China Turns to Underwater Data Centers to Fuel AI Boom
April 7, 2026
Facing the dual pressures of skyrocketing energy consumption from artificial intelligence and ambitious carbon neutrality goals, China is pioneering a novel solution: submerging data centers in the ocean. This move represents a significant strategic shift in digital infrastructure, aiming to address the critical bottlenecks of cooling and energy efficiency that threaten to constrain the nation's AI ambitions.
The initiative is being led by major Chinese technology and industrial firms. Notably, a subsidiary of the state-owned shipbuilding giant China State Shipbuilding Corporation Limited (CSSC) has successfully deployed an underwater data module in the tropical waters off the coast of Hainan Island. This follows earlier commercial deployments by companies like Beijing Highlander Digital Technology, which has modules operating in the coastal regions of Guangdong and Hainan provinces. These submerged facilities leverage the consistent, low temperature of seawater for natural cooling, a process that can consume up to 40% of a traditional land-based data center's power.
The scale of these projects is substantial. The recently launched module by CSSC's unit has a designed capacity of 1,300 racks, highlighting the serious investment in this technology. Proponents argue that underwater data centers can achieve a Power Usage Effectiveness (PUE) — a key metric where 1.0 is ideal — as low as 1.1, dramatically more efficient than the national average for conventional centers. A company representative involved in the Hainan project emphasized the broader vision, stating, "The development of underwater data centers is an important measure for implementing the national strategy of building a maritime power and expanding marine economy."
The implications for China's tech industry are profound. By drastically reducing cooling costs and energy consumption, underwater data centers could provide a more sustainable and scalable power foundation for the computationally intensive workloads required by AI training and large language models. This approach not only helps manage the local energy grid strain but also positions China as a leader in innovative, green data center solutions. If proven reliable and cost-effective at scale, this technology could redefine global standards for building the infrastructure of the AI era.
Source: techinasia
