TL;DR · 30-second read
The Short Version
Fermi, a Texas company building a giant power-and-computing campus for artificial intelligence near Amarillo, has lost the executive in charge of building its nuclear plants. Its reactor schedule has also slipped.
The surprise is that its biggest new customer, an unnamed large technology company putting about $200 million down, is not waiting on those reactors. The power Fermi says it can deliver soon comes from gas-fired generators and the local utility.
So the computers arrive first, running mostly on gas. The nuclear plants, pitched as the long-term clean power source, move further into the future.
Hart Energy reported that Fermi Inc.’s top nuclear builder has departed and that the company’s reactor timeline has slipped. The news landed alongside Fermi’s third-quarter results, released November 12, 2026. In those results, Fermi (Nasdaq/LSE: FRMI, operating as Fermi America) announced agreements with an undisclosed investment-grade hyperscaler. Hyperscalers are the large cloud companies that build data centers by the hundreds of megawatts. The customer committed about $200 million to advance an initial 600 MW of power capacity at Project Matador beginning in 2027, and a further $90 million steel purchase is contemplated.
The same release disclosed that Fermi and TensorWave had terminated the data center lease they signed in August 2026, along with the associated AMD credit support agreement. Fermi also filed its quarterly report on Form 10-Q on November 12. A separate 8-K filed November 4 reported that neither director nominee at its October 30 annual meeting received the share vote its bylaws require.
Executive Summary
Fermi is developing Project Matador, an approximately 8,400-acre private-grid campus in the Texas Panhandle intended to power AI data centers. Its pitch has paired near-term power with a long-term ambition of roughly 17 GW. A senior nuclear departure and a slipping reactor schedule would ordinarily raise the question of whether that pitch still holds.
Fermi’s own Q3 disclosures point to a more specific answer. The commercial traction it reported is a 600 MW hyperscaler commitment starting in 2027. That commitment is anchored by what CEO Lee McIntire called “permitted power”: a 6 GW air permit for on-site generation, firm natural gas supply, gas turbines on order, and 200 MW from utility Xcel Energy. Nuclear appears in the company’s investor presentation mainly as a licensing risk. The reactors, in other words, are the long-dated layer of the plan, and the near-term megawatts are gas and grid.
That matters beyond one company. AI campus plans increasingly advertise nuclear megawatts. Fermi’s situation suggests that, in practice, those megawatts depend on a small number of specialized people and a federal licensing process. Meanwhile, the bridge generation that serves early customers can end up serving them for considerably longer than the brochure implies.
The 600 MW That Doesn’t Wait for a Reactor
Fermi’s headline commercial win this quarter is an investment-grade hyperscaler committing about $200 million to advance an initial 600 MW at Project Matador, beginning in 2027. McIntire made the logic explicit: “Hyperscalers want sites that are ready with permitted power, and Fermi’s Project Matador is one of few that can meet that demand today.” The customer, the release says, is making commitments “ahead of finalizing a lease,” to secure power “within its targeted delivery timeline.”
The “permitted power” in question is a 6 GW air permit issued by the Texas Commission on Environmental Quality (TCEQ). That is the kind of approval combustion plants such as gas turbines require, and nuclear plants do not. The investor presentation filed with the 8-K itemizes Fermi’s 2.2 GW of current capacity as including 200 MW from Xcel Energy and six SGT-800 gas turbines on order for 2028 delivery. It also lists four turbine platforms from two manufacturers, 220,000 MMBtu per day of firm gas supply, and five major pipelines within 20 miles. Nuclear shows up in the same deck among the risk factors, alongside approvals from the Nuclear Regulatory Commission (NRC), the federal body that licenses reactors.
Fermi has not specified which generation will serve the 600 MW. But every source of near-term power it itemizes is gas or grid. That is why a reactor slip does not directly threaten the 2027 tranche. What it changes is what that customer is likely to be running on, and for how long.
Nuclear Schedules Are Built on a Short List of People
Nuclear projects differ from gas plants in one respect that matters here. Licensing, supply-chain qualification and construction sequencing are handled by a small pool of people with direct reactor experience. When one of them leaves, the schedule can move with them. That is the risk the Hart Energy report puts on the table.
