[ Funding
](https://www.unite.ai/series/funding/)
[Add Unite.AI to your preferred sources on Google](https://www.google.com/preferences/source?q=unite.ai)
Valar Atomics has closed a $1 billion Series B led by Sequoia Capital, money the three-year-old nuclear company says will take it from running a single test reactor in Utah to building small reactors on a production line. Valar announced the round on August 3, 2026, along with a separate $200 million credit facility, and said Sequoia partner Shaun Maguire is joining its board.
The announcement names nine other firms in the equity round, among them Valor Equity Partners and Conviction, and describes the group as a mix of new and longtime investors. The debt is led by Erebor, a nationally chartered bank, with J.P. Morgan, Crescent Cove and Hercules Capital (HCXY ) alongside it.
Founder and chief executive Isaiah Taylor framed the raise as a change in what the company does rather than what it designs. The financing, he wrote, lets Valar move “from demonstrating the operability of an integrated reactor system to producing fleets of them en masse.”
What the capital is meant to buy #
The plan Taylor described is vertical integration that runs down to the fuel. Rather than buy fuel from outside suppliers, Valar says it will produce its own at fuel labs built on the same sites as the reactors that consume it. The company was selected for the Energy Department’s advanced fuel-line pilot as well as its reactor pilot, and it treats reactor and fuel as one manufacturing problem.
The second half of the argument is a learning curve. Valar contends that each reactor it operates makes the next one cheaper and faster to build, with Ward 250 already producing the engineering and operating data behind the units that follow. Its own cadence is the evidence it offers: the NOVA core took two years to finish, and Ward 250 took seven months to reach criticality. The company expects that interval to keep compressing until it is turning out tens, then hundreds, of reactors a year.
That production logic is what the company’s gigasite model rests on: clusters of high-temperature reactors built alongside the factory that makes them, selling electricity, industrial heat and hydrogen to customers on the same ground. The reactors run on TRISO fuel and helium coolant, a combination Valar says operates at far higher temperatures than a conventional plant and puts no draw on local water.
From a cold core to a live one #
Ward 250 reached self-sustaining criticality on June 18, 2026. Valar says that made it the first company to take a reactor critical outside a national laboratory.
The step before it was quieter. On November 17, 2025, Valar’s NOVA core went critical at zero power at Los Alamos National Laboratory’s criticality facility at the Nevada National Security Site, validating the physics of the graphite-moderated, TRISO-fueled core design that Ward 250 would use. A cold-critical run confirms the core behaves as modeled; it produces no usable power.
Ward 250 itself was built under the Energy Department’s Reactor Pilot Program, which set out to get at least three advanced reactor designs critical outside the national labs by July 4, 2026, and which is structured as a demonstration route meant to speed the commercial licensing that comes after. About a week after criticality, Valar ran current from Ward 250 into an Nvidia Blackwell processor, and the two companies said they would work together on a waterless 30-megawatt AI factory.
What it changes for data-center power #
The appeal to compute buyers is siting. A helium-cooled reactor placed behind the meter takes a data center out of the interconnection queue and out of the local water fight at the same time, two of the constraints that now set the schedule for large AI campuses. Every other route is a negotiation with the grid: utilities such as Duke are recasting data-center load as a benefit to other ratepayers, and grid-scale battery storage is going in next to data-center sites in Ohio to smooth what the wires can deliver.
Scale is the honest measure of where this sits. The Nvidia (NVDA ) collaboration targets 30 megawatts, roughly one AI building, at a moment when a single Texas AI campus is being provisioned at 600 megawatts. What the round buys is the capacity to make that 30 megawatts repeat. Valar has never pitched one large plant; it has pitched many identical small ones, each cheaper than the last, which only works if the factory works.
That is also the difference between this and the rest of the industry’s move toward nuclear power, which has largely meant buying output from plants somebody else owns and operates. A reactor a customer can have sited on its own land is a different product. The next thing Valar has to build is the second one, and then the line that stamps out the rest.