> ## Content Index > Fetch the complete content index at: https://nexi.fund/llms.txt > Use this file to discover other available public pages before exploring further. # Why the military, not utilities, is pulling small nuclear to market - URL: https://nexi.fund/nuclear-microreactors-military-power-2026/ - Published: 2026-08-27T13:00:58.000Z - Updated: 2026-08-27T13:00:58.000Z - Description: The US military is becoming small nuclear's first real customer. With Antares' $470M raise, the Air Force's three-vendor ANPI program, and Project Pele's 2028 demo, defence is pulling small reactors to market ahead of utilities. - Author: Nexi.fund Labs - Tags: Defence & Robotics, #mode-6, #hook-character, #track-F BWXT's engineers in Lynchburg are building a reactor that breaks down into four shipping containers and runs for three years without a fuel stop. The customer isn't a utility. It's the Defense Department. ๐Ÿ“Œ **Summary** For two decades, small nuclear struggled to find its first commercial buyer. The buyer has now shown up, and it wears a uniform. Through Project Pele, the Advanced Nuclear Power for Installations (ANPI) initiative, and the Army's Janus Program, the US military is acting as the lead customer for microreactors, pulling a stalled technology toward fielded power by 2028\. ## The buyer that wasn't supposed to exist Small modular reactors and their smaller cousins, microreactors, have been a venture story since the early 2010s. The pitch never changed: carbon-free power, factory-built, resilient. The blocker also never changed. No utility would be the first to buy a design with no operating history, and no financier would back production without a buyer. The stalemate is breaking from a direction the commercial thesis didn't predict. The Department of Defense (DOD) has become the anchor tenant. It is not buying one reactor. It is building three parallel programs that together function as a state-backed first market, complete with long-term power purchase agreements and a regulatory fast lane. The logic is operational, not ideological. Forward bases and remote installations run on diesel, and the fuel convoy is a target. A 2018 Army study tied roughly half of recent ground-transport casualties to the logistics chain that keeps bases powered. A reactor that arrives by truck or aircraft and runs for years without resupply removes both the convoy and the outage risk. Resilience, not carbon, is the product. ## Three programs, one bet Project Pele is the oldest. Run by the DOD's Strategic Capabilities Office and built by BWX Technologies, it is a 1.5 MW transportable microreactor designed to fit into four 20-foot containers and deploy by road or C-17 airlift. Core manufacturing began in 2025, the first TRISO fuel load reached Idaho National Laboratory (INL) in early 2026, and full testing is scheduled at INL with power targeted around 2028\. Rolls-Royce, Northrop Grumman, and Torch Technologies support the build. ANPI, the Advanced Nuclear Power for Installations initiative, is the Air Force's fixed-site track. In April 2026 the service paired three vendors with three bases: Antares Nuclear with Joint Base San Antonio, Radiant Industries with Buckley Space Force Base, and Westinghouse Government Services with Malmstrom Air Force Base. Each vendor owns, operates, and eventually decommissions its reactor under a long-term contract. The Air Force separately picked Oklo to demonstrate a liquid-metal fast reactor at Eielson Air Force Base in Alaska, with a notice of intent to award issued in 2026. The Army's Janus Program is the most ambitious. Launched with the Defense Innovation Unit, it selected nine domestic installations for microreactor deployment and targets first power by 2030\. Janus inherits Pele's lessons but points at permanent base power rather than transportable field units. A 2024 executive order, reinforced in 2025, sets September 30, 2028 as the deadline for an Army-regulated reactor operating at a domestic installation. As we previously wrote, these reactors burn High-Assay Low-Enriched Uranium (HALEU), the same fuel whose supply chain is the binding constraint on the whole build-out. The reactor bet and the fuel bet are the same bet, just at different points in the stack. ## What's actually moving in 2026 The capital milestone came in July. Antares Nuclear closed a $470 million Series C, the largest microreactor raise of the cycle, co-led by Paradigm and Caffeinated Capital, bringing total funding past $600 million. The raise followed a harder milestone: in June, Antares' Mark-0 demonstrator reached zero-power criticality at INL, the first privately developed non-light-water reactor to do so in the United States in over four decades. The company targets electricity production from its Mark-1 in 2027 and initial military deployments in 2028. Radiant's Kaleidos, a roughly 1 MW unit built for truck delivery and a 48-hour power-on, began its journey to INL for DOME testing in August 2026, the first new US reactor design trialed there in nearly fifty years. Westinghouse's eVinci and Oklo's Aurora advance on parallel tracks. None of these is generating electrons for a base yet. All of them now have a named customer, a dated contract, and a testing slot. $470M Lead microreactor raise, Jul 2026 #### Capital raised to field military microreactors The lead microreactor vendor closed a $470M Series C in July 2026, the largest raise this cycle, targeting first deployments in 2028\. ยท *The Defense Post, 2026* 1.5 MW Project Pele (BWXT) #### Transportable microreactor output BWXT's Pele prototype breaks into four shipping containers and runs three years without refueling. ยท *ANS Nuclear Newswire, 2025* 9 bases Army Janus Program #### Sites slated for microreactor power The Army selected nine installations for Janus deployments, with first power targeted by 2030\. ยท *US Army, 2026* 2028 Executive Order 14299 #### Deadline for first operational reactor A 2024 directive, reinforced in 2025, requires an Army-regulated reactor running at a domestic installation. ยท *The National Interest, 2026* ## The vendor field Five names now carry the program risk. They differ less on physics than on packaging and timeline. | Vendor / program | Output | Transportable | Target online | | ------------------------------- | ------------------ | ----------------------- | ----------------------- | | **BWXT: Project Pele** | 1.5 MW | โœ” 4 containers, airlift | \~2028 (INL test 2026) | | **Antares: R1 (ANPI)** | 100 kWeโ€“1 MWe | โœ” >6 years no refuel | 2028 | | **Radiant: Kaleidos (ANPI)** | \~1 MW | โœ” truck, 48h start | 