China has connected its first underwater data center in Shanghai , designed to run on energy from offshore wind farms . Located off the coast of Lin-gang, the facility combines digital infrastructure with renewable energy generation at sea to reduce energy consumption, water usage, and land use.
The project, located in the Lingang Special Area's Pilot Free Trade Zone , starts with an initial operating capacity of 2,3 MW and is expected to reach 24 MW upon completion. The total investment amounts to 1.600 billion yuan (approximately US$226 million), and it integrates wind power generation with IT infrastructure in a coordinated manner.
Location, investment and roadmap
The Lin-gang site allows for the harnessing of resources from the East China Sea , where usable wind hours exceed 3.000 per year, providing a stable electricity supply with a low carbon footprint. In its first phase, the center is already operational and has been designed with a PUE ( power efficiency ) target of no more than 1,15.
More than 95% of the electricity consumed will be from renewable sources , prioritizing offshore wind power. This integration facilitates consuming the generated electricity on the coast itself, minimizing transmission losses and improving the operational stability of the offshore platform.
Regarding the expansion, the entities involved —including Shanghai Hicloud Technology , Shenergy Group , China Telecom (Shanghai), INESA and Third Harbor Engineering— have signed an agreement to promote a 500 MW underwater data center program linked to offshore wind power , marking the leap from the demonstration phase to an industrial scale.
The initiative is part of Shanghai's local smart computing strategy, which aims to raise computing power to 200 EFLOPS in the coming years, and fits with national plans to distribute digital infrastructure and strengthen the efficiency of the computer network.
Efficiency, cooling and lower impact
One of the biggest changes comes with cooling . In land-based facilities, cooling can account for between 40% and 50% of total electricity consumption; in this case, using seawater as a natural heat sink reduces that fraction to less than 10% , with the consequent energy savings.
The design eliminates the need for freshwater to cool equipment and drastically reduces the land requirement by relocating most of the infrastructure to the seabed. According to project managers, the combination of offshore wind and submerged architecture allows for energy savings of 22,8% compared to conventional data centers.
Regulatory standards also play a role: China's 2024 policy requires that large new or upgraded hubs achieve a PUE ≤ 1,25 by the end of 2025 (≤ 1,20 for key nodes). The Lingang project falls below these thresholds, reinforcing a low-carbon approach to digital infrastructure.
Beyond efficiency, the platform is designed to support AI workloads , 5G networks , the Industrial Internet, and e-commerce, supporting the growth of the digital economy without increasing water impact or land occupation.
The deployment, however, presents challenges: the technological maturity of the submersible modules, the logistics of long-distance operation and maintenance, corrosion resistance, and the environmental management of the marine environment. The partners involved emphasize that cost reduction and standardization will be key to scaling up to large volumes.
As a reference for other regions, this model opens the door to similar solutions where offshore wind power is well-established, provided that the technical and environmental requirements are met. Integrating renewable energy generation with data centers located near the energy source can alleviate pressure on terrestrial grids and improve the balance between computing power and sustainability.
With the first phase already active, a target of 24 MW and a framework agreement of 500 MW, Shanghai's underwater wind data center sets a precedent: combining offshore wind and ocean cooling to move towards more efficient digital infrastructures , with less water and lower energy consumption per computing unit.
