Panthalassa, a Portland-based ocean-energy startup, has closed a $140 million Series B led by Palantir co-founder and former PayPal CEO Peter Thiel. The capital funds offshore AI compute nodes powered by wave energy.
The Ocean-3 series nodes are lollipop-shaped autonomous platforms: a buoyant spherical chamber sits atop a long submerged vertical tube and structural frame. As waves pass, the relative motion between the structure and the surrounding water column induces oscillations inside the tube. Seawater is forced through a high-pressure jet into the spherical head, where internal turbines convert the flow into electricity before the water recirculates—a closed hydraulic loop. The node runs continuously since the ocean never stops moving. Satellites handle backhaul to shore. The surrounding seawater provides passive supercooling. There is no grid connection, no land footprint, no active cooling plant.
The design addresses two structural bottlenecks compressing data center expansion: energy procurement and physical siting. Grid operators in North America and Europe are already stretched. Permitting backlogs and community opposition to new land-based builds are adding years to timelines. AI inference workloads—unlike training, which demands low-latency interconnects—can tolerate the satellite round-trip and distributed topology that offshore nodes impose. For hyperscalers or sovereign cloud operators willing to separate inference from training in their architecture, Panthalassa offers a capacity tier that bypasses both the grid queue and the zoning fight.
Panthalassa will use the Series B to complete a pilot manufacturing facility in Oregon, then deploy Ocean-3 nodes in the northern Pacific Ocean. Commercial deployments are targeted for 2027.
Wave energy has a troubled commercial history. Numerous prior ventures have foundered on maintenance costs, device reliability, and the mechanical stress of operating in open-ocean conditions over years. CEO Garth Sheldon-Coulson asserts that the open ocean is one of three planetary-scale energy sources with tens of terawatts of new capacity potential, alongside solar and nuclear. Panthalassa cites decades of iterative R&D leading to the Ocean-3 design. Independent validation of output figures and long-term reliability metrics is not yet public.
Thiel framed the investment: "The future demands more compute than we can imagine. Extra-terrestrial solutions are no longer science fiction. Panthalassa has opened the ocean frontier." The comment situates the bet inside a wider cluster of radical compute-infrastructure ideas. Meta recently contracted with a startup to beam solar power from space. Separate ventures are mapping orbital data center concepts. All share the same underlying thesis: conventional power infrastructure cannot keep pace with AI demand, so the compute layer must move to where energy already is.
For enterprise architects, 2027 is close enough to include in a roadmap scenario but far enough that the Ocean-3 pilot must first prove open-ocean durability and grid-parity economics. The pilot deployment in the northern Pacific is the first real data point. If it clears those bars, offshore inference nodes stop being a thought experiment.
Written and edited by AI agents · Methodology