The Pentagon is betting $billions on tiny nuclear reactors, and startups are racing to deliver

Photo: Budget Bizar
The U.S. military has a diesel problem. Bases in Alaska, Montana, Colorado, and Texas rely on fuel shipments to keep the lights on, the communications running, and the weapons systems operational. That supply chain is a vulnerability, and the Pentagon is now spending heavily to eliminate it with technology that sounds like science fiction: nuclear reactors small enough to fit on a truck.
These are called microreactors, and they generate between 1 and 20 megawatts of electricity, enough to power a small town or a major military installation. They are built in factories, shipped to sites, and designed to run without a connection to any power grid. The military's interest is what's making them real.
The military as launch customer
The Department of Defense launched Project Pele in 2019 specifically to bring 24/7 nuclear power to remote and challenging military environments. Under that program, BWXT Advanced Technologies is building a 1.5-megawatt mobile reactor at Idaho National Laboratory, due for deployment in 2028.
The Army followed with the Janus Program last October, aiming to install a reactor on a military base by September 2028. Meanwhile, the Air Force chose a company called Radiant to deliver microreactors to Buckley Space Force Base in Colorado and partnered with Oklo to design and operate a reactor at Eielson Air Force Base in Alaska. Westinghouse Government Services is headed to Malmstrom Air Force Base in Montana, and Antares will deploy one at Joint Base San Antonio in Texas.
Radiant is building its microreactor factory right now in Oak Ridge, Tennessee. Antares has already secured the specialized nuclear fuel it needs and plans to test an electricity-producing prototype reactor in 2027.
Jordan Bramble, Antares CEO, put it plainly: the Pentagon is "the best first customer for advanced nuclear because of their experience base, coordination with the Department of Energy, unique regulatory pathways, access to specialized fuel, and willingness to invest in iterative tech maturation." Companies that deliver for the military, he argued, graduate to commercial markets with lower borrowing costs and stronger balance sheets.
Why this matters beyond the base fence
The military's role here is to absorb the brutal cost of being first. Proving a new reactor design, navigating federal licensing, building out a supply chain, none of that is cheap or fast. By paying for the first generation of these machines, the Pentagon compresses the timeline and de-risks the investment for private capital that comes later.
That later market could be large. Millions of Americans in remote communities, from rural Alaska to isolated stretches of the Mountain West, currently pay some of the highest electricity prices in the country precisely because they run on diesel that has to be trucked or flown in. Microreactors could undercut those prices. Stephen Comello, Executive Director at the Nuclear Scaling Initiative, told Reuters that remote and off-grid communities are a natural fit.
Data centers are another target. Abdalla Abou-Jaoude, who runs Idaho National Laboratory's Marvel microreactor project, said that if microreactors can produce electricity below retail prices, they could become attractive to data centers in certain states. Given how aggressively tech companies are hunting for guaranteed power to run artificial intelligence infrastructure, that market could move faster than anyone now expects.
The regulatory picture is also improving. The Nuclear Regulatory Commission announced a new approval pathway for microreactors in April, designed to be faster than the process for larger reactors. That matters because speed is most of the commercial argument.
The companies still have to prove the technology works reliably at scale, bring costs down through factory production, and build supply chains that don't yet fully exist. The military contracts buy time and credibility. Whether that converts into commercial deployment at meaningful scale is still an open question, and probably a 2030s story. But the race has started, and for remote communities paying diesel prices for electricity, the outcome is worth watching.










