{"id":12733,"date":"2026-07-06T08:00:09","date_gmt":"2026-07-06T12:00:09","guid":{"rendered":"https:\/\/www.autonocion.com\/us\/?p=12733"},"modified":"2026-07-06T06:40:48","modified_gmt":"2026-07-06T10:40:48","slug":"america-three-brand-new-nuclear-reactor","status":"publish","type":"post","link":"https:\/\/www.autonocion.com\/us\/america-three-brand-new-nuclear-reactor\/","title":{"rendered":"America just took three brand-new nuclear reactors critical in thirty days, a first for any country, machines the size of shipping containers, one of them built in 150 days on a single-digit-million budget with a core its founder drove to Idaho in the back of a Ford F-150"},"content":{"rendered":"

The AI power problem has become one of those things everyone in tech mentions in the same breath as “grid queue,” “seven-year interconnect wait,” and “we’ll figure it out.”<\/p>\n

Hyperscalers keep signing gigawatt deals with existing nuclear plants. Microsoft locked up Three Mile Island’s 835 MW under a 20-year power purchase agreement, and Amazon secured 1.92 GW from Talen’s Susquehanna plant. But those are old reactors being kept alive or switched back on, not new ones being built. New nuclear was always going to take a decade.<\/p>\n

Until last month, when the US quietly took three brand-new reactors critical in thirty days. Not gigawatt behemoths, but microreactors the size of a shipping container, small enough that one founder drove the core of his machine to Idaho Falls in the back of a Ford F-150.<\/p>\n

And the data center industry is already trying to figure out how to buy them.<\/p>\n

Three reactors, one month, one presidential deadline<\/h2>\n

The setup goes back to May 23, 2025, when Executive Order 14301 told the Department of Energy to stand up a Reactor Pilot Program and get at least three advanced reactors to criticality by July 4, 2026, the country’s 250th birthday.<\/p>\n

Criticality is the milestone: the moment the core sustains its own controlled chain reaction. It’s the difference between a nuclear machine and a very expensive metal sculpture.<\/p>\n

The DOE beat the deadline with days to spare. Antares Nuclear’s Mark-0 got there first on June 4 at Idaho National Laboratory, becoming the 53rd reactor ever built at the site since 1951 and, per the DOE, the first privately funded non-light-water design to go critical there in four decades.<\/p>\n

Valar Atomics’ Ward 250 followed on June 22, per the American Nuclear Society<\/a>, and not in Idaho: Valar runs its own test site at the San Rafael Energy Lab in Emery County, Utah. It was actually the startup’s second criticality, after an earlier core-only test at Los Alamos.<\/p>\n

Deployable Energy’s Unity completed the sweep at 11:55 PM on June 30, per the company’s announcement<\/a>, back at Idaho National Laboratory. The DOE says no country had ever taken three distinct advanced microreactor designs critical in a single month. Energy Secretary Chris Wright called it “a significant milestone on a timeline many thought was unachievable.”<\/p>\n

Each machine is a different bet on the physics. Antares’ Mark-0 is a sodium heat-pipe-cooled design running HALEU TRISO fuel, modeled on the fuel compacts BWXT built for the Army’s Project Pele. Valar’s Ward 250, roughly the size of a minivan, is a TRISO-fueled high-temperature gas reactor cooled with helium.<\/p>\n

Deployable took a different route entirely. Unity is water-moderated and helium-cooled, and runs on standard 4.95% enriched uranium dioxide, per POWER magazine<\/a>, with commercially available materials instead of HALEU or exotic components. Which is another way of saying: they built one out of parts you can actually source.<\/p>\n

One more wrinkle worth keeping straight. Antares and Valar ran under the Reactor Pilot Program; Unity was the first criticality under the DOE’s Nuclear Energy Launch Pad, the follow-on initiative. The DOE counts all three toward the presidential goal.<\/p>\n

\n
\n
New reactors critical in June<\/div>\n
3<\/div>\n
Antares (June 4), Valar (June 22), Deployable (June 30). A first for any country, per the DOE.<\/div>\n<\/div>\n
\n
Unity, kickoff to criticality<\/div>\n
~150 days<\/div>\n
On a “single-digit million” dollar budget, per Deployable’s CEO.<\/div>\n<\/div>\n
\n
Unity output<\/div>\n
1 MWe<\/div>\n
A nuclear battery designed to ship in a standard 20-foot container.<\/div>\n<\/div>\n
\n
PER NRC FILING<\/div>\n
Deployable’s LOI pipeline<\/div>\n
$10B+<\/div>\n
Letters of intent ranging from data centers to remote island power.<\/div>\n<\/div>\n<\/div>\n

