Table of Contents
Martina Raveni, senior analyst at GlobalData, previously told RCRTech that the deployment costs of underwater data centers are high as companies need to design, build, and then deploy the infrastructure in the ocean
In sum – what to know:
Direct wind power – The underwater facility is connected to Lin-gang’s offshore wind farm, with modules positioned near turbines and located 10 meters below sea level.
Efficiency gains – Developers said the system cuts electricity consumption by 22.8%, eliminates water usage and reduces land requirements by more than 90%.
AI computing focus – GPU servers inside the underwater modules support big data processing and domestic large language model (LLM) development, with China Telecom among early users.
China has brought what developers describe as the world’s first offshore wind-powered underwater data center into operation near Shanghai, highlighting growing efforts to combine renewable energy generation with AI infrastructure deployment.
The underwater data center (UDC) is located in the Lingang Special Area of the China (Shanghai) Pilot Free Trade Zone and is directly connected to the Lin-gang offshore wind farm. According to project developers and Chinese media reports, the facility is positioned 10 meters underwater between the first and second phases of the offshore wind project.
The project was formally launched in June 2025 through a cooperation agreement involving the Lingang Special Area administrative committee, Shanghai Lingang Special Area Investment Holding Group and HiCloud Technology.
Construction of the facility was completed in October 2025, when Chinese authorities described it as “the world’s first wind-powered underwater data center project.” Earlier this year, Lingang officials said the CNY 1.6 billion ($228 million) project had entered operation.
The deployment is being developed in two phases. The initial demonstration phase has a capacity of 2.3 MW, while the second phase is expected to scale the project to 24 MW.
The facility combines offshore renewable energy generation with seawater cooling to reduce operating costs and environmental impact. Developers said the underwater configuration lowers electricity consumption by 22.8%, removes the need for freshwater cooling and cuts land usage by more than 90% compared with conventional facilities.
The project’s reported power usage effectiveness (PUE) is around 1.15, a figure considered highly efficient for modern data center operations.
The underwater modules house GPU servers designed for AI and high-performance computing workloads, including big data annotation and development of domestic large language models (LLMs). The project also supports coordinated computing resource allocation between offshore and onshore infrastructure.
According to Lingang officials, computing clusters from companies including China Telecom have already been deployed at the site, alongside local computing service providers such as LinkWise.
The project reflects broader industry efforts to reduce the energy and cooling demands associated with AI infrastructure expansion. As hyperscalers and cloud providers increase investments in AI computing capacity, operators are exploring alternative approaches to power sourcing, cooling efficiency and land utilization.
Underwater data centers have been tested previously in pilot projects globally, primarily because seawater can provide natural cooling and potentially lower operational energy requirements. China’s Lingang project combines that concept with direct integration into offshore wind generation infrastructure.
The development also comes as governments and technology companies increasingly focus on locating AI infrastructure near large-scale renewable energy assets to address growing power demand from AI workloads and data centers.
Martina Raveni, senior analyst at GlobalData, previously told RCRTech that the deployment costs of underwater data centers are high as companies need to design, build, and then deploy the infrastructure in the ocean.
“The underwater environment is very challenging because you need specialized maintenance if you need to repair. For example, you need to retrieve the entire module… and of course, it may result in downtime,” Raveni said.
Environmental risks also remain a concern. Raveni noted that temperature fluctuations could damage components, while the ecological impact of ocean-based systems is still under evaluation. “We’re putting data centers in the ocean also because we could use the seawater for cooling, and this might harm the environment… any chemical contamination can be possible.”
