China's ambitious underwater data center has reached full operational capacity, marking a significant milestone in integrating renewable energy with advanced computing infrastructure. Officially launched in June 2025, this facility has begun processing workloads ranging from artificial intelligence to big data annotation, utilizing nearly 2,000 servers housed 35 meters beneath the ocean's surface.
This 24 megawatt (MW) installation represents a pioneering approach in the data center sector, using the surrounding seawater as a passive cooling solution. Traditional land-based data centers typically rely on energy-intensive cooling systems that can significantly increase their Power Usage Effectiveness (PUE). In contrast, the Shanghai underwater data center claims a PUE below 1.15—an impressive figure compared to conventional facilities that often exceed 1.5.
The efficiency achieved through this underwater design addresses a critical issue in modern AI infrastructure: cooling. Dense GPU configurations can consume vast amounts of energy, and this energy conversion into heat requires stable cooling solutions. By harnessing the ocean's natural thermal properties, this facility can lower the overall energy costs associated with cooling, thereby enhancing its operational efficiency.
Strategically connected to offshore wind farms, the facility operates sustainably by sourcing a significant portion of its electricity from renewable resources. This alignment between digital infrastructure and renewable energy is increasingly important as global demand for AI processing power rises alongside electricity consumption.
However, deploying underwater data centers comes with challenges. Issues such as saltwater corrosion, long-term pressure sealing, and maintenance accessibility present significant engineering hurdles. Unlike traditional data centers where technicians can quickly replace failing hardware, the underwater environment complicates physical access to critical components. Operators must implement sealed modular designs and extensive remote monitoring systems to minimize the need for onsite interventions.
This development follows earlier experiments in underwater data center technology, notably Microsoft's Project Natick, which explored submerged data center solutions in various locations but ultimately halted commercial efforts. Despite these setbacks, the Shanghai project showcases the potential for underwater facilities to achieve lower hardware failure rates and improve energy efficiency.
As the demand for AI infrastructure escalates, more projects are emerging worldwide. A recent initiative by Panthalassa, a startup backed by Peter Thiel, focuses on developing wave-powered floating data centers that utilize similar cooling strategies while remaining operational far offshore.
The launch of this underwater data center not only highlights China's commitment to renewable energy and innovative infrastructure but also signals a shift in how the tech industry addresses the growing challenge of energy and cooling management in AI applications. As the industry evolves, unconventional solutions will play a critical role in meeting the insatiable demand for computational power.
Quick answers
What is the significance of the underwater data center?
It represents a novel integration of renewable energy and passive cooling for AI workloads.
How does the underwater cooling system work?
Seawater is used as a passive heat sink to absorb excess heat generated by the servers.
What are the operational challenges of underwater data centers?
Challenges include saltwater corrosion, hardware maintenance complexities, and ensuring subsea cable reliability.
What is the Power Usage Effectiveness (PUE) of the facility?
The facility claims a PUE below 1.15, significantly lower than traditional data centers.



