TVA Small Modular Reactor Permit Clears Clinch River Construction

A regulatory first, with conditions

The U.S. Nuclear Regulatory Commission has approved a construction permit for the Tennessee Valley Authority to build a GE Vernova Hitachi BWRX-300 small modular reactor at its Clinch River site in Roane County, Tennessee. The agency says the decision clears the way for the country's first commercial small modular reactor.3 The planned unit would produce 300 megawatts, and the NRC says TVA intends to use it to meet rising regional power demand.2 The agency finished its review in 14 months. That is four months ahead of its own schedule, and the NRC counts it as its second major reactor approval this year.19

The headlines suggest more than the permit actually allows. It lets TVA build. It does not let TVA load fuel or run the reactor. Before the plant can produce any electricity, TVA will need a separate operating license from the NRC.9 TVA has not set a construction start date. It says it is still negotiating with industry partners over how to share costs and keep spending accountable during design and construction.1 No final project cost has been made public. TVA is studying as many as four BWRX-300 units at Clinch River, but this permit covers only the first.4

So the right way to read this is as a permit to start the hard part, not as proof the technology works. Even so, it matters a great deal for an industry that has long struggled to get from paper designs to poured concrete.

How fast the review moved, and why

The NRC is using the timeline as evidence that it has changed. Chairman Ho Nieh said the decision shows the "new NRC" can make rigorous safety calls that let major nuclear projects proceed.1 Jeremy Bowen, who heads the agency's Office of Advanced Reactors, said at the signing that staff finished the safety review in 12 months and the permit was signed in under 14.8

Outlets agree on the overall arc but differ on some details. GE Vernova Hitachi and other accounts say TVA filed its application in April 2025. World Nuclear News says May 2025.5131 The Federal Register notice dates the permit September 28, while the NRC announced it on September 29.8 None of this changes the substance, but it shows how much of the early coverage came straight from press releases.

The groundwork goes back further. TVA got an early site permit for Clinch River in 2019.2 That permit is valid for 20 years and covers any light-water SMR design, which helps explain why the construction-permit review could move quickly.13 The reactor design itself also helped. World Nuclear News describes the BWRX-300 as a water-cooled reactor that circulates coolant naturally, uses passive safety systems, and borrows the design and licensing basis of GE Hitachi's larger ESBWR.1 One analysis argues that this long boiling-water-reactor lineage gave regulators decades of operating data to work from. That is an advantage newer, more experimental reactor types do not have.6 This is a plausible explanation, but it is the outlet's interpretation, not something the NRC has said.

The AI demand backdrop

The permit comes as electricity demand is rising for the first time in a generation, and the reporting is clear that data centers are a big reason. One summary of Lawrence Berkeley National Laboratory research says data centers could use about 12% of U.S. electricity by 2030, up from 4.4% in 2023.12 One tracker estimates U.S. data centers used about 192 terawatt-hours in 2024, roughly 4.7% of national demand, and expects that to reach 464 to 649 terawatt-hours by 2028–2030. An industry roundup adds that grid interconnection delays of five to seven years are pushing developers toward on-site power such as gas turbines and nuclear.18

The pressure is already visible in TVA's territory. Former TVA chief executive Don Moul said earlier this year that the utility expects data center load in its region to double by 2030, and xAI's Colossus supercomputer in Memphis is on TVA's system.12 Interim president Mike Skaggs linked the project to national goals directly, saying TVA's partnerships would support new nuclear for "energy security and AI dominance."8

The engineering case for nuclear in this market is simple. Many AI data centers need steady, firm power around the clock. Solar and wind vary with the weather and may need storage or backup to fill the gaps, while a reactor runs at constant output.11 Much coverage of SMRs for data centers makes the same point: hyperscale AI campuses run continuously, and renewables alone cannot promise 24/7 supply.14

A 300-megawatt unit and a gigawatt problem

The sense of urgency hides a mismatch of both size and timing. One sizing analysis estimates that a data center with 1 gigawatt of computing load, at typical efficiency, would draw about 1,300 megawatts. Meeting that would take about five BWRX-300s, plus a sixth as backup.16 The scale is already showing up in real proposals. Blue Energy wants to pair up to five BWRX-300s with GE Vernova gas turbines to serve a Texas data center campus directly, with a final investment decision targeted for 2027.11 Another report describes this as the first request ever made to a grid operator to connect SMRs, at up to 1.5 gigawatts.17 Blue Energy filed the first part of its own NRC construction permit application in September.1

Timing is the bigger problem. Clinch River is aiming for commercial operation in the early 2030s.12 One analysis notes that this target comes from earlier industry statements and has not been confirmed by TVA or the Department of Energy.6 An industry roundup expects SMRs to become widely available only around 2032–2035, which leaves a gap for AI demand in the meantime.18 Put simply, the data center growth driving this permit will arrive years before Clinch River can deliver power. Gas, grid upgrades and existing nuclear plants will have to cover most of the next decade.

Cost is the real test

The coverage is most united on one point: the permit settles the licensing question but not the economic one. The warning everyone cites is Georgia's Vogtle plant, whose two newest reactors cost about $35 billion combined and came online years late.11 SMR developers argue that smaller, standardized units built more like factory products could avoid that kind of overrun.11 TVA says modular design lets major components be built in factories and shipped to the site, cutting time, cost and complexity.8

The early cost numbers suggest caution. Using public figures, one analysis calculates that Ontario's first BWRX-300 at Darlington costs about C$25,667 per kilowatt. It also notes that NuScale's Utah project for UAMPS was cancelled in 2023 after its estimate rose to about $9.2 billion for 462 megawatts.16 Public money is already in the project. The Department of Energy committed $400 million and TVA about $150 million before the permit was granted.6 Skaggs's statement that Valley ratepayers should not "shoulder the costs of first-of-a-kind technology" suggests TVA knows the first unit will cost more than later ones.1

Where this fits in the pipeline

The claim of being "first" needs some context. In March 2026 the NRC issued a construction permit for TerraPower's Natrium plant in Wyoming, which pairs a 345-megawatt sodium-cooled reactor with molten-salt storage. Kairos Power also holds permits for its Hermes 2 project, which is meant to supply power to Google through TVA's grid.13 TVA itself calls Clinch River only the second advanced reactor construction permit ever issued in the U.S.8 The most defensible way to put it is that Clinch River is the first commercial light-water SMR permit and the first BWRX-300 permit in the U.S., not the first advanced reactor of any kind.

The BWRX-300 does have one clear advantage: a sister unit is already ahead of it. The first BWRX-300 is being built at Ontario's Darlington site1, and Ontario Power Generation applied for its operating license in March 2026.5 If Darlington stays on budget and on schedule, Clinch River gets a template. If it does not, U.S. buyers will notice.

The verdict

The permit is a real milestone, and the NRC's speed matters on its own terms. It weakens the long-standing argument that U.S. licensing is the main obstacle to new nuclear. But the AI-driven demand behind the project will not be met by Clinch River this decade. The project's real job is to show whether a 300-megawatt reactor can be built at a cost that utilities and data center developers will accept. TVA describes the permit as a repeatable licensing path for future applicants.1 What happens to all the reactors that might follow will depend on what TVA spends once construction starts.