China has officially begun commercial operations at a pioneering underwater data centre near Shanghai, marking a major step in sustainable computing infrastructure. Located around 35 metres beneath the ocean surface in Shanghai’s Lingang Special Area, the facility uses naturally cool seawater to regulate temperatures, significantly reducing energy consumption while supporting the growing demands of artificial intelligence (AI) and cloud computing.

The project combines subsea computing with offshore renewable energy, offering an innovative solution to the rising electricity requirements of modern data centres.

Data centre designed for AI-era computing

The underwater facility houses nearly 2,000 servers, including high-performance GPU clusters supplied by China Telecom and LinkWise. These servers are designed to process resource-intensive workloads such as artificial intelligence applications, 5G services and large-scale data annotation.

Developed jointly by Chinese authorities and engineering firm HiCloud Technology, the project represents an investment of approximately $226 million. The 24-megawatt installation is expected to provide reliable computing power while improving energy efficiency through its unconventional underwater design.

Seawater provides natural cooling

Unlike conventional land-based data centres that rely on energy-intensive cooling systems, the underwater facility takes advantage of the ocean’s naturally stable temperatures.

The servers are housed inside sealed, pressure-resistant modules submerged beneath the sea. The surrounding seawater acts as a passive cooling medium, eliminating much of the need for mechanical air-conditioning systems that consume significant amounts of electricity.

Cooling has become one of the biggest operational challenges for modern data centres, particularly those running advanced AI models that require powerful graphics processing units (GPUs). These processors generate enormous amounts of heat during continuous operation, making efficient cooling essential for performance and reliability.

Improved energy efficiency

According to Chinese media reports, the underwater data centre has achieved a Power Usage Effectiveness (PUE) rating of below 1.15. PUE is a widely used industry metric that measures how efficiently a data centre uses electricity.

A lower PUE indicates that a larger proportion of energy is dedicated to computing tasks rather than supporting infrastructure such as cooling, lighting and ventilation. By comparison, the global industry average is around 1.5, making the Shanghai facility significantly more energy-efficient than many conventional data centres.

This improved efficiency could help reduce operational costs while lowering the environmental impact of large-scale computing facilities.

Meeting growing AI energy demands

The launch comes at a time when countries worldwide are seeking sustainable ways to support the rapid expansion of artificial intelligence. As AI models become increasingly sophisticated, demand for high-performance computing infrastructure continues to grow, placing significant pressure on electricity grids.

Industry experts have been exploring alternative cooling technologies, renewable energy integration and innovative infrastructure designs to address these challenges. China’s underwater data centre combines passive ocean cooling with offshore wind energy, offering a potential model for future AI infrastructure.

If successful, the project could inspire similar developments in other coastal regions, helping meet rising computing demands while improving energy efficiency and reducing reliance on traditional cooling systems. As AI adoption accelerates globally, such innovations may play a key role in making data centres more sustainable and environmentally responsible.