Colocating Compute with Wave Energy

This is the key economic bet in Panthalassa, because it turns offshore power from a transmission problem into a compute product. Most ocean energy projects still need export cables, shore interconnection, and land based facilities before they can sell anything. Panthalassa instead plans to use wave power where it is generated, run chips on the same platform, and send back model outputs by network, which cuts out one of offshore energy’s hardest and most failure prone steps.

• The avoided step is not trivial. Offshore projects usually need subsea export cables and onshore grid connection, both of which add major capex, permitting work, and outage risk. Cable shortages and cable repair costs have become a real bottleneck even in much larger offshore wind markets.
  4 offshoremag 5 semafor
• There is a real precedent for putting compute in the ocean, but usually without local generation. Microsoft’s Project Natick showed submerged servers could run with low failure rates, and Nautilus built floating data center infrastructure cooled by water. Panthalassa combines those ideas with onboard power generation, which is the distinctive piece.
  2 microsoft 3 nautilusdt 6 microsoft
• That changes who pays and for what. Instead of selling electrons into a utility grid at commodity prices, Panthalassa can sell scarce AI inference capacity to buyers blocked by grid delays, cooling limits, or land siting fights. DOE notes data centers could reach up to 9% of U.S. electricity generation by 2030, which makes power constrained compute more valuable.
  1 sacra 7 energy

If this works, offshore energy developers will have a new template, skip the cable, keep the load local, and sell a higher value digital service instead of raw power. That would move wave energy out of the utility procurement lane and into the AI infrastructure lane, where speed to available capacity matters more than selling the cheapest kilowatt hour.