The First Customer
The U.S. Air Force just bought a nuclear reactor.
That sentence probably needs some context. The Air Force didn't buy a nuclear reactor the way you buy a refrigerator. They awarded a contract — an SBIR Phase I through AFWERX, their innovation arm — to a company called NANO Nuclear Energy (NASDAQ: NNE) to advance a microreactor called KRONOS MMR for base power applications.
If you don't follow defense procurement or advanced nuclear, that probably sounds like a routine government contract. It isn't. It's one data point in a much larger pattern — and the pattern explains something about where real infrastructure investment is happening right now, who's paying for it, and why the technology that powers the next economy may be built for the military first. Let me explain why.
What They Actually Bought
KRONOS MMR is a fourth-generation high-temperature gas-cooled reactor. It produces 15 megawatts of electricity and 45 megawatts of thermal power per unit. It's cooled by inert helium gas under passive circulation. It requires zero external water for normal operation or decay heat removal. It uses TRISO fuel — tiny uranium particles coated in layers of carbon and silicon carbide that essentially cannot melt at operating temperatures. It runs for five to ten years without refueling.
You can cluster multiple units together. Four KRONOS units give you 60 megawatts of electricity from a facility that fits on a modest footprint and was built in a factory, not constructed in the field over a decade.
If you know anything about traditional nuclear power, the contrast is stark. A conventional gigawatt-scale nuclear plant takes 7 to 15 years to build, costs $5 to $10 billion, requires billions of gallons of cooling water, and needs active safety systems. KRONOS is factory-built, passively safe, air-cooled, and designed to be transported to the site and set up in months.
If you know anything about diesel generators — which is what most military forward operating locations actually run on — the contrast is even starker. Diesel requires continuous fuel logistics. Microreactors run for a decade without a single fuel delivery.
The Vulnerability
Here's the part that most civilians don't know: approximately 99% of domestic U.S. military bases depend on the civilian commercial power grid for their day-to-day operations.
That means the same grid that experienced rolling blackouts in Texas during Winter Storm Uri, that has been physically attacked at substations multiple times, and that foreign adversaries have been probing through cyber operations — that grid is what powers American military installations. Drone operations, satellite command centers, missile defense radar, C4ISR logistics — all of it plugs into the same wall your refrigerator plugs into.
The Department of Defense has been aware of this for years. Congress has been aware of it. The NDAA now mandates that critical military facilities maintain islanded microgrid capability — the ability to disconnect from the civilian grid entirely and operate independently for 14 to 30 days without external supply.
Diesel generators can do that. But diesel generators require fuel, and fuel requires convoys, and convoys in a conflict zone are where people die. During operations in Iraq and Afghanistan, approximately one in every 24 fuel convoys resulted in a casualty. Up to 70% of total supply weight delivered to deployed forces consists of fuel and water.
A microreactor that runs for five to ten years without refueling eliminates that logistics tail entirely. No convoys. No diesel. No vulnerability to a fuel supply chain that an adversary can interdict.
That's why the Air Force is interested. That's why AFWERX exists — to fast-track commercial technologies into operational military capabilities.
The Bypass
Here's something that matters more than most people realize: the Department of Defense doesn't have to go through the Nuclear Regulatory Commission to deploy a reactor on a military base.
Under Section 91(b) of the Atomic Energy Act of 1954, the DoD has authority to license and operate nuclear reactors on military property for national defense purposes. That bypasses the standard commercial NRC licensing pathway, which is a multi-year process designed for large-scale civilian power plants.
The military can move faster. They can build, test, and deploy under their own safety oversight. That doesn't mean they're reckless — the military has some of the most rigorous nuclear safety protocols in existence, built around the Naval Reactors program that has operated hundreds of submarine and carrier reactors for decades without a significant incident. But it means the timeline compresses dramatically.
While commercial microreactor companies are working through NRC licensing processes that can take three to five years, the military can field a prototype in 18 to 24 months. The Air Force Research Laboratory and the 11th Civil Engineering Squadron are already working with NANO Nuclear to map real-world base electrical load profiles and thermal requirements.
The Army has its own program — JANUS — targeting operational microreactor deployment at candidate domestic installations by 2028. BWX Technologies is assembling and fueling the first U.S. military microreactor prototype at Idaho National Laboratory under Project Pele.
This isn't theoretical. It's happening.
The Pattern
If you step back, there's a pattern here that's bigger than nuclear power.
The Department of Defense has a long history of being the first customer for technologies that later transform the civilian economy. ARPANET became the internet. GPS was a military navigation system before it was in your phone. Commercial jet engines came out of military aviation programs. The silicon microchip industry was built on defense procurement contracts.
The dynamic is simple: the military has a problem, a budget, and the authority to move faster than civilian regulators. They become the anchor tenant. They pay for early development. They establish safety protocols, operating procedures, and supply chains. They de-risk the technology. Then it crosses over to commercial use.
Microreactors are following the same arc.
The Air Force needs islanded power for base resilience. The Army needs to eliminate fuel convoys. The Navy has been operating nuclear reactors at sea for decades. The technology that solves these military problems — a 15-megawatt, factory-built, passively safe, water-free reactor that runs for a decade without refueling — happens to be exactly the technology the civilian world needs right now too.
