AI's soaring demand drives firms to test ocean-based data centers
The surge in AI workloads is prompting operators to consider underwater data centers as an alternative to land, but significant environmental and technical hurdles remain.
Growing AI workloads are driving unprecedented demand for computing capacity, leading operators to explore ocean-based data centers that promise lower energy consumption, reduced freshwater use and minimal land footprint. Early experiments began with Microsoft's 2018 seafloor test near Scotland's Orkney Islands, which showed lower server failure rates but was discontinued in 2024, likely due to permitting and upgrade issues.
More recent initiatives include a wind-powered underwater facility in Shanghai, a floating container site near Yokohama powered by solar and batteries, Singapore's four-story floating center under construction by Keppel, and plans for a massive submersible hub in Ulsan, South Korea, as well as a tidal-turbine-driven concept in Maine's Bay of Fundy. Proponents argue these designs can cut electricity demand by up to 30 % and lessen reliance on fresh water, yet critics warn that discharging warmed seawater could disturb oxygen levels, pH and marine life, especially as many ecosystems are already stressed.
Maintenance remains a challenge because faulty hardware may require the entire sealed module to be retrieved. The ultimate viability of underwater data centers will depend on addressing these environmental, technical and regulatory obstacles.
Why it matters
AI's growth demands greener computing solutions; ocean data centers could cut energy and water use but must avoid harming marine ecosystems.
In this story
