America's Boldest Nuclear Bet Just Passed Its First Real Test

On Friday, the U.S. Nuclear Regulatory Commission published a draft environmental review for four new nuclear reactors in the Texas Panhandle, and its staff recommended approving them. It is one sentence in the Federal Register, but it is the most significant licensing milestone for American nuclear power in years.

The project is Project Matador, Fermi America's planned energy and AI computing campus in Carson County, Texas, about 17 miles northeast of Amarillo. Four Westinghouse AP1000 reactors would supply roughly 4 gigawatts of baseload power to a private campus grid that Fermi says could ultimately reach 17 gigawatts. The draft environmental impact statement, published October 9 (91 FR 64702), gives a preliminary recommendation that the NRC issue the requested combined license. That recommendation is contingent on the safety review, which is far from finished.

The speed is the surprise. Fermi submitted the first portion of its application in June 2025. The draft review arrived roughly 15 months later, after a process that typically takes years. But speed and licensing are different things, and the fine print matters as much as the headline.

The draft arrived in about 15 months. The final license, if it ever comes, is still years of hearings and safety review away.

What the NRC actually said

Precision matters here, because this was not an approval. The NRC staff's draft gives a preliminary recommendation on environmental grounds only: build these four reactors, and the environmental impacts are acceptable. A combined license still requires a final environmental review, a completed safety review, a mandatory hearing, and a final vote by the Commission. None of that has happened.

Each of the four units would be a Westinghouse AP1000, rated at 3,415 megawatts thermal and at least 1,000 megawatts electric net. They would sit on an 881-acre site within a 5,236-acre campus, next to the Department of Energy's Pantex Plant, on land Fermi leases from Texas Tech University under a 99-year agreement.

The review's details reveal the trade-offs of building big reactors in a semiarid region. The units would use dry cooling, with 160 fans per unit and no cooling water, because the nearest surface water source of sufficient capacity, the Canadian River, is about 15 miles away. The fans themselves would draw 20 to 40 megawatts per unit just to keep the condensers cool, a real efficiency cost for avoiding water use. Process water demand is estimated at 1.4 million gallons per day from municipal supply.

Construction impacts are rated large where it counts locally. The draft projects up to 8,200 workers on site at the peak, which would overwhelm short-term housing and push traffic on Farm to Market Road 683 beyond its capacity. Those impacts get the "large" rating; land use, air quality, and housing land in the moderate to small range. Fermi estimates 2,200 construction workers per unit, with starts staggered about two years apart across a roughly 12-year build. Operations would need 600 workers for the first unit and 230 for each additional one, plus 1,000 more during refueling outages.

The public has until November 23, 2026 to comment on the draft, under Docket ID NRC-2026-0100.

A private grid, not a power plant

The reactors are only part of the story. Project Matador is conceived as a private energy system built to power AI computing, not a merchant power plant selling into the grid. Nuclear generation is described as the behind-the-meter baseload for the campus, with natural gas, solar arrays, and utility-scale battery storage filling out the rest.

The scale is hard to overstate. Original plans called for an 11-gigawatt campus anchored by about 4.4 gigawatts of nuclear. The ambition has since expanded to nearly 17 gigawatts. Back in February, Texas environmental authorities granted the project a 6-gigawatt Clean Air Permit, the second-largest air permit in U.S. history, which paves the way for the gas-fired generation that will support the campus alongside the reactors. Fermi has said it wants 1 gigawatt of non-nuclear generation operating by the end of 2026.

That gas component is worth stating plainly. This is not a fully clean-energy campus. It is a hybrid energy complex designed around one objective: round-the-clock power for computing. Intermittent renewables cannot carry that load alone, and the campus wants its own electrons rather than competing for grid capacity. The private-grid model, building your own power system instead of waiting for the public grid to catch up, is the bet being tested here.

Why the licensing process itself is the news

The milestone is also a test of a new way of doing environmental reviews. Fermi's application went through the NRC's applicant-prepared environmental review pilot: under a federal provision (42 U.S.C. 4336a(f)), the applicant drafts the environmental impact statement under the agency's supervision, and the NRC staff then evaluates it independently and takes responsibility for the content. Fermi's parent company says Project Matador was the first project to complete the pilot's required steps.

The NRC's July schedule letter estimated the remaining environmental work at about 1,000 staff hours and $175,000 in contractor support, a small price for the agency if the pilot works as designed. The model shifts the drafting burden to the applicant, which is why the draft arrived in 15 months instead of the typical multi-year crawl.

It is arriving alongside a broader push to compress nuclear licensing. Under Executive Order 14300, the NRC has said it will finish the review within 18 months of receiving a complete application. That clock only starts once Fermi submits the site-specific safety analysis, the final part of its application, which is due December 31, 2026. The safety review has no schedule yet.

What comes next

The remaining milestones are stacked and none of them is guaranteed. The NRC is targeting a final environmental impact statement in December 2026. Then comes acceptance of the safety analysis, a formal notice of opportunity for a hearing, the safety review itself, the mandatory hearing, and the Commission's decision, all unscheduled.

Fermi's own timeline has already slipped once. Its quarterly filing plans to commission one reactor per year from 2033 to 2036, about a year later than its earlier target for the first unit. It has signaled intent to begin nuclear construction next year, contingent on licensing, financing, and construction execution. Hyundai Engineering and Construction finished a front-end engineering design study in July 2026. The reactor coalition also includes Westinghouse, ENERCON, and Preferred Licensing Services.

The AP1000 design carries a track record. It has held NRC design certification since 2006, and units already operate at the Vogtle plant in Georgia, the only completed large reactors in the U.S. in a generation. That pedigree is exactly why Fermi chose it: a standardized, licensed design is supposed to be easier to build. Whether "supposed to" survives contact with a 12-year, four-unit construction program in the Texas Panhandle is the multi-billion-dollar question.

For now, the draft review is a milestone, not a verdict. It says the environmental case for four new reactors next to the Pantex Plant holds up. The harder questions, safety, cost, financing, and whether the power will arrive when the AI boom needs it, are still entirely open.