Wave-powered sea data centers aim to curb AI energy use

Panthalassa is betting that the next frontier for AI infrastructure sits in the middle of the ocean, where wave motion would generate electricity and seawater would handle cooling. The company calls itself a public-benefit corporation building a planetary-scale energy platform, and says its Ocean-2 wave-energy converter and newer Ocean-3 concept are meant to be fast to deploy, scalable, ultra low-cost and zero emissions. CBS News reported that Ocean-3 has no anchor and no cables.

That pitch arrives as the energy bill for artificial intelligence keeps climbing. The International Energy Agency estimates data centers used about 415 terawatt-hours of electricity in 2024, roughly 1.5% of global electricity consumption, and says electricity supply to data centers could rise to more than 1,000 TWh in 2030 and 1,300 TWh in 2035 in its base case. The IEA also says investment in data centers nearly doubled since 2022 and reached half a trillion dollars in 2024, underscoring how quickly capital is flowing into a sector whose power needs are still expanding.

The U.S. picture is just as stark. The Energy Information Administration estimates commercial computing accounted for 8% of commercial-sector electricity consumption in 2024 and will rise to 20% by 2050. It says U.S. electricity demand has been rising steadily since 2020, with data centers driving the increase. That makes any new power source, whether grid-connected or offshore, part of a much larger fight over where the country will find enough electricity to keep the digital economy running.

Panthalassa’s case is that ocean-based systems could sidestep some of the constraints now dogging land-based projects. The company argues that putting compute on offshore energy nodes could avoid some grid bottlenecks while using ocean water for cooling, a major appeal as tech companies, policymakers and investors search for alternatives to land-bound power buildouts. Garth Sheldon-Coulson and Brian Moffat are among the names attached to the project, which has drawn attention in the broader conversation around AI infrastructure, climate claims and deployment speed.

Still, the scale problem remains hard to ignore. The IEA says data centers are a small share of global electricity use overall, but their local impacts are far more pronounced, and it warns that supply-chain bottlenecks and grid-connection delays are already straining efforts to add capacity. A wave-powered platform may trim emissions and reduce pressure on terrestrial grids, but it does not make the underlying demand disappear. For AI’s energy boom, the real test is whether ocean power can move from a glossy workaround to a system that can actually be built, maintained and scaled fast enough to matter.