Why Fermi Stock Collapsed 41% Last Month

Shares collapsed 41% in March and are more than 80% off their highs for newly public Fermi (FRMI), which raised $750M in its IPO but currently generates zero revenue and burned about $600M in free cash flow in its first year. Management targets 17 GW via Project Matador using gas, renewables and prospective nuclear, but an initial tenant withdrew a $150M upfront commitment, there is no near-term path to profitability, and upcoming IPO lock-up expirations increase downside risk.

Analysis

Translating a bespoke power platform for hyperscale compute into reality requires capital density that is frequently underappreciated: at typical combined‑cycle or integrated gas/renewables build costs of ~$0.8–1.2M per MW, each gigawatt of delivered capacity implies roughly $0.8–1.2B of up‑front capex before transmission and land costs, so a multi‑GW program sits in the low‑double‑digit billions. That math forces a binary financing dynamic — meaningful non‑dilutive project financing, long‑dated offtakes, or strategic JV equity — and absent any one of those the equity will act like a finance call option that rapidly dilutes.

Key Decisions for Investors

• Short FRMI equity (or buy 6–12 month put spreads if available): target 40–70% downside if no secured long‑term anchor offtake or project financing is announced within 6 months; use a 25–30% stop to limit single‑name risk.
• Pair trade: short FRMI / long an established data‑center REIT (e.g., DLR or EQIX) for 6–24 months — rationale is tenant migration toward providers with immediate, contracted capacity. Size as 1:1 notional; take profits if FRMI prints a financing package that covers >50% of forecast capex.
• Long NVDA 9–15 month call spread (e.g., buy A, sell B with strikes that create ~2–3x asymmetric payoff): NVDA benefits if bespoke power delays push customers to optimize compute per watt and accelerate chip refresh cycles. Hedge with a 20% allocation cap given market beta.
• Event play: buy out‑of‑the‑money FRMI puts 3–6 months into a lock‑up/financing window or ahead of major permitting decisions; reward is high convexity vs limited premium paid, while maximum loss is the premium.