Washington-state startup Panthalassa is developing autonomous floating data centers powered by wave energy to support artificial intelligence workloads at sea.
The project addresses a critical infrastructure bottleneck as AI computing requires massive amounts of electricity that often exceed the capacity of land-based power grids. By moving data centers offshore, the company aims to reduce reliance on fossil fuels and bypass land-use constraints.
Investment in the venture has reached approximately $200 million [1], with reports indicating Peter Thiel has contributed $140 million [2] toward the effort. These funds support the development of the Ocean-3 prototype, which underwent testing throughout 2024 and 2025. The company is targeting commercial deployment for the early 2030s.
Panthalassa is currently operating wave-energy converters off the northern tip of Scotland in the North Atlantic [3]. These converters currently feed the local grid while the company prepares for larger-scale AI integration. Additional test sites are planned for the Pacific Ocean to evaluate performance across different maritime environments [4].
The push for offshore solutions comes as energy projections for the sector climb. AI data-center electricity demand could reach 945 terawatt-hours per year by 2030 [5]. This surge has pressured tech companies to find sustainable energy sources that do not compete with residential or industrial power needs on land.
Wave energy provides a consistent power source that can be harvested directly at the site of the computing hardware. This proximity minimizes transmission loss, a common issue with remote renewable energy farms, and allows the ocean to provide natural cooling for the high-heat AI servers.
“AI data-center electricity demand could reach 945 terawatt-hours per year by 2030”
The transition to offshore AI infrastructure represents a strategic shift in how Big Tech manages the physical requirements of generative AI. If successful, Panthalassa's model could decouple AI growth from terrestrial energy grid limitations, potentially accelerating the deployment of large-scale models while mitigating the environmental impact of land-based data center sprawl.
