# The Army Wants Reactors on Its Bases Again. Project Pele Is the $300 Million Test. A June 2022 contract worth about $300 million to build a single prototype anchors the Army's push to put nuclear microreactors on nine US bases by 2028. Author: J.A. Watte Published: July 18, 2026 Source: https://jwatte.com/blog/army-microreactors-project-pele/ --- In a plain building out in the Idaho high desert, workers are bolting together something the United States has not seriously attempted in almost fifty years: a working nuclear reactor small enough to travel. It fits inside a cluster of modules no larger than shipping containers, it runs on fuel pebbles the size of poppy seeds wrapped in layers of ceramic, and it is built to make roughly one to five megawatts of electricity far from any power line. This is Project Pele, and it is the visible tip of a much larger bet. In May 2025 the White House made the Army the Department of Defense lead for putting reactors on and around military bases, a function the Pentagon calls Installation and Operational Nuclear Energy. The single hardest number in the whole effort is about $300 million, the June 2022 contract the Strategic Capabilities Office awarded BWX Technologies to build one Pele prototype. That figure is a ceiling to deliver a single machine, not money already spent, and keeping that distinction straight is most of the honest work here. ## What the Army is trying to build, and why now The idea is old, but the paperwork behind it is new. On May 23, 2025 the President signed Executive Order 14299, which directed the Secretary of Defense to name the Army as the department's executive agent for both installation and operational nuclear energy. Installation power means fixed reactors that keep a base running when the civilian grid fails. Operational power means reactors that can be moved to a forward position. The same order set a hard clock: an Army regulated reactor operating at a domestic military installation no later than September 30, 2028. The Army's answer to that clock is the Janus Program, announced in October 2025. Janus does not ask the government to design or own the reactors. It invites commercial vendors to build, own, and operate microreactors of roughly one to twenty megawatts at up to nine bases across the continental United States, with the Army buying the power under long term agreements. That commercial ownership model is the whole point, and it is why no firm lifecycle cost exists yet. The plan is to push the heavy capital cost onto industry and pay it back slowly through the electricity bill. Underneath Janus sits an older and more concrete demonstration. Project Pele is run by the Department of Defense Strategic Capabilities Office, not by the Army executive agent function, and it predates the 2025 order. It is a transportable microreactor designed to produce one to five megawatts of electricity, fueled with TRISO pebbles that lock each speck of uranium inside a ceramic shell rated to survive very high temperatures. The prototype is being assembled for a multiyear demonstration at Idaho National Laboratory, where ground was broken on September 24, 2024. By the Department's own account, Pele is the first Generation IV reactor to begin construction anywhere outside China. ## The designation and the people who carry it An executive agent designation is quiet but powerful. It names one part of the government as the lead for a mission that touches many parts, and hands that lead the authority to plan, spend, and set the rules. The official DoD Executive Agent registry lists the Army as the executive agent for installation and operational nuclear energy under a Department of Defense directive on nuclear energy, and that registry, not any single news story, is the authority for the designation itself. Executive Order 14299 is the instrument that told the Secretary of Defense to make it happen. Section 318 of the National Defense Authorization Act is cited in coverage as the statutory basis for the executive agent role, though that citation belongs to the paper trail rather than to any dollar figure. One name runs through both the demonstration and the program. Dr. Jeff Waksman managed Project Pele for the Strategic Capabilities Office, and he now oversees the Army's Janus Program as a principal deputy assistant secretary. That continuity matters, because it means the people who built the transportable prototype are the same people writing the rules for the commercial fleet that is supposed to follow it. One bookkeeping change from this period is worth a line. In 2025 the Department of Defense was renamed the Department of War, and some current budget and contract documents now live on war.gov rather than defense.gov. For historical accuracy this article uses Department of Defense and DoD, the name the department carried through almost all of the events described here. ## Three piles of money that must not be mixed The temptation with a program like this is to add every number into one scary total, which would be wrong, because the figures cover different scopes and years. The first pile is budgeted appropriations. The Pentagon reported putting about $63 million into Project Pele in fiscal 2020 and about $70 million in fiscal 2021. These are research and development budget lines for the demonstration, and they should not be confused with any contract. The second pile is contract value. On March 9, 2020 the Strategic Capabilities Office made design phase awards totaling about $39.75 