These things are genuinely small<\/h2>\n

When people say “small modular reactor,” they usually mean something like GE Vernova’s BWRX-300: 300 megawatts, enough for roughly 300,000 homes, and requiring a 953-tonne concrete basemat lowered into a 35-meter shaft. That’s what Canada just did at Darlington<\/a>, and even that machine fits on two soccer fields. Grid-scale, in the SMR world, is still a serious piece of civil engineering.<\/p>\n

These microreactors are a different animal. Deployable, founded barely a year ago by Bobby Gallagher, a former Australian Army soldier who used to build offshore oil rigs, partnered with Texas A&M’s nuclear engineering department on its 1-megawatt container reactor, the Unity Nuclear Battery.<\/p>\n

To demonstrate portability, Forbes reports<\/a>, Gallagher drove the core to Idaho Falls in the back of a Ford F-150. One megawatt is small, power for maybe 800 homes or one modest server room. But it’s designed to be dropped in wherever you need it and left alone.<\/p>\n

Antares runs even smaller in raw output but broader in mission. The three-year-old startup has been testing at NASA’s Marshall Space Flight Center with the aim of providing 100 kilowatts to a future moon base, and it already has an order to deliver reactors to Joint Base San Antonio in 2027, per Forbes. Which is a strange sentence to write in 2026, but here we are.<\/p>\n

The portability claim isn’t theoretical either. In February, the Pentagon and the DOE airlifted a small reactor from California to Utah, a first for the country.<\/p>\n

The data center pitch, in plain math<\/h2>\n

Why is anyone in AI infrastructure watching a 1 MW reactor? Because one megawatt isn’t the point. The point is that if you can make one, you can make fifty, and stack them next to a compute campus without waiting years for a transmission line.<\/p>\n

The demand curve behind that logic keeps steepening. The International Energy Agency estimates data centers consumed about 415 TWh of electricity in 2024 and projects that figure will more than double to around 945 TWh by 2030, slightly more than Japan’s entire annual electricity consumption today.<\/p>\n

Deployable is already pointing at that market. In its letter of intent to the Nuclear Regulatory Commission, the company said it has achieved commercial traction with over $10 billion in LOIs, ranging from data centers to remote island community power, per the American Nuclear Society.<\/p>\n

And Valar didn’t wait long to make the connection literal. On July 1, it generated 100 kilowatts from its reactor, enough electricity to feed an Nvidia Blackwell processor, as Forbes put it, timed for the announcement of a new deal with Nvidia to build a 30-megawatt nuclear datacenter in Utah. Timed for the announcement. Nothing accidental about that.<\/p>\n

Criticality is not the same as commercial<\/h2>\n

Here’s where everyone selling a microreactor gets a little quieter. What these three machines achieved is a zero-power criticality: a low-power physics demonstration in which the core sustains a controlled chain reaction without producing meaningful heat or electricity, per POWER magazine. It proves the design works as physics. It doesn’t prove the design works as a business.<\/p>\n

People in the sector are being polite about this but firm. Emily Tucker, vice president on the energy team at advisory firm Capstone, cautioned that the criticality milestones matter as technology demonstrations but don’t signal that commercialization is near. Alison Hahn, former head of advanced reactors at the DOE and now at the Nuclear Energy Institute, noted the pilot program is explicitly for demonstration reactors.<\/p>\n

The commercial route runs through the NRC, a different beast than the DOE. A proposed rule would give advanced reactors a faster path to approval if they’ve already been reviewed by the DOE or the Defense Department, and Gallagher says Deployable is likely to apply for a commercial license later this year, once a new microreactor licensing rule is finalized, with the NRC review expected to take six to twelve months.<\/p>\n

Six to twelve months is fast for the NRC. It’s still not plug-it-in-next-quarter. Gallagher’s own stated target, per POWER, is commercial reactors by 2028.<\/p>\n

