fundraising · the lease agreement not filed — termination provisions unconfirmed.

The directional assessment is **"the moment blueprints become buildings is the real scoring day"** — the score's direction depends on the start of the first delivery phase and the first recognition of rental revenue. **Atlas v2 classification: ignition in progress (pre-ignition) — the deepest credibility discount.** In pre-ignition's decision sequence — (a) contract binding force → (b) revenue-conversion timing → (c) demand credibility — FRMI is recorded as the case showing that when (c) breaks, (a) becomes meaningless.

2. Blitzscaling Lens

**Rule 4. Launch an Embarrassing Product — it sold the embarrassing blueprint first.** Fermi is an extreme variant of Rule 4. Not an embarrassing "product" but an embarrassing "blueprint" — it sold a 15-year lease before building anything. The same species as CIFR's "pre-sale" pattern and NBIS's "build with customer money" pattern. The difference from NBIS: no customer prepayments yet — what TensorWave paid was a signature, not money. Rule 4's premise is "launch, then patch," but data centers cannot be patched (the CRWV research's "heavy blitz" grammar), and Fermi hasn't even launched. **The risk of the "signed blitz" is that the signature is everything** — the first anchor's departure is the evidence.

**Rule 8. Raise Too Much Money — raising works, but gets more expensive each time.** The raising speed is fast: the 2025-10-01 IPO (32.5M shares × $21 = $682.5M) → 2026-07-14 converts of $431.25M (5.00%, 2031 maturity, $34.5M capped call) → ~$785M in equipment financing (press-reported). Dual listing on Nasdaq + LSE. The basis of A2 4 points. But the other side is steep: the stock $21 → $3.25 (−85%). **Raising works but gets more expensive each time** — dilution's price eats the stock, and the lower stock makes the next raise more expensive. Rule 8 says "running out of money mid-race is fatal, so accept dilution" — but for Fermi, trust depletion is arriving before cash depletion. In a structure where an off-book completion guaranty sits atop $62.5M of cash, the terms of the next raise are this company's lifespan.

**Rule 5. Let Fires Burn — what this company sells is not electricity but demand credibility.** Fermi's true big fire is **demand credibility**. This company's core asset is not electricity but the belief "that demand exists" — and exactly that belief is being litigated (*Lupia*, alleging "the IPO materials inflated Matador demand"). The first anchor's termination (−34%) is the first crack in that belief. The small fires are forward options not yet fires: the tax-credit (45V·45Q·48C·45J) application/monetization plans and the nuclear phase's NRC path. From Rule 5's view, the big fire of demand credibility is not one the company can put out — only the first delivery and first rental-revenue recognition can. Until then, D4 −3 stands.

3. Competitive Landscape

**Within the power-bottleneck cohort**, its position is "preempted but unproven." Matador's 5,236 acres + up-to-11 GW vision + a 222 MW first anchor — the same game of "land and contracts first" as SEI·TLN·LBRT. The difference is delivery track record: TLN holds cash-generating assets plus a 17-year contract, SEI is under construction, and LBRT has its first delivery scheduled for Q4 2027. FRMI is at 0. **The earliest stage of the same game** — hence 31/D.

**The comparison with NBIS·CIFR** reveals the pattern. NBIS ("build with customer money") is the proven case of the prepayment blitz — FRMI is the "signature without money" version of that pattern. CIFR ("pre-sale") scores 32/D on the gap between ~$11.4B in contracts and $191M in TTM revenue — FRMI is the more extreme version: $6.5B in contracts vs $0 in revenue. What the three show is the spectrum of the word "contract": NBIS's prepayments → CIFR's definitive contracts → FRMI's conditional signatures.

**The structural competitor is the whole power-bottleneck cohort.** Data-center power demand is real, but it can flow to competitors (SEI·TLN·LBRT et al.) rather than Fermi. Dependence on the single anchor TensorWave is FRMI's own risk — if the anchor leaves, the contract is paper.

4. Inflection Points

**Inflection Point 1. The first delivery phase begins + first rental revenue is recognized — the moment blueprints become buildings.** What to watch: post-Q3 2026 earnings (scheduled 2026-11-10) construction-progress disclosures, the announcement of the first delivery phase's start, and the first rental revenue appearing on the income statement. The moment recognized revenue moves from $0 to $1 is the resolution of G1's conditionality. Which of the 222 MW blocks goes live first, and whether take-or-pay's start timing follows the contract, are the observation points. Link: A1 3 points ("zero proven growth") and B4 2 points ("no measured revenue or margins") both hang on this inflection point. When delivery begins, both move at once — the "signed blitz" becomes the "under-construction blitz."