million, split among three vendors: about $13.5 million to BWX Technologies, about $11.95 million to Westinghouse Government Services, and about $14.3 million to X-energy. That was to draw up competing designs. Then, in June 2022, the office picked a winner and awarded BWX Technologies a cost type contract worth around $300 million, depending on options selected, to design, build, and deliver one full scale transportable prototype. The phrase "depending on options selected" is the tell. That $300 million is a ceiling to build a single unit, not a check that has been cashed, and it is separate from both the earlier design awards and the annual budget lines. The third pile is projected spending that does not exist yet. Army Secretary Dan Driscoll said in 2025 that he expects to spend "hundreds of millions" on the Janus effort over five years. Because Janus runs on the commercial ownership model, there is no firm lifecycle figure to cite, only an expectation. For a grounded sense of what base reactors actually cost across their whole life, the closest primary anchor is the last attempt. The 1950s era SM-1 reactor cost roughly $2 million to build and is estimated near $68 million to clean up, and decommissioning the related SM-1A was estimated at about $243 million. Those cleanup figures are the part of the ledger that never shows up in a groundbreaking ceremony, and they are the reason the whole spent versus projected distinction is not academic. ## The honest case against The strongest argument against this effort is that the Army already tried it and it went badly. From 1954 to about 1977 the Army Nuclear Power Program built and ran a series of base and portable reactors. On January 3, 1961 the SL-1 reactor suffered a prompt criticality steam explosion that killed all three of its operators, the first fatal reactor accident in United States history. The truck mounted ML-1 portable reactor, the closest ancestor of what Janus wants, never really worked. The program was wound down on cost grounds and left long, expensive cleanups that taxpayers are still paying for decades later. Independent safety experts make a technical case on top of that history. Edwin Lyman of the Union of Concerned Scientists argues that microreactors need far more uranium mined and enriched for each unit of electricity they produce, will likely cost more to fuel, and generate more waste per kilowatt hour than large plants. With no national repository open for spent fuel, putting reactors on many bases mostly spreads radioactive material across more sites rather than solving anything. A reactor on a base is also a target, one an adversary could try to turn into a radiological hazard. When the military airlifted a small reactor in a February 2026 demonstration, Lyman called it a "dog and pony show" that proved only that the armed forces can move heavy equipment, not that the reactors are feasible, economic, workable, or safe. The independent watchdog record aimed squarely at Pele and Janus is still thin, mostly because the effort is so new. The Government Accountability Office has published a science and technology spotlight on nuclear microreactors, but it raises the open questions on waste and regulation for the technology class as a whole, not for this program by name. That gap is worth watching. A program moving this fast toward a 2028 deadline has not yet faced the kind of sustained outside audit that usually catches the expensive mistakes. ## The honest case for The case in favor is real, and it is not just contractor sales talk. Base electricity today rides on a civilian grid that is aging, overloaded in places, and exposed to both weather and deliberate attack. A self contained reactor sitting inside the fence could give a critical mission baseload power that keeps running through a regional blackout or a targeted strike on a substation, resilience that is hard to price but easy to understand. The operational case is even more direct, and it is measured in lives. Moving fuel to forward positions gets people killed. The Army's 2009 Sustain the Mission study found roughly one casualty for every 24 fuel resupply convoys in Afghanistan, with fuel and water together making up 70 to 80 percent of the tonnage those convoys hauled. Every gallon of diesel that a reactor displaces at a forward base is a gallon that does not have to ride a truck down a road where someone can bury a bomb. Cutting forward fuel demand is not an abstraction. It is a way to keep soldiers alive. There is an industrial argument too. If Project Pele works, it will be the first Generation IV reactor to begin construction outside China, and the supply chain, the fuel fabrication, and the engineering talent it pulls together do not vanish when the demonstration ends. In a decade when advanced reactor leadership is itself treated as a national security question, being able to build one of these at all has value beyond the single machine at Idaho National Laboratory. ## Reading the ledger Both readings are true at once, and the honest thing is to let them sit next to each other. On one side is a technology with a lethal history, an unsolved waste problem, and a security profile that turns a power plant into a target, pushed toward a political deadline before any independent audit has caught up. On the other side is a genuine grid vulnerability, a body count from fuel convoys documented in the Army's own research, and a real chance to keep a piece of the advanced reactor industrial base inside the