Canada took the other bet<\/h2>\n

The contrast with what’s happening at Darlington is worth sitting on for a second. Ontario Power Generation isn’t building a nuclear battery you can move around; it’s building a full grid asset.<\/p>\n

The basemat there weighs close to 953 metric tons, about 2.1 million pounds, or more than three fully loaded Airbus A380s. It measures 37 meters across, was welded together in one piece, and one of the world’s largest crawler cranes lowered it 35 meters below grade. The whole project runs CAD$20.9 billion, and none of the four planned units will generate a watt until the end of 2030.<\/p>\n

Same underlying idea: a smaller, factory-built reactor deployed faster than a conventional plant. Different scale entirely. Darlington’s BWRX-300 will produce 300 MW per unit, enough for a city’s worth of homes. Deployable’s Unity produces 1 MW, enough for a warehouse.<\/p>\n

The Canadian bet is that hyperscalers eventually want to be plugged into a proper grid backed by proper reactors. The American bet, at least for these three startups, is that hyperscalers don’t want to wait that long and will take power in whatever form arrives first.<\/p>\n

Where this actually goes<\/h2>\n

The next test isn’t another criticality announcement. It’s whether any of these companies can get an NRC commercial license, deliver a working reactor to a paying customer, and run it long enough to prove the economics.<\/p>\n

Unity went from project kickoff to criticality in roughly 150 days, on what Gallagher describes as a single-digit-million-dollar budget. A decade ago, both halves of that sentence would have sounded made up. But going from a zero-power test at a national lab to a revenue-generating machine at a hyperscaler campus is a much longer walk.<\/p>\n

Meanwhile the queue keeps growing. The DOE accepted 11 projects from 10 companies into the pilot effort back in August 2025, including Aalo Atomics, Deep Fission, Last Energy, Oklo, Radiant Industries, and Terrestrial Energy.<\/p>\n

Radiant is expected next with Kaleidos, another shipping-container reactor, and it intends to go further than zero power: full heat, then 1 MW of sustained output for 150 hours without operator intervention, per Forbes. Aalo just flew all 200 of its employees to Idaho Falls for its own attempt, an event its CEO describes as “a nuclear Burning Man.” Every one of these companies is talking to somebody in AI.<\/p>\n

So the honest read is this: America has three tiny reactors that work as physics, an executive order that hit its deadline, and a set of AI companies desperate enough for firm power that they’ll sign LOIs with startups that were empty parking lots a year ago. Canada has one very large slab of concrete and a plan that doesn’t produce electrons until 2030.<\/p>\n

Both bets can win. But only one of them fits in a Ford F-150.<\/p>\n

Image Credit: Deployable Energy – Axios<\/strong><\/em><\/p>\n","protected":false},"excerpt":{"rendered":"

Three US microreactors just hit criticality in a single month while Canada pours its first SMR basemat \u2014 and AI hyperscalers are already circling the shipping-container designs.<\/p>\n","protected":false},"author":8,"featured_media":12741,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"footnotes":""},"categories":[116],"tags":[],"class_list":["post-12733","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-energy","resize-featured-image"],"acf":[],"_links":{"self":[{"href":"https:\/\/www.autonocion.com\/us\/wp-json\/wp\/v2\/posts\/12733","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.autonocion.com\/us\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.autonocion.com\/us\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.autonocion.com\/us\/wp-json\/wp\/v2\/users\/8"}],"replies":[{"embeddable":true,"href":"https:\/\/www.autonocion.com\/us\/wp-json\/wp\/v2\/comments?post=12733"}],"version-history":[{"count":3,"href":"https:\/\/www.autonocion.com\/us\/wp-json\/wp\/v2\/posts\/12733\/revisions"}],"predecessor-version":[{"id":12742,"href":"https:\/\/www.autonocion.com\/us\/wp-json\/wp\/v2\/posts\/12733\/revisions\/12742"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.autonocion.com\/us\/wp-json\/wp\/v2\/media\/12741"}],"wp:attachment":[{"href":"https:\/\/www.autonocion.com\/us\/wp-json\/wp\/v2\/media?parent=12733"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.autonocion.com\/us\/wp-json\/wp\/v2\/categories?post=12733"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.autonocion.com\/us\/wp-json\/wp\/v2\/tags?post=12733"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}