China has activated what it describes as the world’s first commercial-scale wind-powered underwater data center, adding a new dimension to the country’s push for lower-resource AI infrastructure. Located in the Lin-gang Special Area off Shanghai, the facility sits about 35 meters beneath the sea and carries a planned capacity of 24 megawatts. The project represents an estimated $226 million investment and brings offshore energy, seawater cooling and high-density computing into a single infrastructure model. Its significance extends beyond the novelty of placing servers underwater, as operators increasingly face tighter constraints around electricity, cooling water and land availability.

The project brings together Shenergy, China Telecom’s Shanghai branch, INESA and China Communications Construction Company’s Third Harbor Engineering division. Its power strategy centers on offshore wind, with the facility targeting roughly 95% of its electricity demand from wind generation. The design also aims to cut electricity consumption by 22.8% compared with comparable land-based data centers while maintaining a target Power Usage Effectiveness below 1.15. That combination gives the project a more strategic role in China’s infrastructure planning, where renewable generation and computing capacity increasingly need to develop in tandem.

Seawater Cooling Changes the Infrastructure Equation

The facility uses surrounding seawater as its primary cooling resource, removing the need for freshwater consumption associated with conventional cooling systems. Its submerged architecture also cuts land requirements by more than 90%, creating an alternative for markets where suitable development land remains constrained. In addition, the cooling architecture could offer an important pathway for high-density computing, where rising thermal loads increasingly shape the economics of AI infrastructure. The project therefore connects three increasingly important infrastructure pressures: power availability, thermal management and physical space.

The Lin-gang system has been engineered around high-density GPU computing and workloads that demand substantial processing capacity. Those applications include big-data processing and the development of large language models, positioning the installation closer to an AI infrastructure experiment than a conventional subsea computing project. The architecture also fits China’s broader “East Data West Computing” strategy, which seeks to coordinate computing resources with energy availability across different regions. For data center developers, that alignment matters because future AI deployments will increasingly depend on how efficiently computing capacity connects with available power and cooling resources.

China Tests Whether Subsea Computing Can Scale

Wang Shifeng, chairman of China Communications Construction Company’s Third Harbor Engineering division, said the technology remains at an early stage even as the sector moves toward commercial deployment. “While the sector is moving towards commercial deployment, the technology remains in its initial stages,” he said. His comments highlight the gap between demonstrating technical feasibility and proving that underwater infrastructure can deliver competitive economics, reliability and maintainability over decades. Operators and investors will therefore need to examine long-term performance before treating subsea facilities as a mainstream alternative to conventional data centers.

That evaluation will also extend beyond computing performance. Data center operators and investors need to assess the technology’s economic viability, while policymakers and regulators will have to consider potential environmental and operational effects on coastal development and energy policy. At the same time, engineering and technology companies can examine the project’s seawater cooling systems and subsea construction techniques for applications in future sustainable infrastructure. The commercial question is no longer simply whether servers can operate underwater, but whether the full infrastructure stack can outperform land-based alternatives under the right power and resource conditions.

Lin-gang Builds On China’s Earlier Underwater Push

The Shanghai project follows an earlier underwater data center deployment off Hainan Province in December 2022, which was designated the world’s first commercial underwater data center. Lin-gang now moves the concept toward a larger commercial-scale configuration that combines subsea computing with offshore renewable power. If the facility delivers its targeted efficiency and operational performance, it could provide China with a model for deploying computing capacity where land and freshwater resources are under pressure. The project could also attract attention from countries confronting similar constraints as AI workloads accelerate.

The broader implication lies in whether underwater infrastructure can become a repeatable platform rather than a demonstration project. Success would require reliable subsea operations, predictable economics, maintainable equipment and measurable environmental performance over an extended operating period. For the data center industry, Lin-gang offers a real-world test of whether cooling can move outside the traditional mechanical plant and whether renewable power can become an integral part of the physical computing architecture. China is now putting that proposition through a commercial-scale test, with the results likely to shape how developers view the next generation of resource-constrained AI infrastructure.