2028 (INL DOME 2026) | | **Westinghouse: eVinci (ANPI)** | microreactor class | โœ” | Malmstrom, 2030 horizon | | **Oklo: Aurora (Eielson)** | fast reactor | โœ— fixed site | NOITA 2026, 30-yr PPA | Program pairings as of mid-2026; output figures from vendor and service disclosures. The pattern is deliberate. The DOD selected a portfolio rather than a winner, spreading technical and vendor risk across gas-cooled, heat-pipe, and fast-reactor designs. A procurement office builds an industrial base this way. A utility only buys the finished product. ## Where the promise meets the schedule The promised dates are aggressive and the history is not kind. Every prior microreactor timeline slipped. Pele was first discussed as a concept years before core manufacturing began. Janus talks about 2030, but the Army's own program leads have said early 2030s is the realistic fielding window once licensing and fuel readiness are counted. Critics inside the defense-policy community make a simpler point. A reactor on a base is a high-value, fixed target, and the resilience case weakens if the shielding and security footprint outweigh the diesel it replaces. Cost is the other open question. No microreactor has yet produced a kilowatt-hour at a price that beats the grid or a diesel set, and the first units will be the most expensive reactors ever built per watt. The 2028 deadline is also a licensing test. The Nuclear Regulatory Commission (NRC) is engaged in pre-application work with several vendors, but no microreactor has cleared a full combined operating license for a military installation. The executive order's September 2028 date is a policy marker, not a construction guarantee. ## Why this matters beyond the fence line The military buyer solves the problem that stalled small nuclear for a generation: it absorbs first-of-a-kind cost and regulatory risk that a utility never could. If the deployments hold to even a loose schedule, the vendors exit the program with licensed, demonstrated designs and a cost curve bent by volume. That is the real investment signal. The DOD is not the end market. It is the proof customer. The same reactors sized for a remote base are sized for data centers, Arctic communities, and industrial sites that the grid does not reach. Executives across the advanced-reactor sector have said openly that a successful military deployment is the trigger for private capital to follow. ๐Ÿ“Š **Key signals to track** First electrons: which vendor produces power from a microreactor before 2028\. The Mark-1 is the stated candidate. Fuel readiness: HALEU and TRISO output scaling, since every program above depends on the same constrained fuel line. NRC milestone: the first full license or permit for a military-installation microreactor. Spillover orders: the first non-defense power purchase agreement signed off the back of a Janus or ANPI deployment. ## The model that makes it bankable The procurement structure is as important as the hardware. Every ANPI and Janus reactor is contractor-owned and contractor-operated. The vendor builds, fuels, runs, and eventually decommissions the plant; the service pays a fixed tariff under a power purchase agreement that can run for decades. That shifts construction and operating risk off the public balance sheet and onto the developer's, which is exactly what a private nuclear builder needs to raise cheap capital. For vendors, the military contract is a revenue floor, not the ceiling. The vendors have all framed their defense work as the bridge to commercial orders. The same reactor that powers a remote base can power a hyperscale data center, and the data-center buyer is the one with both the capital and the urgency. Several advanced-reactor firms are already pitching defense-proven designs directly to compute operators chasing round-the-clock clean power. This is why the $470 million raises matter more than the headline suggests. They are not bets on a single base. They are bets on a financing template: prove it on a government site, then replicate it where the margin is higher. The defense program de-risks the first unit; the commercial market pays for the next hundred. The template is not unique to reactors. It mirrors how military demand pulled drones, composites, and satellite imaging into commercial markets decades earlier. The difference this time is scale and capital intensity. A microreactor costs more to certify than a drone does to fly, and the regulatory tail is longer. The programs underway in 2026 are effectively a government-funded certification engine for an entire reactor class. ## The bottom line for the thesis The defence-energy microreactor story is not a power story. It is a demand-creation story. The institution with the weakest cost sensitivity and the strongest resilience need has decided to be the first buyer, and it is doing so with contracts, deadlines, and a portfolio of vendors. The technology works in a lab. The open question is whether 2028 holds. Three programs, five vendors, one fuel bottleneck, and a licensing path that has never been walked. If even one deployment lands on schedule, the commercial market small nuclear has chased for twenty years finally has its reference customer. ## Sources [ First criticality for US microreactor under DOE programme Antares Nuclear completed a zero-power criticality test of its Mark-0 demonstrator at Idaho National Laboratory on June 4, 2026, the first privately developed non-light-water reactor to do so in the US in four decades. World Nuclear News ](https://www.world-nuclear-news.org/articles/first-criticality-for-us-microreactor-under-doe-programme?ref=nexi.fund) Primary trade source for the June 2026 criticality milestone and the Janus Program timeline. [ Antares to Fast-Track US Military Base Nuclear Microreactors With $470M Injection Reports Antares' $470 million Series C in July 2026, the largest microreactor raise of the cycle, aimed at first military deployments from 2028. The Defense Post ](https://thedefensepost.com/2026/08/03/us-nuclear-microreactors-base/?ref=nexi.fund) Trade-press confirmation of the capital event that anchors the 2026 inflection. [ Department of Air Force picks bidders for nuclear microreactors, assigns locations Details the ANPI vendor-to-base pairings: Antares at Joint Base San Antonio, Radiant at Buckley SFB, and Westinghouse at Malmstrom AFB. Breaking Defense ](https://breakingdefense.com/2026/04/department-of-air-force-picks-bidders-for-nuclear-microreactors-assigns-locations/?ref=nexi.fund) Maps the fixed-site vendor field and base assignments referenced in the comparison table.