China has unveiled the first commercial offshore underwater AI data center, located off Shanghai’s coast, combining renewable energy and innovative cooling to dramatically enhance energy efficiency for AI and big data workloads.
- Nearly 2,000 servers powered by offshore wind energy and seawater cooling.
- Facility reports a highly efficient PUE rating of 1.15, reducing electricity use by over 20%.
- Eliminates freshwater usage and reduces land footprint by more than 90% compared to traditional data centers.
What happened
China officially launched the world’s first offshore underwater AI data center near Shanghai’s Lingang Special Area. The facility contains nearly 2,000 servers housed in pressure-resistant underwater modules situated beside offshore wind turbines. This project is the result of a partnership between the Lingang Special Area Administrative Committee, local investment entities, and technology firms including HiCloud Technology, China Telecom, and others.
Construction began mid-2025, completed by October, with commercial operations commencing in May 2026. The data center is positioned within Lingang’s offshore wind farm area and is directly connected to the renewable energy infrastructure. The project cost around ¥1.6 billion ($226 million). This phased development started with a smaller demonstration facility before scaling to a full 24 MW capacity capable of supporting AI training, big data processing, and 5G services.
Why it matters
This underwater data center introduces a novel approach to AI computing infrastructure by integrating renewable offshore wind energy and a seawater-based cooling system. Traditional data centers consume significant electricity on cooling and often depend heavily on freshwater resources. By using passive seawater cooling, the project slashes the energy spent on temperature regulation and eliminates freshwater consumption completely.
Energy efficiency is reflected in the facility’s Power Usage Effectiveness (PUE) rating of approximately 1.15, meaning the vast majority of power is used directly for computing rather than overhead. This is a significant improvement compared to many conventional data centers, which often exceed PUE values of 1.5 or 2.0. Such efficiency gains contribute to sustainability goals amid rising global demand for AI infrastructure.
What to watch next
The success of the Lingang underwater AI data center could set a precedent for future data center designs worldwide, especially in regions with access to offshore renewable energy resources. Observers will be interested in how scaling and operational reliability evolve over the coming months, especially under peak AI workloads and big data processing demands.
Additionally, the project may stimulate further innovation in green AI infrastructure, including expanded offshore deployments or integration with other renewable assets. Monitoring how well the facility maintains its low energy consumption metrics and adapts to increasing computational demands will be key indicators for the potential broader adoption of similar underwater data centers.