The Case for Putting SMRs on Military Bases
The U.S. military runs on diesel it can barely protect, and small modular reactors offer the most compelling fix anyone has proposed in decades.
Here is a number worth sitting with: in Afghanistan in 2007, one American soldier or civilian was killed or wounded for every 24 fuel resupply convoys. According to the UC Institute on Global Conflict and Cooperation, roughly 900 of those convoys ran that year alone. The fuel wasn’t going into jets or tanks. Much of it was running diesel generators to keep the lights on, the radios live, and the air conditioners churning at forward operating bases. Fuel to make electricity. Electricity that a small reactor could have provided without a single convoy.
That’s the whole case, right there. Everything else is detail.
The U.S. military has known about its energy problem for years, done occasional things about it, and never really solved it. Most bases at home still pull power from the civilian grid. Abroad, they run on diesel that costs up to $15 per gallon delivered and requires an unbroken chain of trucks through hostile territory to keep flowing. President Trump signed Executive Order 14299 in May 2025, titled “Deploying Advanced Nuclear Reactor Technologies for National Security,” which mandated a working reactor on a U.S. military base by September 30, 2028. That’s not a vision statement. That’s a deadline.
The vulnerability that diesel created
The problem goes deeper than convoy casualties, though those alone should have settled the argument years ago. 🛡️
Most U.S. military bases are directly connected to the civilian electricity grid, which is a large, complex, and genuinely fragile piece of infrastructure. A U.S. Department of Energy study from 2017 concluded that a serious grid failure was a matter of “when,” not “if,” and the grid’s vulnerability to both cyberattacks and physical attacks on transformers has not improved dramatically since then. Defense News published an assessment in early 2025 noting that the military faces two distinct energy failures waiting to happen: the civilian grid going down, and the diesel backup system failing because its fuel supply gets cut.
Neither of these is hypothetical. The consequences include:
Command and control systems going dark when grid power fails 📡
AI-enabled targeting and radar arrays losing power during a conflict
Communications infrastructure failing at exactly the moment it’s needed most
Patriot missile batteries and other air defense systems running on generators that depend on the same fuel convoys they’re there to protect 🚛
The Atlantic Council’s analysis from May 2026 frames it plainly: the Department of Defense has a target of achieving 99.9% energy availability for critical missions by 2030. That standard allows for less than nine hours of downtime per year. With current infrastructure, that target is something between aspirational and delusional. 💡
The military has tried solar and wind. Useful in the right contexts, genuinely not suitable as primary power for a base running 24/7 surveillance, logistics, and weapons systems regardless of weather. What it needs is baseload power that doesn’t require a supply chain to operate. That is exactly what a nuclear reactor provides.
What Project Janus actually is
In October 2025, Secretary of the Army Dan Driscoll and Secretary of Energy Chris Wright jointly announced Project Janus, the Army’s formal program to put nuclear microreactors on military bases. ⚛️ The following month, the Army named nine candidate installations:
Fort Benning, Georgia
Fort Bragg, North Carolina
Fort Campbell, Kentucky
Fort Drum, New York
Fort Hood, Texas
Fort Wainwright, Alaska
Holston Army Ammunition Plant, Tennessee
Joint Base Lewis-McChord, Washington
Redstone Arsenal, Alabama
These weren’t picked arbitrarily. The Army evaluated them on mission criticality, existing energy infrastructure, resiliency gaps, and suitability for reactor deployment, according to statements from the Army at the time. Fort Wainwright in Alaska is worth a particular look: it’s remote, cold, and expensive to supply, making it precisely the kind of installation where eliminating the fuel dependency matters most. 🌨️
Janus builds on Project Pele, a 1.5-megawatt gas-cooled demonstration microreactor whose core is now under construction at Idaho National Laboratory, expected to begin producing electricity in 2028. The program is structured as a public-private partnership, meaning commercial vendors own and operate the reactors on Army installations under military oversight, following a contracting model borrowed explicitly from NASA’s Commercial Orbital Transportation Services program.
The Defense Innovation Unit issued its solicitation for reactor designs in November 2025, seeking designs up to 20 MWe. ⚙️ Companies pre-qualified under the broader Advanced Nuclear Power for Installations program include BWXT, Westinghouse, Kairos Power, Oklo, X-energy, and Radiant Industries, among others. The Air Force is running a parallel track, having selected Eielson Air Force Base in Alaska for a microreactor pilot and announced its intent to award a contract to Oklo for a 5-megawatt sodium-cooled reactor under a 30-year fixed-price arrangement.
