Panthalassa reportedly seeks fresh capital for ocean AI compute

The Information reports that Panthalassa, the Portland-based startup building wave-powered floating data centers for AI inference at sea, is in talks for a fresh funding round at a higher valuation. Take the specifics as reported, not settled, since the new terms are single-sourced and have not been independently confirmed elsewhere yet.

What is confirmed is the trajectory. In May, Peter Thiel's Founders Fund led a $140 million Series B that pushed Panthalassa's valuation close to $1 billion, according to Fortune, with John Doerr, Marc Benioff's TIME Ventures, Max Levchin's SciFi Ventures and Hanwha Group joining. The pitch is that autonomous nodes moored offshore can pair onsite wave power with cold seawater cooling and run trained models without ever touching the mainland grid.

Why this matters if you are not building offshore compute: the binding constraints on AI infrastructure right now are power and water, and both get harder to secure onshore every quarter. If a floating node can deliver useful inference throughput at cost, it sidesteps the interconnect queue, the local permit fight, and the freshwater tradeoff at once. That is why investor appetite appears to be stepping up before any commercial hardware is in the water.

The honest caveats are large. The open ocean is a punishing environment for anything with a compute die inside it, and the company has not yet run its Ocean-3 pilot in the northern Pacific, which GeekWire reported is scheduled for this year, ahead of planned commercial deployments in 2027. What the reporting does not give you is the customer pipeline, the unit economics on inference at sea, or how nodes get serviced when a swell wins.

The thing to watch is that Ocean-3 deployment. If it survives a Pacific season and returns real inference numbers, offshore compute stops being a thought experiment and becomes a live procurement option for hyperscalers already rationing power onshore.