Antares raises $470M Series C to power US military bases

A privately built nuclear reactor going critical in Idaho is one milestone. A $470 million growth round priced against it, with the Pentagon lined up as first customer, is a different signal, and that is what closed this month. Bloomberg reported that Antares raised $370 million in equity and $100 million in debt in a Series C co-led by Paradigm and Caffeinated Capital, with Point72 Ventures, Shine Capital and Industrious Ventures also participating.

The context that makes the round pricable is the criticality event on June 4, 2026. Antares' Mark-0 microreactor went critical at Idaho National Laboratory at around 12:30 MDT, the first test reactor to go critical under the Department of Energy's Reactor Pilot Program. That gave the company something no other startup in the microreactor space had, an actual physical demonstration to point at rather than a design deck. CEO and co-founder Jordan Bramble now has capital to march from that demonstration toward a first electricity-producing reactor in 2027 and initial deployments to US military installations in 2028.

Why investors of Paradigm's profile write this cheque is where the story stops being about nuclear and starts being about compute. Behind-the-meter power, generation sited next to the load rather than delivered through a utility grid, is the constraint AI infrastructure buildouts keep running into and cannot solve on the timelines they want. A microreactor that can be manufactured in a factory and dropped onto a base, or later a data center campus, is the cleanest answer anyone has fielded. The military is the natural first customer because it has both the mission need and the licensing latitude to move faster than a commercial utility.

The honest caveat is what the reporting does not settle. There is no per-unit megawatt figure disclosed for the deployments, no named base as the pilot site, and no public procurement contract behind the 2028 target. Nuclear timelines slip, HALEU fuel supply is tight, and a Series C priced on a zero-power criticality event is exposed if Mark-0 stumbles when it has to produce real electricity next year. Take the 2028 date as the company's plan, not a settled fact.

If Antares clears those hurdles, the more interesting story is the second-order one. A working DOE pilot template, a defense customer that de-risks the technology, and a factory-built form factor is exactly the shape of thing hyperscalers will want quoted the moment military deployment ships.