Fermi Inc.
Business Overview: Fermi Inc. (NASDAQ: FRMI)
Executive Summary
Fermi Inc. is a Texas-based, development-stage company formed in January 2025 to build one of the largest privately-powered data center campuses in the United States — Project Matador, a planned 5,236-acre (expanding toward ~7,570-acre) site in Carson County, in the Texas Panhandle, leased on a 99-year ground lease from the Texas Tech University System. The company's premise is that AI and high-performance computing demand is now constrained by power availability, not chip supply, and that owning grid-independent generation — gas, and eventually nuclear — next to the data center itself is the way to win hyperscaler tenants.
Fermi matters not because of current revenue (it has none) but because of the sheer scale of what it has already lined up before signing a single binding tenant lease: a path to as much as 11 GW of on-site power capacity, roughly 15 million square feet of planned AI-ready compute space, and hundreds of millions of dollars in committed equipment financing, positioned to IPO'd onto Nasdaq as a REIT-structured power-and-shell landlord rather than an operator.
1. Core Business Model & How They Work
Fermi intends to lease "powered shell" space to AI/compute tenants, with the power itself bundled into the lease rather than sold separately to the grid.
[ Secure land (Texas Tech ground lease) ] ➡️ [ Build on-site power (gas → nuclear) ] ➡️
[ Construct powered-shell data center buildings ] ➡️ [ Lease to AI/HPC tenants, rent tied to reserved power (per kW/month) ] ➡️
[ Collect rent + pass-through reimbursements ] ➡️ [ Operate as a REIT ]Key Operational Drivers
- Power-first development: Rather than waiting on a utility interconnection queue (often a multi-year bottleneck for hyperscale data centers), Fermi is building its own phased generation stack — grid power, gas turbines, eventually nuclear — to guarantee tenants power on a predictable timeline.
- Phased capacity buildout: Phase 1 targets up to 200 MW of grid power plus ~1.5 GW of owned/leased gas turbines (initial tenant energization expected 1H 2027); Phase 2 adds ~2.5 GW of combined-cycle gas by end-2028; Phases 3–4 layer in Westinghouse AP1000 nuclear units toward ~11 GW total.
- REIT structure, triple-net economics: Rent is planned around reserved power capacity (per-kW-per-month) plus a capital-based component and pass-through reimbursements — closer to an industrial landlord's economics than a data-center operator's.
- Heavy reliance on external financing and counterparties before any revenue: The company has stacked equipment and warehouse financing facilities (MUFG, Keystone National Group, CSG Investments/Beal Bank, Yorkville) to fund construction ahead of signed, binding leases.
2. Business Segments
Fermi is a single-project, pre-revenue development company at this stage; there is no meaningful multi-segment breakdown to report, so this section is omitted per the guide.
3. Product Portfolio
| Offering | Category | Purpose | Why It Matters |
|---|---|---|---|
| Powered-shell data center space | Leasable real estate | Core product: physical building + guaranteed on-site power, leased to AI/HPC tenants | The entire business model in one line — power-guaranteed shell space is the product |
| Phase 1 gas generation (~1.7 GW incl. grid) | On-site power | Bridges tenants to power well before utility grid capacity would otherwise be available | Enables first tenant energization as early as 1H 2027, years ahead of typical grid-dependent projects |
| Phase 3–4 AP1000 nuclear (up to ~6 GW) | On-site power | Long-duration, carbon-free baseload power for later-phase tenants | Differentiates Fermi from pure gas-powered competitors on both scale and emissions profile, pending NRC license approval |
| Ground leases to self-building tenants | Alternative leasing structure | Option for tenants to build their own facilities on leased land | Gives Fermi a lower-capex path to monetize excess acreage |
4. Competitive Landscape
Grid-Independent Power
│
Fermi (on-site gas──→┼←── Other private-power/behind-the-meter
then nuclear, REIT) │ data center developers
│
Hyperscaler-built─────┼── Traditional colocation operators
campuses, utility PPAs│ (grid-dependent, interconnection queues)
Grid-Dependent Power- Traditional colocation/data center operators: Depend on utility interconnection queues, which can take years in power-constrained regions — Fermi's pitch is explicitly built around skipping that bottleneck.
