Valar Atomics raises $1B Series B at $6B for Nvidia data centers

A nuclear startup going from a $2 billion valuation to $6 billion in a matter of months is the kind of price move that usually happens in software, not in a business that has to pour concrete and satisfy the NRC. That is what happened to Valar Atomics this week, according to The Next Web, which reports Sequoia Capital led a $1 billion Series B, with Atreides Management, Point72 and Snowpoint Ventures participating, plus a $200 million credit facility on top. Sequoia partner Shaun Maguire took a board seat.

The interesting part is what the money is for. CEO Isaiah Taylor framed it plainly in the piece: "One reactor can be built as a project. A fleet has to be manufactured." Valar's design is a helium-cooled microreactor, and the company says it hit a self-sustaining chain reaction in July and used the resulting electricity to power an Nvidia AI chip and host a website. That demo is the pitch. The next step, per the reporting, is a 30-megawatt nuclear-powered AI facility in Utah with Nvidia as the first commercial application of the partnership.

Why this matters if you are not in the nuclear business: the compute build-out is running headfirst into the grid. Hyperscalers keep announcing multi-gigawatt AI campuses that utility interconnect queues cannot serve on the timelines the AI roadmaps assume. A private, factory-built reactor sitting next to a data center campus is one of the few answers that does not involve waiting on transmission. If Valar can actually ship reactors like products rather than one-off projects, the power side of the AI stack gets a new default option, and Nvidia gets to say its chips run on carbon-free electrons it effectively owns.

The honest caveat is that a fission demonstration and a fleet of licensed commercial reactors are very different problems, and the reporting does not give you a commercial operation date for the Utah site, the price Nvidia is paying, or how the $1 billion splits between factory buildout and regulatory work. Valar is also, per the article, one of four companies that reached self-sustaining fission in 2026, so the AI-native power slot is contested. Take the $6 billion price as a bet on execution, not a settled market position. If it works, though, the buyers who benefit first are the hyperscalers who need dedicated power now, and the states willing to permit it fastest.