China Submerges Data Centers to Solve the AI Power Crisis

The Shanghai Lingang undersea demonstration, led by HiCloud Technology and China Communications Construction, highlights a growing trend toward marine-based computing. By utilizing natural seawater for cooling, the facility sidesteps the energy-intensive process of pumping freshwater through server systems—a method that accounts for up to 40% of electricity usage in land-based centers. According to government data, this design cuts energy consumption by nearly 23% and eliminates land usage requirements entirely.

This shift arrives as international concern mounts over the environmental footprint of artificial intelligence. The United Nations University Institute for Water, Environment, and Health warns that global data centers could consume 9.3 trillion liters of water by 2030. While Microsoft pioneered underwater testing in 2018 near Scotland, China has moved toward rapid commercial deployment, leveraging state funding and integrated marine engineering capabilities to scale the technology.

Competition is intensifying across Asia. South Korea’s Samsung Heavy Industries is currently developing floating data centers, having secured design approvals for 50-megawatt models that can operate on self-generating power. Meanwhile, Japanese firms Hitachi and Mitsui O.S.K. Lines are exploring the conversion of existing ships into floating server farms. Despite potential risks to marine ecosystems, the industry maintains that thermal impact remains localized and manageable, pushing underwater and floating infrastructure to the forefront of the green energy transition.