United States. The money, read carefully, is still modest by Pentagon standards: budget lines near $63 million in fiscal 2020 and $70 million in fiscal 2021, one design round of about $39.75 million in 2020, and a single build contract capped around $300 million in 2022, with everything past that a projection. The number that should worry a taxpayer is not the one on the contract. It is the one that never appears at the groundbreaking, the cleanup cost, which for the last generation of Army reactors ran from tens of millions to nearly a quarter billion dollars per site. Whether Pele and Janus land on the lifeline side of this series or the boondoggle side will not be decided by the reactor that runs in Idaho. It will be decided by who pays to shut the fleet down, and whether anyone counted that bill before the first one switched on. ## Related reading - [The couriers who move America's nuclear weapons](/blog/office-of-secure-transportation/): the other end of the nuclear enterprise, where moving radioactive material safely is the entire mission. - [The ICBM program that breached its own budget](/blog/sentinel-icbm-cost-breach/): a reminder of how fast a nuclear modernization price tag can run away once the meter starts. - [The classified material the government forgot how to make](/blog/fogbank-nuclear-material/): what happens when a nuclear program loses the know-how and has to pay to relearn it. - [The index of programs, lifelines, and boondoggles](/blog/public-money-programs-index/): the full catalog this article belongs to. - [The working ledgers](/blog/the-working-ledgers/): the running tally of spent, projected, and ceiling figures across the series. ## Fact-check notes and sources - The designation is real and was directed by executive order. Executive Order 14299, signed May 23, 2025, told the Secretary of Defense to name the Army as executive agent for installation and operational nuclear energy and set a September 30, 2028 deadline for an Army regulated reactor at a domestic installation. The designation itself is recorded in the official DoD Executive Agent registry, which is the authority for it; the "Section 318" statutory basis appears in coverage but was not independently confirmed against the live registry for this piece. [Executive Order 14299 (White House)](https://www.whitehouse.gov/presidential-actions/2025/05/deploying-advanced-nuclear-reactor-technologies-for-national-security/); [DoD Executive Agent registry](https://dod-executiveagent.osd.mil). - The load-bearing number is a ceiling, not an outlay. The June 2022 Strategic Capabilities Office award to BWX Technologies is a cost type contract worth around $300 million "depending on options selected" to design, build, and deliver one full scale transportable prototype. It is separate from the March 9, 2020 design phase awards, which totaled about $39.75 million across BWX Technologies, Westinghouse Government Services, and X-energy, and separate again from the fiscal 2020 (about $63 million) and fiscal 2021 (about $70 million) budget lines, which are reported appropriations that could not be independently confirmed from a primary source here. Project Pele is Strategic Capabilities Office led, uses TRISO fuel, is being assembled and tested at Idaho National Laboratory where ground broke on September 24, 2024, and is described as the first Generation IV reactor to begin construction outside China. [DOE Office of Nuclear Energy](https://www.energy.gov/ne/articles/department-defense-breaks-ground-project-pele-microreactor). - The safety and cost history of Army base reactors is documented. The Army Nuclear Power Program ran from 1954 to about 1977; the SL-1 reactor exploded on January 3, 1961, killing all three operators in the first fatal reactor accident in United States history, and the portable ML-1 never worked reliably. [The Conversation](https://theconversation.com/the-us-army-tried-portable-nuclear-power-at-remote-bases-60-years-ago-it-didnt-go-well-164138). - The fuel convoy casualty figure comes from the Army's own research. The 2009 Sustain the Mission study found roughly one casualty for every 24 fuel resupply convoys in Afghanistan, with fuel and water making up 70 to 80 percent of resupply tonnage. [Army Environmental Policy Institute, via Army Technology](https://www.army-technology.com/features/feature77200/). - The independent critique and the open regulatory questions are on the record. Edwin Lyman of the Union of Concerned Scientists argues military microreactors raise fuel cost, waste, and security problems, and called a February 2026 airlift demonstration a "dog and pony show" that did not answer whether the project is feasible, economic, workable, or safe. The Government Accountability Office's microreactor spotlight flags unresolved waste and regulatory questions for the technology class rather than for this program by name. [NPR](https://www.npr.org/2026/02/21/nx-s1-5721761/us-military-airlifts-small-reactor); [GAO-20-380SP](https://www.gao.gov/assets/gao-20-380sp.pdf). *This post is informational and journalistic, drawn from public records, and is not legal, financial, or policy advice; dollar figures are attributed to their fiscal year and, where noted, are contract ceilings or projections rather than money already spent.* --- Canonical HTML: https://jwatte.com/blog/army-microreactors-project-pele/ RSS: https://jwatte.com/feed.xml JSON Feed: https://jwatte.com/feed.json Hero image: https://jwatte.com/images/army-microreactors-project-pele.webp