Panthalassa and Thiel: Revolutionary Marine Data Centers
Le brief IA que les pros lisent chaque soir
Les 7 actus IA du jour, décryptées en 5 min. Gratuit.
Inclus dès l'inscription : notre sélection des meilleurs guides & comparatifs IA.
Choisis ton rythme
Gratuit · Pas de spam · Désabonnement en 1 clic
Panthalassa and Wave Energy: A New Era for Data Centers
In response to the growing electricity demand for land-based data centers, Panthalassa proposes a bold solution: relocating these infrastructures offshore. The company aims to harness wave energy to power these data centers while using the ocean for cooling. This innovative project has caught the attention of Peter Thiel, who has invested $140 million to support this vision.
The Portland-based startup, founded in 2016, recently succeeded in raising $140 million in a Series B funding round. This round was led by Peter Thiel and included participation from influential figures such as John Doerr, Marc Benioff via TIME Ventures, Max Levchin from SciFi Ventures, as well as companies like Hanwha, Fortescue Ventures, and Super Micro Computer. With this funding, Panthalassa has now accumulated a total of $210 million to realize its ambitious project.
The Concept of Wave Energy Plants
Panthalassa is betting on an innovative concept: an autonomous 85-meter steel node designed to float in deep waters. This device captures wave energy to produce the electricity needed to operate the artificial intelligence chips it houses. The results from the calculations performed by these chips are then transmitted to the mainland via low Earth orbit satellites. Seawater plays a crucial role in cooling the installations, thus eliminating the need for cables, anchors, or fuels.
This approach addresses two of the main challenges faced by land-based data centers: limited access to the electrical grid and managing the heat generated by servers. Garth Sheldon-Coulson, co-founder and CEO of Panthalassa, emphasizes that three energy sources on Earth have the potential to provide tens of terawatts: solar, nuclear, and ocean. Between 2021 and 2024, Panthalassa tested several generations of prototypes, including Ocean-1, Ocean-2, and Wavehopper, in real-world conditions. The Ocean-3 nodes are expected to be deployed in the North Pacific in 2026 to validate AI inference at sea, with commercial deployments planned for 2027.
A Project Among Other Attempts
Panthalassa is not the first to explore the idea of offshore data centers. In 2018, Microsoft submerged a container containing 864 servers off the coast of Scotland as part of Project Natick. After two years, the results were promising, with a failure rate eight times lower than that of land-based data centers. However, Microsoft did not pursue the industrialization of this concept.
Meanwhile, China is currently deploying an underwater data center composed of 100 modules near Hainan Island. In Japan, shipping company MOL and Hitachi are collaborating to transform cargo ships into floating data centers, with a launch planned for 2027. In the United States, startup Starcloud raised $170 million in March to develop orbital data centers.
Panthalassa stands out for its ability to design Ocean-3 nodes capable of autonomously navigating high-energy density areas, without reliance on a home port or terrestrial network.
Challenges to Overcome
Despite the excitement surrounding these innovations, several technical challenges remain to ensure the long-term viability of these installations. Among the obstacles are corrosion from seawater, satellite connectivity in adverse weather conditions, and maintaining the installations without human intervention for extended periods.
In France, pressure on the electrical grid is already noticeable. During a meeting in Bercy in January 2025, 35 data center projects were identified, representing a total capacity of 28.6 GW. Alternative solutions, whether marine or orbital, are no longer mere futuristic concepts but concrete options that still require technical validations.
Brief IA — L'actualité IA en français
L'essentiel de l'actualité de l'intelligence artificielle, décrypté et expliqué chaque jour.