Panthalassa is going after the constraint that increasingly limits AI: not chips, but power and somewhere to put them. It was founded in 2016 by Garth Sheldon-Coulson, a former AI and energy researcher at the hedge fund Bridgewater, and it has spent close to a decade building floating steel structures that turn ocean waves into electricity. Those same structures now run AI computing at sea.
What it does. Each of Panthalassa’s nodes is a long steel structure, most of it below the surface, where the motion of the waves drives a turbine and generates electricity that powers AI chips in a sealed, seawater-cooled container onboard. The results are sent back by low-earth-orbit satellite, with no cable to shore and no connection to the grid. The nodes are mass-produced from plate steel in coastal factories and deployed far out in the parts of the ocean where waves are strongest. The cold seawater handles cooling for free — one of the hardest and most expensive problems in a normal data center — and the plan is to sell computing capacity rather than electricity.
Where it stands. This is still early. Panthalassa has put earlier prototypes — named Ocean-1, Ocean-2, and Wavehopper — through sea trials in 2021 and 2024. It plans to deploy its first pilot series, Ocean-3, in the northern Pacific in 2026, with the first commercial systems targeted for 2027. The recent funding is meant to finish a pilot manufacturing facility near Portland, Oregon, and get those first nodes in the water.
Funding. Panthalassa has raised about $210 million to date. Its most recent round, a $140 million Series B in May 2026, led by Peter Thiel, valued the company close to $1 billion. Its investors include Founders Fund, John Doerr, Marc Benioff’s TIME Ventures, Lowercarbon Capital, Fortescue Ventures, Hanwha, and Super Micro Computer.
Worth keeping in mind. This is a hardware company doing something difficult in a hard place, and it is earlier than most names we cover. Commercial deployments are not planned until 2027, so the technology still has to prove itself at scale and survive the open ocean for years at a time. Building factories and fleets is slow and expensive. Others are working on floating and ocean-based computing too. And there is no sign of an IPO, so this is a patient, long-term position rather than one with near-term liquidity.