China Activates First Commercial Underwater Data Center, Reshaping Global Computing Strategy
June 20, 2026
China Activates First Commercial Underwater Data Center, Reshaping Global Computing Strategy
The global tech industry is confronting a dual crisis: a shortage of power and, more critically, a shortage of water. Traditional terrestrial data centers are hitting thermodynamic limits, consuming vast amounts of electricity and municipal water just to keep hardware from overheating. Metrics such as Power Usage Effectiveness (PUE) and Water Usage Effectiveness (WUE) are under immense strain, with hyperscale facilities struggling to push PUE below 1.1 using conventional air cooling. Water consumption has also become a geopolitical concern—Microsoft alone reported using 5.8 billion liters of water in 2024, while Meta consumed 3.1 billion liters globally. The long-term solution, experts argue, is not to build larger air conditioners on land, but to relocate infrastructure entirely.
This strategic pivot is now taking shape off the coast of Hainan Island, where China has activated the world’s first commercial underwater data center. Led by the Chinese company Highlander, with substantial backing from the Chinese government, the deployment represents the first truly scalable commercial application of underwater data center (UDC) technology. A single 1,400-tonne module has been submerged 35 meters onto the seafloor, using the surrounding seawater as an infinite natural cooling sink. While this initial module serves as a proof of concept, development plans call for 100 modules to be deployed at the site, fundamentally altering the economics of data center operations. By eliminating mechanical chillers, heat exchangers, and evaporative cooling towers, Highlander effectively removes the cooling penalty that burdens terrestrial competitors. The approach also reclaims billions of liters of municipal water that land-based data centers evaporate daily—a critical advantage in an era of escalating water scarcity.
The concept of underwater data centers is not entirely new. In 2018, Microsoft launched Phase II of its Project Natick, deploying a container-sized module off the coast of Scotland’s Orkney Islands. That experimental facility housed 864 servers in a nitrogen-rich environment, powered entirely by renewable energy. After two years, Microsoft found that the underwater data center exhibited a failure rate roughly one-eighth that of traditional land-based facilities. Without oxygen, fluctuating humidity, or human technicians bumping into racks, the hardware thrived in isolation. Despite this technical success, Microsoft did not scale the project commercially. The reason lies in the logistical friction of deployment: underwater systems require specialized maritime vessels, deep-sea rigging crews, and massive upfront capital. If a module fails, a multi-ton pressure vessel must be winched from the seafloor. Highlander and the Chinese government are betting that long-term operational savings and strategic security advantages will offset these costs—a long-term infrastructure bet that Western tech giants, constrained by quarterly earnings pressures, have so far hesitated to make at full commercial scale.
The ocean is not the only frontier being considered for alternative data center locations. Space-based computing, including orbital and lunar data centers, has gained attention through projects like the European Commission’s ASCEND study, led by Thales Alenia Space. Advocates argue that space offers a flawless vacuum for insulation and passive radiative cooling, with continuous solar power that can generate roughly 40% higher peak output than terrestrial solar farms. However, the physics and logistics of space computing remain prohibitive. The cost per kilogram to launch hardware into Low Earth Orbit has dropped with reusable rockets, but it still cannot compete with the cost-effectiveness of loading a module onto a marine barge. Space computing also faces extreme radiation hazards, requiring expensive shielding estimated at $1.2 million per kilowatt of compute for launch costs alone. Lunar data centers introduce latency issues, with a round-trip communication delay of 2.6 seconds—an eternity for agentic AI systems or high-speed financial trading. Underwater data centers, by contrast, can be located mere miles from major metropolitan hubs, maintaining sub-millisecond fiber-optic latency while leveraging the ocean’s thermal mass.
The terrestrial data center industry has not stood still. Advanced cooling technologies such as direct-to-chip liquid cooling and full immersion cooling are rapidly replacing traditional air cooling. These methods are effective at moving heat away from high-density processors, but they still require heat exchangers and secondary loops that dump heat into the atmosphere or use massive evaporative cooling towers, keeping them tied to the power grid and municipal water supplies. In drought-prone regions, this consumption is drawing intense regulatory scrutiny, with states like California exploring stringent water reporting mandates for data centers. Underwater data centers bypass these constraints entirely, offering a zero-water-discharge solution where heat is dissipated directly into ocean currents. While concerns about localized thermal pollution exist, the sheer volume of the ocean diffuses heat far more efficiently than the atmosphere handles output from terrestrial cooling towers. With proper spacing and environmental monitoring, UDCs represent a pinnacle of sustainable data center design.
Looking toward 2030, the race for infrastructure dominance is shifting to the ocean. Based on current trajectories, China is positioned to become the global leader in underwater data center deployments. While European and American companies largely treat the concept as a research project, China is executing it as national infrastructure policy. The plan for 100 modules off Hainan Island is just the beginning. Commercially, this gives Chinese technology firms a competitive advantage in operational costs, allowing them to allocate more of their power budgets directly to AI computing, translating into cheaper AI training and more robust digital services. Militarily, the implications are profound. Data centers are the nerve centers of modern warfare, controlling autonomous drone swarms, logistics, and cyber operations. A terrestrial data center is a massive, static, easily identified target. An underwater data center is hidden, hardened against electromagnetic pulse attacks by seawater, and difficult to target with conventional weaponry. However, this physical isolation introduces new risks. The underwater domain is difficult to secure against espionage or sabotage, requiring a new ecosystem of surveillance technologies, including autonomous underwater vehicles, remotely operated vehicles, and smart sensor buoys. The nation that masters both the deployment and defense of underwater data centers will possess a resilient, nearly invisible computing backbone capable of surviving disruptions that would cripple terrestrial networks.
The activation of Highlander’s 1,400-tonne underwater data center off Hainan Island marks a watershed moment. It proves that the thermodynamic and resource challenges of the AI era can be met by utilizing the planet’s largest natural heat sink. While space-based computing remains a theoretical pursuit hindered by launch costs and latency, the deep sea offers immediate, scalable solutions to energy and water scarcity. Microsoft proved that hardware could survive in a sealed subsea environment; China is now proving that the business model can scale globally. As the industry barrels toward 2030, the strategic advantage will shift to those who recognize that the future of the cloud lies at the bottom of the ocean.
Source: techspective
