Fermi Inc. (FRMI) Company Overview

Fermi Inc., operating as Fermi America, is a pre-revenue developer of a private-power and artificial-intelligence campus in Carson County, Texas. Its FRMI shares trade on Nasdaq and the London Stock Exchange. The core asset is Project Matador, planned beside Texas Tech University land. Fermi’s official project description combines data-center buildings with dedicated natural-gas, nuclear, solar, wind, and battery resources rather than primary reliance on the public grid.

Project Matador in plain English

The investment case is not based on operating data centers. It depends on converting land, permits, equipment, fuel access, engineering, and tenant discussions into phased capacity faster than grid-connected alternatives. Fermi’s July 2026 investor presentation targeted 1.5 GW by year-end 2027 and 2 GW during 2028, with a long-range mix of 11 GW gas and 6 GW nuclear.

Why this is not a conventional data-center REIT

Fermi plans to become a landlord and power provider, but it remains in development. It had generated no revenue and signed no tenant lease through March 31, 2026. Investors are therefore underwriting entitlement, construction, financing, procurement, and customer-conversion risk before recurring rent or property margins can be measured.

Fermi’s model centers on long-duration leases and power charges. The latest Form 10-K describes ground leases and powered-shell leases. Ground-lease tenants would finance their buildings while paying for land and power access; powered-shell tenants would occupy substantially completed facilities, increasing Fermi’s rent opportunity and capital burden.

Lease economics and tenant capital

The model shifts capital burden between Fermi and the tenant. Ground leases reduce building expenditure but capture less building economics. Powered shells require more capital and execution while supporting higher rent. Tenant prepayments or construction contributions could reduce debt and dilution, but only after commercial commitments are secured.

Which economic milestone comes first?

The first decisive milestone is a bankable anchor lease. The Texas Tech ground lease requires a notice to proceed by December 31, 2026, with Phase 1 conditioned on a tenant lease of at least 200 MW. A signed lease would therefore trigger development, support financing, and validate demand.

Fermi’s brief history is a compressed sequence of land, equipment, permits, financing, and governance events. Each step increased both the project’s credibility and the capital required.

Seven turning points that still shape the model

1. January 2025
   Fermi LLC was formed. The company began with no operating history, making execution credibility a continuing valuation issue.
2. July-August 2025
   The Firebird acquisition added six Siemens SGT-800 turbines and related equipment, while an amended 99-year ground lease secured the core Texas Tech site. These transactions converted the concept into controlled physical assets.
3. June-September 2025
   Fermi submitted portions of its nuclear combined-license application and received Nuclear Regulatory Commission acceptance, establishing nuclear power as a real permitting workstream rather than a presentation-only aspiration.
4. October 2025
   The company completed its initial public offering and dual listing. Net IPO proceeds of $745.6 million funded early construction, procurement, and corporate liquidity.
5. February-March 2026
   Texas approved a 6-GW clean-air permit, Fermi filed for another 5 GW, and the NRC advanced environmental review. Permitting progress strengthened the site’s option value but did not create revenue.
6. April 2026
   Founder Toby Neugebauer left the chief-executive role and the board. An interim Office of the CEO and interim chief financial officer took over during a capital-intensive development phase.
7. July 2026
   Fermi raised $431.25 million of convertible notes and reported continued turbine-delivery progress. The financing extended runway but introduced interest expense and potential dilution.

Fermi advanced faster on asset control, procurement, and permitting than on tenant contracting. That may reassure hyperscalers about deliverability, but it also commits capital before recurring revenue is visible.

Project Matador targets a central AI constraint: enormous reliable loads face utility interconnection delays and generation shortages. Fermi proposes one coordinated package of generation, fuel, land, cooling, and data-center shells. The strategy is valuable only if private power delivers speed and economics unavailable elsewhere.

Site and power control are the proposed moat

The site combines acreage with nearby gas infrastructure. July 2026 materials cited five major pipelines within 20 miles and 220 MMBtu per day of firm gas supply. Fermi also holds a final 6-GW air permit and has filed for 5 GW more. These assets create barriers because rivals cannot instantly replicate permits, equipment slots, fuel access, and land.

Why equipment ownership matters

Long-lead turbines can determine energization dates. Fermi uses multiple platforms and two manufacturers to reduce single-model dependence. At March 31, 2026, purchase commitments were $192.4 million; $276.6 million had been paid on a $324.4 million fixed-price Siemens F-class order. Procurement supports schedule certainty but creates cancellation, storage, integration, and financing exposure.