Fermi has leaned on its leadership as a differentiator. The company describes itself as combining technology with “a deep bench of proven world-class multi-disciplinary leaders with a combined 25 GW of experience.” A claim about bench depth is tested precisely when a senior builder departs. The material question is whether the remaining team has comparable nuclear-specific delivery experience, not whether it has experience in aggregate.
Fermi’s own disclaimers already push the nuclear layer into conditional territory. The presentation says any capacity beyond roughly 6 GW, up to 11 GW or 17 GW, depends on further permits, financing, interconnection capacity and land, “and there can be no assurance that such capacity will be developed.” A leadership gap on the nuclear side lengthens the path to that conditional tier. It does not change the near-term gas-and-grid core.
Bridge Generation Starts to Look Like the Base Plan
“Bridge” power is the interim supply, usually gas turbines or utility grid, that serves a site until a permanent source arrives. If reactors slip, more of Matador’s life is spent on the bridge. That concentrates exposure on risks Fermi itself lists: natural gas availability and pricing, water, equipment procurement, and approval of the additional 5 GW air permit application it has filed with TCEQ. The deck notes lead times of 2 to 6 years for gas turbines and other long-lead equipment, so turbine deliveries become critical-path items rather than stopgaps.
Several groups are affected. Hyperscale customers that weigh lower-carbon supply over a data center’s lifespan get a longer gas-heavy period than a nuclear-forward pitch implies. Turbine makers and gas suppliers gain a more durable role. Investors valuing the 17 GW figure should treat the nuclear share as an option with a lengthening expiry, not as scheduled capacity.
The broader point for anyone evaluating AI campus proposals is practical. Ask which generation serves which megawatts in which year. At Fermi, the honest answer for the first tranche appears to be gas and grid.
Pre-Lease, Pre-Revenue, and a Contested Board Vote
McIntire described the company as “pre-revenue, pre-lease,” and the quarter reinforced that. The TensorWave lease signed in August 2026 has been terminated, along with the AMD credit support agreement tied to it. The hyperscaler arrangement precedes a definitive long-term lease, and the $90 million steel and building materials agreement is still being finalized. McIntire said converting the prospect pipeline “would completely fill our 2.2 gigawatts of current capacity.” That is a conditional statement, not a contracted one.
Leadership continuity also extends to the board. Fermi’s November 4 8-K reports that at the October 30 annual meeting, nominees Governor Perry and Mr. Stein each won a majority of votes cast but not a majority of shares outstanding, as Fermi’s bylaws require. Both continue as holdover directors with full authority. Fermi says entities affiliated with Toby Neugebauer were obligated under a director nomination agreement to vote for Governor Perry and believes they voted against him. Those investors, through counsel, dispute that the agreement required it. Fermi has not linked this dispute to the nuclear departure. Still, for a project whose longest-dated element depends on sustained, specialized leadership, stability at both the executive and board level is part of the execution picture.
Background
Fermi Inc., headquartered in Dallas and listed on Nasdaq and the London Stock Exchange as FRMI, operates as Fermi America. It is an emerging growth company developing Project Matador, a private electric grid and data center campus in the Texas Panhandle on land that is owned or under long-term lease, including a land lease with the Texas Tech University System. The model is “behind the meter”: power is generated on site and delivered over a private grid, rather than purchased through the public grid and its long interconnection queues.
The pitch targets what Fermi calls the biggest constraint in AI infrastructure: access to large-scale, reliable power on the timelines hyperscalers need. The company combines near-term gas-fired generation and utility supply with a longer-term ambition of up to 17 GW. Its investor presentation says that ambition is contingent on permits, financing, customer commitments and nuclear licensing. Source: Fermi’s Top Nuclear Builder Departs as Reactor Timeline Slips (Hart Energy), reporting on a senior nuclear departure and reactor schedule slip at Fermi. Primary sources: Fermi Inc. Form 10-Q for the third quarter of 2026 (filed November 12, 2026); Fermi Inc. Form 8-K current report (filed November 12, 2026); Exhibit 99.1: Fermi Q3 2026 earnings release, “Fermi Announces New Agreements with Investment-Grade Hyperscaler and Third Quarter 2026 Financial Results”; Exhibit 99.2: Fermi Investor Presentation, Third Quarter 2026; Fermi Inc. Form 8-K on 2026 Annual Meeting voting results (filed November 4, 2026).Sources

