Atomarine raises $19m for nuclear-powered floating data centers

By Marcus Bennett

Atomarine has raised $19 million in seed funding. The round backs a plan to put modular data centers on ships. The plan calls for small modular nuclear reactors to provide power in a later phase.

The startup is based in Cambridge, Massachusetts. Y Combinator led the investor group. Pantera Capital, Soma Capital, Seven Seven Six, Liquid 2 Ventures, Audacious Ventures and E14 Fund also contributed.

MIT students Dimitris Koutentakis and Emile Germonpré founded the company. Its proposed system places SMRs on ships. The company intends to use them to power floating platforms carrying modular data centers. Atomarine estimates that 150 MW could support two platforms.

That figure describes the intended nuclear configuration. It is not a measure of present output. No reactor from a commercial SMR vendor is yet available for the system. Atomarine intends to begin with natural gas generators instead.

What does the $19 million change for Atomarine?

The immediate effect is recruitment. Atomarine is hiring engineers with specialties in “naval architecture, nuclear, or data center design.”

Koutentakis said the company has built a functional proof of concept. He said the company was now working toward its next major milestone.

The funding does not remove the commercial and regulatory constraints attached to nuclear reactors. The initial power plan calls for natural gas generators mounted on the ships. SMRs belong to a later phase.

The offshore siting argument

Koutentakis wrote on LinkedIn after the funding announcement:

“By moving offshore, we do not compete with communities for water, power or land, and we do not drive up their electricity bills or rent. This allows America to stay competitive in the AI race without putting the burden on local communities, while fighting climate change by scaling AI on nuclear instead of fossil fuels.”

His argument separates the data-center build-out from local infrastructure demands. Offshore power and computing capacity would sit outside host communities, under the company's stated design.

The company also frames nuclear power as a way to expand AI computing without fossil fuels. That rationale applies to the later reactor stage. The initial gas-generator plan differs from the long-term fuel proposition.

The reactor constraint

Atomarine faces a basic availability problem. Small modular reactors have not reached the commercial market. None of the vendors working in the sector has released a commercial model. None has obtained regulatory approval in the United States or any other market.

Speaking to DCD earlier this year, Koutentakis said:

“We’ve been talking to some potential partners about getting access [to an SMR]. But with the minor engineering changes that would be required and the regulatory timelines, we’re targeting a first deployment in 2031 or 2032.”

That statement sets the first SMR deployment target for 2031 or 2032. Reactor access, engineering modifications and regulatory timing remain dependencies. Atomarine has discussed potential partners, while its nuclear timeline remains conditional.

The intended scale

Atomarine's target of 150 MW could power two floating data-center platforms. The figure applies to its proposed nuclear ships, not current operations.

Atomarine is one of numerous companies seeking to place AI compute at sea. The effort spans the United States and other markets. DCD examined this emerging sector in depth in issue 62 of its magazine. The seed round gives Atomarine capital for its next phase. That phase combines a gas-first ship approach with a stated path to nuclear power.

Near-term execution now matters. The test is whether Atomarine can turn its proof of concept into a gas-powered ship. It must also recruit the required engineers. It still faces a 2031 or 2032 reactor target.