As terrestrial power grids struggle to keep up with the soaring energy and cooling demands of artificial intelligence, Portland-based ocean technology startup Panthalassa has closed a $140 million Series B funding round to take AI compute entirely off the grid and into the open ocean.
Led by billionaire investor Peter Thiel and joined by high-profile backers including John Doerr, Marc Benioff's TIME Ventures, Max Levchin's SciFi Ventures, Super Micro Computer, and returning funds like Founders Fund and Lowercarbon Capital, the financing propels Panthalassa to near a $1 billion valuation. The round signals growing investor conviction that the solution to AI's infrastructure crisis may not be found on land at all.
Ocean-3 Nodes | Autonomous Floating AI Factories
Panthalassa's core innovation abandons traditional terrestrial data centers entirely. The public benefit corporation builds autonomous, self-propelled, floating steel platforms called Ocean-3 nodes that operate far off the coastline in the open ocean, requiring no land, no anchors, and no tethered power cables back to the mainland.
Each node functions as a self-contained AI compute factory. As ocean waves heave the floating structure up and down, water is forced into an onboard pressurized reservoir that drives a turbine, generating continuous, zero-emission electricity. Instead of sending the generated power back to land, the node uses the electricity immediately to run high-density AI inference processors housed directly within the structure.
The surrounding ocean water acts as a massive, natural heat sink, providing free ambient thermal cooling to the high-performance computing racks. This solves one of land-based data centers' most expensive operational headaches: cooling infrastructure can account for up to 40 percent of a traditional facility's total energy consumption. Processed AI inference results are then transmitted to land users via low-Earth-orbit (LEO) satellite networks, completing a fully offshore compute pipeline.
Bypassing the Land-Based Grid Crisis
Panthalassa's offshore deployment model targets the massive infrastructure bottlenecks currently choking terrestrial AI data centers. Severe regional grid congestion has made securing power allocations for new data center campuses a multi-year negotiation in major markets like Northern Virginia, Silicon Valley, and Dublin. Skyrocketing capacity clearing prices, land zoning pushback from local communities, and water-scarcity concerns in drought-prone regions have further constrained the buildout.
"There are three sources of energy on the planet with tens of terawatts of new capacity potential: solar, nuclear, and the open ocean," said Garth Sheldon-Coulson, co-founder and CEO of Panthalassa. "We've built a technology platform that operates in the planet's most energy-dense wave regions, far from shore, and turns that resource into reliable clean power."
The open ocean represents a largely untapped energy reservoir. Wave energy density in prime regions like the North Pacific can reach 30 to 70 kilowatts per meter of wave front, and unlike solar or wind, wave patterns are highly predictable days in advance. For AI inference workloads that can tolerate the latency of satellite uplinks, the tradeoff of slightly higher network round-trip time for effectively unlimited, zero-carbon power and cooling is increasingly compelling.
Deployment Roadmap | From Prototype to Commercial Fleet
The $140 million infusion will be used to complete Panthalassa's pilot manufacturing facility near Portland, Oregon, and scale up production for its fleet of Ocean-3 series nodes. The company has been methodically iterating on its platform: Ocean-1 prototypes were deployed in 2021, followed by the more capable Ocean-2 nodes in 2024, each generation validating the core thesis that wave-powered offshore compute is not just theoretically possible but commercially viable.
Initial offshore deployments of the Ocean-3 nodes are scheduled to begin in the Northern Pacific Ocean in 2026. Building on the data and operational experience from earlier prototypes, Panthalassa is targeting full commercial fleet availability and compute delivery by 2027. The manufacturing facility near Portland positions the company close to both deep-water port access for deployment and the Pacific Northwest's concentrated engineering talent pool.
The investor syndicate itself hints at the operational strategy. Super Micro Computer, a major server and storage hardware manufacturer, brings direct supply chain integration for the high-density compute racks that will populate each Ocean-3 node. Founders Fund and Lowercarbon Capital, both returning investors, signal continued confidence from the earliest institutional backers who have watched the technology progress from concept through multiple ocean-tested prototype generations.
Panthalassa's approach also sidesteps the growing political friction around data center construction. In 2025 and 2026 alone, proposed data center campuses in Virginia, Ireland, and the Netherlands faced organized community opposition over noise pollution, aesthetic blight, and competition for grid power that drives up residential electricity rates. An offshore node operating beyond the horizon eliminates every one of these objections. There is no neighbor to complain about generator hum, no farmland to pave over, and no local grid to strain.
The company's public benefit corporation structure is also notable. Unlike a traditional Delaware C-corp, a PBC is legally required to balance shareholder returns with a stated public benefit mission. For Panthalassa, that mission is delivering zero-carbon compute infrastructure that does not compete with communities for land, water, or power. The structure may also make the company more resilient to acquisition pressure from hyperscalers that might prefer to bury the technology rather than see it commoditize the data center market.