- Cloud providers' own data centers: Hyperscalers increasingly build or co-invest in their own power-adjacent campuses, which could reduce their need to lease from a third party like Fermi.
- Other private/on-site generation developers: A growing category of competitors pursuing the same "power is the bottleneck" thesis, some with more capital or faster permitting progress.
- Renewable, storage, and nuclear developers: Compete for the same long-duration, carbon-aware power story that Fermi's later nuclear phases are targeting.
- Fermi explicitly states that many competitors are larger and better capitalized, an unusually candid admission for a pre-revenue company this early in its buildout.
5. Strategic Strengths & Risks
Strengths
- Scale and permitting progress ahead of peers: A TCEQ air permit for roughly 6 GW (issued February 2026) and an NRC combined license application under review for four AP1000 reactors put Fermi meaningfully further along the regulatory path than most "paper" data-center-power proposals.
- Long-term land control: A 99-year ground lease with Texas Tech University System on a multi-thousand-acre site removes land-assembly risk, historically one of the hardest parts of large-scale data center development.
- Diversified, committed financing already in place: Over $900 million in equipment and warehouse financing facilities (MUFG, Keystone, CSG/Beal Bank, Yorkville) secured before any tenant revenue shows real capital-markets access despite the pre-revenue stage.
Risks
- No signed, binding tenant leases: As of the filing date, Fermi had no binding leases; its only disclosed letter of intent's related advance agreement was terminated in December 2025 — the entire revenue thesis still depends on converting "active talks" into contracts.
- Execution and timeline risk across every phase: Each phase (gas turbines, combined-cycle conversion, nuclear) depends on equipment delivery, permitting, and construction timelines that have historically slipped industry-wide, and nuclear approval in particular (NRC licensing) is a multi-year, non-guaranteed process.
- Capital intensity and financing dependence: The company has no operating revenue and is reliant on continued access to debt/equipment financing until tenants are signed and paying rent; a capital-markets shock could stall the buildout.
- Larger, better-capitalized competitors: The company itself flags that many rivals have more resources, a notable risk given how capital-intensive power-and-compute infrastructure is.
6. Financial Overview
| Metric | Value | Strategic Context |
|---|---|---|
| Operating revenue | $0 | A true pre-revenue development-stage company; thesis rests entirely on future leasing |
| Planned capacity | ~11 GW (up to ~17 GW) | Scale alone would place Fermi among the largest privately-powered campuses proposed in the U.S. |
| Committed financing | $900M+ across four facilities (MUFG, Keystone, CSG/Beal Bank, Yorkville) | Shows real capital access, but all debt taken on before any tenant revenue exists |
| TCEQ air permit | ~6 GW approved (Feb 2026); 5 GW more filed (Mar 2026) | Concrete regulatory progress that differentiates Fermi from earlier-stage peers |
| Initial tenant energization target | 1H 2027 | The key near-term milestone that will validate (or undercut) the entire investment case |
7. Summary Conclusion
Fermi's business model is a bet that power, not land or chips, is the binding constraint on the AI buildout, and that a developer willing to build its own gas-and-eventually-nuclear generation stack next to a data center can leapfrog the multi-year utility interconnection queues that slow conventional data-center development. The company has made unusually concrete progress for its age — land under long-term lease, air permits in hand, an NRC license application under review, and nine-figure financing facilities secured — but none of it yet translates into a signed, paying tenant, which remains the single event markets will watch for. The central risk is sequencing and execution: Fermi must deliver power on schedule (Phase 1 energization by 1H 2027) and convert its hyperscaler conversations into binding leases before its financing runway or the broader AI-infrastructure capital cycle turns, all while competing against better-capitalized incumbents chasing the same power-constrained opportunity.