Panthalassa Raises $140M to Build Wave-Powered AI Data Centers at Sea
Oregon startup Panthalassa closed a $140 million Series B led by Peter Thiel to build floating, autonomous ocean platforms that generate electricity from wave energy and use seawater cooling to run AI inference workloads. The company plans to deploy its Ocean-3 pilot nodes in the northern Pacific by end of 2026, with commercial operations in 2027, aiming to deliver compute power at as little as $0.02 per kWh.
When engineers talk about the bottlenecks constraining AI’s expansion, two problems dominate every conversation: where to put the power and where to get it. Oregon startup Panthalassa thinks both answers are the same place — the ocean.
On Monday, the company announced a $140 million Series B led by Peter Thiel to build floating, autonomous data center platforms powered entirely by wave energy. If the bet pays off, it could represent the most radical rethinking of AI infrastructure since hyperscalers started building their own chips.
The Problem Panthalassa Is Solving
AI data center power demand has exploded to 29.6 gigawatts globally — roughly the peak electricity consumption of the entire state of New York — and the buildout shows no signs of slowing. Hyperscalers have pledged a combined $725 billion in capital expenditure for AI infrastructure through 2026, with data centers consuming an ever-larger share.
On land, that translates to fierce competition for sites near reliable power grids and water sources. Proposed data centers in Arizona and Nevada have triggered community backlash over freshwater cooling demands in drought-prone regions. Permitting, grid interconnection queues, and land-use disputes routinely add years to construction timelines.
Panthalassa’s answer is to leave land entirely. Each Ocean-3 node — roughly 85 meters in diameter — is a self-contained floating platform that generates its own electricity from wave motion and cools its servers with surrounding seawater.
How the Technology Works
The wave energy conversion mechanism is elegant in its simplicity. Each node consists of a buoyant spherical head connected to a long, submerged vertical tube and structural frame. As ocean swells pass, the node bobs up and down, driving oscillations within the internal tube. That oscillating water column passes through turbines, generating electricity without any combustion or chemical process.
The platform carries no anchors or seabed cables. It operates fully autonomously, using onboard AI systems to manage power generation, thermal regulation, and equipment health. Compute results — running AI inference workloads on the onboard servers — are transmitted back to clients via satellite uplink rather than a direct fiber connection.
The cooling system requires no freshwater infrastructure at all. Cold deep-ocean water circulates through the server racks, providing a heat sink that makes the costly chillers and cooling towers of land-based data centers unnecessary.
At scale, the company projects power generation costs as low as $0.02 per kilowatt-hour — a figure that would undercut most grid-connected commercial tariffs and virtually eliminate the energy cost component that dominates the total cost of AI inference at scale.
Who’s Betting on It
The $140 million Series B drew a roster of high-profile names alongside Thiel’s lead check. John Doerr, the legendary Kleiner Perkins partner whose green-energy investments helped establish the cleantech VC category, joined the round. Marc Benioff’s TIME Ventures and Max Levchin’s SciFi Ventures participated alongside hardware manufacturers Super Micro Computer, industrial conglomerate Hanwha Group, and Fortescue Ventures, Andrew Forrest’s clean-energy vehicle.
Dylan Field, co-founder of Figma, also invested individually, as did Planetary VC, Portland Seed Fund, and the Intrepid Oregon Fund, reflecting the company’s deep roots in the Pacific Northwest startup ecosystem.
The breadth of the investor syndicate is notable. It spans traditional tech venture (Doerr, Thiel), enterprise software (Benioff), fintech (Levchin), hardware manufacturing (Supermicro), and industrial decarbonization (Fortescue) — a coalition that mirrors the multi-sector nature of the problem Panthalassa is trying to solve.
2026 Roadmap
Proceeds from the round will fund two near-term priorities. The first is completing Panthalassa’s pilot manufacturing facility near Portland, Oregon, where the Ocean-3 nodes will be assembled and tested before ocean deployment. The second is executing the first open-ocean pilot, planned for the northern Pacific later in 2026.
The pilot is designed to demonstrate AI inference capabilities at sea and refine the manufacturing process ahead of commercial-scale deployments targeted for 2027. Panthalassa intends to operate without teleoperation from shore — autonomy isn’t just a selling point, it’s a core constraint of the offshore environment where human intervention is impractical and expensive.
A New Category of Infrastructure
Panthalassa isn’t the first company to imagine moving computing offshore. Microsoft’s Project Natick submerged a data center pod on the Scottish seafloor in 2018 and retrieved it in 2020 with encouraging reliability results, though the project was quietly shelved. The difference with Panthalassa is the combination of self-generated renewable power and satellite-based connectivity, which eliminates the two dependencies — grid power and undersea cables — that made prior concepts impractical at scale.
The addressable opportunity is vast. Oceanic regions cover 71% of Earth’s surface, and the wave energy resource in coastal and open-ocean zones is estimated at orders of magnitude more than current global electricity consumption. The limiting factor has always been cost-effective conversion — a problem that large-scale AI computing economics, which price every kilowatt-hour intensely, may finally justify solving.
For Thiel, the investment fits a pattern of bets on infrastructure that challenges incumbent assumptions. For the hyperscalers watching their land-based permitting timelines stretch toward 2030, a sea-based alternative that can be manufactured, deployed, and operational within a year may soon look less exotic than it does today.
“The data center industry has optimized relentlessly for cost and reliability within a fixed set of constraints,” said one investor in the round who requested anonymity. “Panthalassa is questioning whether those constraints are actually fixed.”
The Ocean-3 pilot later this year will be the first real-world test of that question.