Texas developer pitches recycled Navy nuclear reactors as power source for AI infrastructure – News

A Texas power developer is proposing to use repurposed retired U.S. Navy nuclear reactors from aircraft carriers and submarines to supply continuous power to large-scale artificial intelligence data centers.

The proposal comes from HGP Intelligent Energy LLC, which, according to filings with the U.S. Department of Energy, is asking for a loan guarantee under the federal government’s energy financing programs to help launch what it calls the “CoreHeld Project.”

The proposal would see two decommissioned naval reactors deployed at a dedicated data center campus near Oak Ridge National Laboratory in Tennessee, the home of the Manhattan Project and today the largest multi-program science and technology lab in the U.S.

The reactors HGP has its eyes on are naval pressurized water reactors that have been used to power U.S. aircraft carriers and submarines for decades. The reactors, built by companies including Westinghouse Electric Co. and General Electric Co., are designed for long operational lifetimes and have a record of safe use within the U.S. Navy.

The reactors would produce about 450 and 520 megawatts of baseload power, enough to support intensive AI training and inference workloads that require constant, predictable electricity.

HGP estimates the project would cost $1.8 billion to $2.1 billion, which works out to roughly $1 million to $4 million per megawatt. The figures, if accurate, work out significantly lower than the equivalent cost of building a new nuclear plant or a small modular reactor.

In theory and likely in practice, it shouldn’t be too difficult to reposition the reactors for civilian electricity generation, but the problem arises with regulations that were not designed with provisions to recycle naval nuclear reactors.

One core regulatory issue is that naval reactors typically use highly enriched fuel and are engineered as sealed systems, which means they do not align neatly with existing Nuclear Regulatory Commission licensing frameworks for commercial power plants.

HGP, though, is confident it can make the proposal happen. “We already know how to do this safely and at scale and we’re fortunate to have a solid base of investors and partners who share that vision,” HGP Chief Executive Gregory Forero said in a statement to Bloomberg.

The proposal reflects growing pressure on the U.S. power grid as AI workloads drive rapid expansion of hyperscale data centers. Cloud providers and infrastructure operators are increasingly exploring nuclear energy as a source of carbon-free, always-on power, with recent interest spanning traditional reactors, small modular reactors and long-term power purchase agreements tied to nuclear facilities.

Companies exploring and developing nuclear power solutions to power AI data centers are also gaining increasing attention from venture capital. In November, X-energy Reactor Co., a startup developing miniature nuclear reactors, raised $700 million in funding. On Dec. 18, Radiant Industries Inc., a startup developing portable, mass-produced nuclear microreactors, raised more than $300 million.

Image: News/Ideogram