Why the military is a better first customer than the grid
The commercial SMR market faces a genuine chicken-and-egg problem: reactors need to get cheaper to attract customers, and they get cheaper by getting built in volume, but they can’t build in volume without the customers. The military, with its unique combination of capital, patience, and regulatory authority, may be the best entity alive to break that loop. 🔑
Consider what the military brings that commercial buyers cannot: 🏛️
Long time horizons: The Army is explicitly thinking about 2030 and beyond, not quarterly earnings
Independent regulatory authority: The Army can operate nuclear reactors under its own regulatory framework, distinct from the NRC’s civilian licensing process, which compresses timelines significantly 📋
Predictable demand: A base draws a known amount of power continuously, making sizing and contracting straightforward
Tolerance for first-of-a-kind costs: The commercial market killed NuScale’s Utah project when costs rose. The military makes different calculations about value
Strategic urgency: The threat to diesel supply chains is real and quantified, not a hypothetical 🚨
There’s also a spillover effect worth taking seriously. Neutron Bytes, which covers nuclear developments in depth, noted in October 2025 that if military microreactor deployment takes off alongside civilian AI data center demand in the 2030s, the combined economic effect on the U.S. nuclear supply chain could be significant, though both sectors competing simultaneously for the same fabrication capacity could also push costs up. The history of advanced technology programs suggests the military’s early adoption creates the production base that later makes civilian deployment cost-effective. That is more or less exactly how GPS, the internet, and aircraft manufacturing all worked.
The real obstacles, honestly assessed
None of this means it’s straightforward. There are genuine obstacles, and pretending otherwise would be a mistake. 🔧
Fuel supply is probably the biggest. ⚠️ Many advanced reactor designs require high-assay low-enriched uranium, or HALEU, which is enriched to between 5% and 20% uranium-235. The OECD Nuclear Energy Agency reported in September 2025 that more than half of the SMR designs planning to use HALEU had not progressed beyond non-binding agreements with national laboratories about fuel supply. The U.S. nuclear industry has warned openly that some SMR deployment timelines could slip by years if HALEU production doesn’t scale. Centrus Energy launched commercial HALEU enrichment in Ohio in late 2025, and Urenco USA produced its first batch of enriched uranium above 5% in New Mexico around the same time, but the supply chain is still early.
There are also real questions about:
Security at the installation: A reactor is not a diesel generator, and hardening protocols for small reactors on military bases are still being developed
Operator training: Janus is structured so commercial companies run the reactors, which means military personnel are not reactor operators, raising questions about continuity during deployment and conflict 👷
Community relations: All nine candidate bases sit near civilian communities, and the Army has explicitly committed to “transparency with host communities” as part of the program
The 2028 deadline: Getting a reactor licensed, built, and operating in under three years is aggressive. Project Pele, a demonstration reactor, has been in development since 2020 and won’t produce power until 2028 at the earliest. The compressed timeline for Janus reflects political will more than a technical schedule
To be fair, the designs being evaluated for Janus mostly use standard low-enriched uranium fuel, not HALEU, which sidesteps the fuel supply problem for the initial wave. Oklo’s Aurora uses metal uranium fuel with a different supply chain dynamic entirely. So the HALEU constraint matters more for second- and third-generation commercial SMR builds than for the first military installations.
What success actually looks like here
If Project Janus delivers an operating reactor at, say, Fort Wainwright by late 2028, a few things happen simultaneously. ⚡
The Army gets actual data on what it costs and what it takes to run a microreactor under military conditions, data that is currently unavailable anywhere. 📊 The commercial supply chain gets an anchor customer and a production run that justifies investment in manufacturing capacity. The NRC and the Army’s own nuclear oversight bodies develop the regulatory precedents that future civilian deployments will rely on. And the broader case for SMR deployment in difficult environments, from remote industrial sites to island communities, gets a proof point that no simulation or white paper can match. 🌏
The U.S. Energy Information Administration’s latest survey of SMR development, updated in April 2026, notes that DOE has allocated $900 million to accelerate SMR deployment and is running a parallel Energy Reactor Pilot Program through commercial vendors. The military deployments and the civilian deployments are moving in parallel now, feeding the same supply chain. If both arrive in the 2028-2030 window as planned, the economics of the third and fourth units will look substantially different from the first.
None of this is guaranteed. The history of large technical programs inside the U.S. government contains a sobering number of over-promised, under-delivered projects with 2028 milestones that somehow became 2035 milestones. But the drivers here are different from, say, a Pentagon software procurement. The fuel convoy death toll is a documented, quantified cost that is not going away. The grid vulnerability is real and worsening. And the reactor designs are not conceptual, they’re physical hardware with known supply chains being fabricated right now in Idaho.
The question isn’t whether the military has good reasons to want nuclear power on its bases. It obviously does, and has for decades. The question is whether the combination of political will, commercial urgency, and available technology that exists in 2026 is finally sufficient to get it there.
Based on everything visible right now, I think the answer is probably yes. For the first time in a long time, “probably” feels like more than optimism.
What’s your read on the most likely obstacle? HALEU supply, regulatory timing, cost overruns, or something nobody is talking about yet?