The quarter ended March 31, 2026 shows rapid physical development without revenue. Fermi’s latest Form 10-Q reported $1.777 billion of assets, up from $1.413 billion at year-end. PP&E rose 53.0% to $1.431 billion, almost entirely construction in progress; no depreciable asset was in service.

The income statement is dominated by development-stage costs

Q1 2026 G&A was $166.2 million, equal to operating loss because revenue was zero. Expensed share-based compensation was $134.0 million, with another $31.1 million capitalized. A $24.8 million debt-extinguishment loss contributed to a $188.7 million net loss, or $0.30 per diluted share. Current margins are therefore not meaningful.

Asset growth is the key quarterly signal

The most useful quarterly signal is the $495.6 million PP&E increase. Construction in progress reached $1.423 billion while land was $7.8 million. Fermi has committed substantial resources, but those assets must still become energized capacity and contracted rent. Future return on invested capital matters more than current EPS.

Fermi had $207.5 million of cash and $35.8 million of restricted cash at March 31, 2026, but management estimated more than $3 billion may be needed for Phase 0 and Phase 1, including about $2 billion over the following twelve months if leasing advances. The filing said existing liquidity was insufficient without financing plans, creating substantial going-concern doubt before those plans.

The financing stack is expensive and evolving

Gross debt was $439.1 million at March 31. MUFG accounted for $396.6 million at a 12.1% effective rate; Keystone carried $39.5 million at 13.3%; and Beal Bank carried $3.0 million at 14.2%. Repayment of the Macquarie facility removed a 48.9% effective-cost instrument, but remaining debt was still expensive for an unleased development.

Cash flow shows the funding gap more clearly than net income

Q1 operating cash use was $7.3 million, compared with $441.2 million of investing outflow and $283.3 million of financing inflow. The July 2026 convertible-note financing added $416.8 million of net proceeds before $34.5 million of capped-call spending. It extended runway without closing the multibillion-dollar funding gap.

Fermi competes across overlapping markets. Data-center landlords offer operating portfolios; powered-land developers pursue similar customers; utilities and power producers sponsor generation; hyperscalers can self-build; and alternative technologies compete for tenant capital. No single public peer captures the whole model.

The relevant comparison set is broader than data-center REITs

What must be proved before the moat is durable?

Permits and equipment create a head start, not a complete moat. Fermi must prove competitive delivered power costs after fuel, maintenance, reserves, and debt service; meet schedules as AI hardware and cooling requirements change; and negotiate returns with a few powerful hyperscalers. Buyer power remains high.

This qualitative scorecard interprets disclosed progress rather than reproducing a company rating. It highlights the gap between advancing physical readiness and unproven contracted demand.

Ownership is concentrated among founder-related holders, executives, and early investors. The latest Form 10-K amendment reported 637.6 million shares outstanding on April 26, 2026. Directors and officers beneficially owned 208.5 million, or 32.7%. Toby Neugebauer’s 21.8% beneficial stake includes overlapping trust and affiliated holdings.

Founder and early-holder concentration matters

Current leadership and board structure increase transition risk

After Toby Neugebauer’s April departure, Anna Bofa and Jacobo Ortiz became co-presidents in an interim Office of the CEO, and Robert Masson became interim CFO. The current board roster changed again when Miles Everson resigned on July 10, 2026. Independent oversight coexists with concentrated ownership, a classified board scheduled to begin declassification in 2029, and founder-affiliated nomination rights.

Fermi offers a large option on AI infrastructure demand, but scale creates severe execution sensitivity. Upside requires long tenant contracts, upfront customer capital, and pricing that compensates development risk. Downside grows if leasing lags while equipment, debt, and construction obligations continue.

The major opportunities and constraints are tightly linked

The filing also identifies material weaknesses in internal control. That matters while Fermi manages construction capitalization, debt, procurement, stock compensation, and public reporting. Remediation is an operating milestone. The investor-relations site provides new filings, lease announcements, financing updates, and leadership changes.

A conventional DCF is highly sensitive because Fermi has no revenue, stabilized margin, or disclosed tenant lease. Modeling should begin with probability-weighted milestones: contracted and energized megawatts, revenue per megawatt, tenant contributions, construction cost, timing, financing cost, residual capital needs, and dilution.

Valuation should separate three phases: probability of a first tenant and Phase 1 delivery; stabilized cash flow from contracted capacity; and later expansion supported by land, generation, permits, fuel, and demand. This avoids valuing the full 17-GW concept as already funded and leased.

Fermi matters because it tests whether AI power bottlenecks can be solved through a privately powered campus. Land, equipment, gas access, permits, and public capital support the story. No revenue, no disclosed tenant lease, multibillion-dollar funding needs, expensive debt, leadership turnover, and control weaknesses constrain it.