Three U.S. Microreactors Just Reached Criticality. Data Centers Are Next.

A third experimental microreactor reached criticality this year under a Department of Energy program designed to push next-generation nuclear designs toward commercial use, the agency confirmed. The unit, called Unity and built by Houston-based Deployable Energy, joins demonstration reactors from Antares of Torrance and Valar Atomics of El Segundo that cleared the same milestone in June, giving the United States three advanced test reactors running in the same push for nuclear designs aimed at high-demand customers such as data centers.

All three are far smaller than traditional nuclear plants and are being used to test new reactor designs and fuels that industry hopes will be easier and cheaper to deploy than today’s large reactors, which provide roughly one-fifth of US electricity. Criticality is the point at which a reactor sustains a controlled nuclear chain reaction, the first operating benchmark any new design must hit.

What Unity and the program cover

Unity is one of two demonstration efforts tied to DOE work announced this year. The reactor from Valar and an Aalo Atomics unit are part of a pilot program launched last year, while Deployable’s project sits in a separate DOE program that began in March. Both are designed to move companies from concept to demonstration and, eventually, toward commercial deployment.

The agency has given Antares and Deployable access to its laboratories and overseen detailed safety reviews for their demonstration units, and it also provided the final authorization that allowed the reactors to go critical. Bobby Gallagher, CEO and co-founder of Deployable, said meeting the criticality benchmark would not have been possible “without the firm dedication of the DOE for safety, quality and speed” or without support from staff at Idaho National Laboratory, where Unity is being tested.

The data-center demand pulling on the program

The administration’s push comes as power demand in the United States rises, especially from data centers built for artificial intelligence and cloud computing. Those facilities use enormous amounts of electricity, and some developers are looking at microreactors and small modular reactors as possible dedicated power sources.

Michael Goff, principal deputy assistant secretary in DOE’s Office of Nuclear Energy, told Politico that he expects near-term announcements tying companies in the pilot to specific data-center customers. “I know there are a number of companies that are in the pilot program that are already in discussions on data centers, and I expect in the very near future them to start making some announcements of, ‘This technology is going to deploy a data center for this company,'” he said. “I think it’ll be just a matter of a few years.”

Goff framed the recent pace of work as a break with past practice. “The fact that they can do this and do this so quickly, it shows that the preconceived notions we’ve had about nuclear are no longer valid anymore,” he said. “Nuclear doesn’t have to take a long time if we have the right enabling environment to move forward. Nuclear can move forward fairly rapidly.”

How far criticality is from the grid

The authority to grant commercial licenses still rests with the Nuclear Regulatory Commission. The agency is undergoing changes prompted by a White House directive and the bipartisan ADVANCE Act, passed in 2024, which directs regulators to streamline reviews of new reactor designs. A proposed rule would give advanced reactors a faster path to NRC approval if they have already been reviewed by DOE or the Defense Department.

Gallagher said Deployable is likely to apply for a commercial license later this year, once a new microreactor licensing rule is finalized, and he expects the NRC review to take six to twelve months. Even so, people in the sector caution against assuming that successful tests mean commercial power is close at hand. Alison Hahn, former head of advanced reactors at DOE and now a senior director at the Nuclear Energy Institute, said the demonstration work is bounded by design. “The pilot program itself is for demonstration reactors. It is not for commercial power to the grid,” she said. “However, it does move the needle technically and in terms of the supply chain, it does move the industry forward.”

The fuel, the units, and the money

Three constraints sit between today’s milestones and data-center power purchases. The first is fuel: some designs depend on high-assay low-enriched uranium, or HALEU, which is in short supply and has no commercial source in the United States today. The second is scale: some of the smallest units may not produce enough power on their own to meet the needs of large data centers. The third is capital: DOE’s pilot program does not provide construction funding, leaving companies to raise money for projects that can still cost billions of dollars, even at smaller sizes. Deployable did not receive federal dollars for its reactor.

James Richards, manager of economics and project development at the Nuclear Innovation Alliance, said moving from a handful of test units to a deployable fleet remains the harder step. “It’ll be incumbent on these companies to really take what they’ve learned and actually kind of apply that in a scaled-up production,” he said. “That’s not a given. That’s still a very difficult task.”

Where critics land

Financing has drawn the sharpest objections. Edwin Lyman, director of nuclear power safety at the Union of Concerned Scientists, described the pilot program as a “performative exercise” without federal financing attached. “The program doesn’t actually come with any financing for these projects, and that is probably the number one obstacle to reactor deployment – the lack of sufficient financing for the multibillion-dollar capital costs of even a small reactor,” he said.

The focus on hitting criticality has drawn criticism from some nuclear advocates as well. The center-left nonprofit Third Way has called DOE’s emphasis on the pilot program “an unhelpful diversion” from broader nuclear ambitions, including more than $17 billion in federal loans aimed at large reactors. The group says that while criticality yields useful data on new fuel and reactor designs, it does not amount to a commercial breakthrough. Goff has held the line that criticality is the “first step” on a longer path. “We’ve got to go beyond criticality,” he said. “If we were just focused on criticality, I think it might just be misplaced.”

What the broader policy is asking of the industry

The administration’s underlying target is to revive nuclear power after years of slow growth. It has set an objective of quadrupling net nuclear generating capacity to 400 gigawatts by 2050, a steep climb for an industry that has seen only a handful of new reactors come online in three decades. Support from both parties in Congress, the recent start-up of two large reactors in Georgia, and concern over rising electricity costs have all fed interest in small modular designs and microreactors.

DOE has cut much of the red tape that critics say slowed nuclear development in the past, and officials including Energy Secretary Chris Wright have said NRC staff have been involved throughout the demonstration process. Goff said he hopes that early involvement translates into an “accelerated process” when companies seek full licenses. For now, even with three reactors running, none of these microreactors or small modular reactors is yet in commercial use in the United States, and the next announcement Goff expects is a data-center deal rather than a licensed power plant.

Frequently Asked Questions

How many experimental microreactors does the US have running?

Three demonstration reactors backed by DOE have reached criticality this year, from Deployable Energy, Antares, and Valar Atomics, according to the Department of Energy.

What does reaching criticality mean?

Criticality is the point at which a reactor sustains a controlled nuclear chain reaction. DOE’s Michael Goff described it as the “first step” on a longer path to commercial deployment.

Why are data centers interested in microreactors?

Data centers built for AI and cloud computing use large amounts of electricity, and some developers are looking at small nuclear units as possible dedicated power sources, the Energy Department says.

What still stands between these reactors and commercial use?

Three things: a commercial license from the Nuclear Regulatory Commission, a US supply of high-assay low-enriched uranium fuel, and private financing for projects that can still cost billions of dollars. Deployable did not receive federal dollars for its reactor, and DOE’s pilot program does not provide construction funding.

What is the administration’s nuclear capacity goal?

The administration has set an objective of quadrupling net nuclear generating capacity to 400 gigawatts by 2050.