The Connection You're Probably Already Making
AI data centers need enormous amounts of constant, clean power. The grid can't deliver it fast enough — interconnection queues now average five to seven years. Hyperscalers have strict zero-emission mandates. Diesel won't work. Natural gas won't work. Coal is dead. Solar and wind are intermittent.
A cluster of four 15-megawatt KRONOS units gives you 60 megawatts of 24/7 carbon-free baseload power behind the meter — meaning it doesn't touch the utility grid at all. You put the data center next to the reactor, not next to a substation.
Microsoft signed a 20-year deal to restart Three Mile Island Unit 1. Amazon bought a nuclear-powered data center campus at Susquenhanna. Google signed a PPA with Kairos Power for 500 megawatts of advanced reactors. Equinix signed agreements with Oklo for up to 500 megawatts of microreactor power.
The commercial demand is real. But the commercial regulatory pathway is slow, and the supply chain is immature. The HALEU fuel that these reactors need — high-assay low-enriched uranium — has almost no domestic production. Centrus Energy is currently the only licensed U.S. producer. TRISO fuel fabrication requires specialized manufacturing capacity that barely exists at commercial scale.
The military can solve part of that problem by being the first customer. By ordering reactors, establishing operational protocols, and building the fuel supply chain through defense contracts, the DoD creates the infrastructure that commercial deployment depends on.
That's the pattern. The military builds the bridge. The commercial market crosses it.
What This Actually Means
The interesting part isn't that a startup won a government contract. Startups win government contracts every day.
The interesting part is that the U.S. military — the largest single energy consumer in the world — has decided that small nuclear reactors are the answer to its most strategic vulnerability: dependence on a civilian power grid it doesn't control and a fuel logistics chain that gets people killed.
And the interesting part is that the answer to that military problem — factory-built, passively safe, water-free, decade-long-refueling microreactors — is the same answer to the civilian problem of powering AI infrastructure without waiting a decade for grid interconnection.
The military doesn't care about the AI data center angle. They care about mission assurance. They care about a base that stays operational when the grid goes down. They care about not sending fuel convoys through hostile territory.
But the technology they're de-risking — the reactor design, the fuel supply chain, the regulatory precedent, the operating procedures, the safety framework — all of it transfers. The same KRONOS MMR that powers an Air Force base can power an AI inference campus. The same TRISO fuel. The same helium cooling. The same walk-away passive safety.
The Department of Defense is building the infrastructure layer for the next economy. They're doing it because they have to. The civilian market will benefit because the technology crosses over.
The Constraint Nobody's Talking About
There's a bottleneck in this story, and it's not the reactor design.
It's the fuel.
HALEU — high-assay low-enriched uranium, enriched between 5% and 20% U-235 — is what these microreactors run on. Before 2022, Russia's Tenex was the sole commercial supplier. The U.S. essentially outsourced its advanced nuclear fuel supply chain to the country it now considers its primary strategic adversary.
Centrus Energy, operating out of Piketon, Ohio, is currently the only licensed domestic HALEU producer. They've produced roughly one metric ton through early 2026 under DOE contracts. Commercial scale remains constrained through at least 2027-2028.
TRISO fuel fabrication — the specialized process of coating uranium kernels in layers of pyrolytic carbon and silicon carbide — requires manufacturing capacity that barely exists at scale.
You can design all the reactors you want. If you can't fuel them, you've built very expensive paperweights. The fuel supply chain is the real infrastructure gap, and it's the one that's hardest to close because it requires enrichment capacity, regulatory licensing, and specialized manufacturing that takes years to stand up.
This is where defense procurement could matter most. The military doesn't just need reactors — it needs fuel for those reactors. If defense contracts help build a domestic HALEU supply chain at scale, that infrastructure serves both military and commercial deployment. The fuel is the bridge.
Where This Goes
The timeline is not fast. Anyone telling you microreactors will be powering data centers next year is selling something. The realistic window for operational military deployment is 2028-2030. Commercial crossover at scale is probably 2030-2035.
But the pattern is what matters. The military is moving first because it has to. The technology it's de-risking — microreactors, TRISO fuel, HALEU supply chains, passively safe designs, factory fabrication — is the same technology that will eventually power the infrastructure layer of the AI economy.
We've seen this movie before. The government builds the thing nobody else can afford to build. The commercial market inherits it. The internet. GPS. Jet engines. Microchips. Now, possibly, nuclear microreactors.
The Air Force didn't just buy a reactor. They bought the first unit in a supply chain that might end up powering both the base that launches the drones and the data center that trains the models running them.
That's the story. The military is the first customer. The commercial market is the second. The infrastructure is the same.
The constraint used to be computing power. Then it was data. Then it was model capability. Now it's electricity.
Follow that logic far enough, and the next great technology platform doesn't start with a software release. It starts with a reactor that was built for a military base and ended up powering an AI data center next door.
Rich Washburn is a technologist and strategist working at the intersection of AI, infrastructure, and capital. He is Managing Partner and Chief AI Officer at Eliakim Capital and CIO of Data Power Supply.
Comments