**Inflection Point 2. The lease agreement's exhibit filing — measuring the termination provisions' binding force.** What to watch: the lease agreement's filing as an SEC exhibit. The termination provisions of take-or-pay, the penalty scale, and the TensorWave guaranty's execution conditions are confirmed for the first time. This is where C2 3 points' ("the contract's binding force itself is the litigated matter") discount is either removed or confirmed. Link: the representative case of v1.1 candidate 2 ("the binding-force grading of anchor contracts"). Every point of this company stands on one termination clause — as the score file puts it.

**Inflection Point 3. The class-action outcome + a second anchor — restoring credibility.** What to watch: the progress and outcome of *Lupia v. Fermi Inc.*, and the exercise of the 650 MW expansion option or a second anchor contract's signing. Only a second anchor heals the first anchor's wound. Also observe the post-ouster management's (current CEO Lee McIntire, press-reported) execution continuity. Link: D3 −3 (founder removal) + D4 −3 (demand-credibility litigation) = −6, nearly half of the −13 in deductions. An adverse litigation outcome reaches the watch criterion (15 points) — the inflection point directly wired to the blitz-failing watch trigger.

**Observation checklist (quarterly):** ① construction progress and whether the first delivery phase begins ② the first recognized rental-revenue amount ③ whether the lease agreement's exhibit is filed ④ the class action's progress ⑤ the 650 MW expansion option's exercise and a second anchor contract ⑥ the cash-burn rate vs the converts-proceeds balance. ①·② are "substance" metrics, ③·④ are "credit" metrics, ⑤·⑥ are "survival" metrics. The longer the substance metrics stay at 0, the heavier the credit metrics weigh — the signed blitz's time is finite.

5. One-Line Risk

If the anchor leaves, the contract is paper — the first anchor already left (−34%), the second anchor's guarantor is a quarter-scale startup against the contract, and the 15-year take-or-pay's binding force hangs on a termination clause whose one line has not been disclosed. Every point of this company stands on that one line.

*BlitzLab v1.0 (frozen). Unconfirmed figures must carry estimate/press-reported labels.* *For education and research purposes only. Not a recommendation to buy, sell, or hold any security.*

Site Summary

Fermi is a company at the contract stage of blitz Act 1, at 31/D — 34 points below the threshold. No revenue, no buildings, no deliveries: 5,236 acres (Texas Tech ground lease), a $6.5B contract, and a ~$2.08B market cap (press-reported). A "signed blitz" — financing by selling blueprints. - The contract's substance is confirmed: the TensorWave lease (signed 2026-08-09, SEC 10-Q confirmed) — 222 MW · ~311,496 sqft · 15 years (build-to-suit, modified net lease), ~$6.5B in expected contract revenue, take-or-pay, guaranty by TensorWave Inc., Fermi's completion guaranty, plus a 650 MW expansion option (2028-12-31 go-live). Unlike OKLO (LOI level) or JOBY (no quantified backlog): a signed, quantified, bilaterally guaranteed contract — the basis of G1's conditional PASS. - Two conditions attach: ① delivery not begun — recognized revenue $0 ② the credibility wound — the first anchor's $150M AICA termination (2025-12-12, −34%) and the *Lupia* class action (S.D.N.Y., pending) alleging inflated Matador demand. The guarantor is ~$1.55B in market cap against a $6.5B contract — about a quarter-scale — and the lease's termination provisions are undisclosed. Recognized revenue is $0, so no XBRL revenue concept exists and there is no FRMI data in financials.json or the site's Financials section. - Governance spread the wound: founder Toby Neugebauer's employment terminated and removed from the board (2026-04-30, ~22.47% voting influence), a director resigning over a "books access dispute." D3 −3 + D4 −3 = −6, nearly half of the −13 in deductions — 2 points below the 15-point blitz-failing watch criterion. - The cohort comparison shows the spectrum of the word "contract": NBIS's prepayments → CIFR's definitive contracts → FRMI's conditional signatures. It is the earliest stage of the same "land and contracts first" game (TLN has cash-generating assets, SEI is under construction, LBRT's first delivery is Q4 2027). - Inflection points to watch quarterly: ① the first delivery phase begins and first rental revenue is recognized — the resolution of G1's conditionality ② the lease agreement's exhibit filing — measuring the termination clause's binding force ③ the class-action outcome and a second anchor — the only path to healing the first anchor's wound. Atlas v2 classifies this as pre-ignition with the deepest credibility discount. - One-line risk: if the anchor leaves, the contract is paper — and every point of this company stands on one undisclosed termination clause.

*Translated from the Korean edition (2026-10-15). For education and research purposes only. Not a recommendation to buy, sell, or hold any security.*

Next earnings

Nov 12, 2026 · Before open · Confirmed

Coverage

As of October 2026 · Model v1.0 (frozen) · Confidence: High (listed, SEC filings) · Not investment advice — a scholar's assessment.

BlitzLab scores businesses against a scholar’s standard — not investment advice. No price targets, no buy/sell recommendations. Figures are from SEC filings (XBRL) as labeled in the source scorecard.