AI Data Centers Are Moving Underwater: Here's Why the Ocean Might Be the Answer to a Growing Power Crisis
AI's explosive growth is forcing data centers to look beyond land, with companies worldwide testing underwater and floating facilities that promise to slash energy consumption and reduce freshwater demand. China's first wind-powered underwater data center, which launched commercial operations in May 2026, uses seawater cooling to cut electricity use by at least 30% compared to traditional facilities. Japan, Singapore, South Korea, and the United States are all pursuing similar ocean-based approaches, signaling a major shift in how the tech industry plans to power artificial intelligence infrastructure.
Why Are Data Centers Moving Into the Ocean?
The numbers tell the story. A Gallup poll conducted in March 2026 found that 70% of Americans oppose building AI data centers in their communities, citing concerns about energy consumption, water use, and environmental impact. That public resistance, combined with the sheer scale of AI's infrastructure demands, is pushing developers to explore alternatives. The ocean offers several advantages: seawater can cool servers without requiring energy-intensive refrigeration systems, coastal locations reduce the need for scarce land, and offshore wind and solar installations can power facilities with renewable energy.
Microsoft pioneered this concept with a research project that began in 2015. The company deployed a waterproof data center on the seafloor near Scotland's Orkney Islands in 2018, containing 864 servers connected to shore by underwater cable. After two years of operation, Microsoft reported that servers in the underwater facility failed at roughly one-eighth the rate of comparable land-based systems. The company hypothesized that the sealed underwater environment, with reduced exposure to oxygen, humidity, and temperature fluctuations, contributed to the improved reliability. However, Microsoft ended the project in 2024, likely due to regulatory hurdles and the difficulty of upgrading equipment without bringing entire modules to the surface.
What Ocean-Based Data Center Projects Are Actually Operating Now?
Several countries have moved beyond research into commercial deployment. China's underwater data center in Shanghai, which began full commercial operations in May 2026, represents the most advanced project to date. The $226 million facility uses seawater as a coolant and offshore wind turbines for power, reducing electricity consumption by at least 30% and cutting carbon emissions significantly compared to traditional data centers.
Japan is testing a different model with floating platforms. A data center housed in shipping containers opened on a floating platform near Yokohama in 2025, powered by solar panels installed on the same platform with battery storage. The test is scheduled to continue through March 2027. Singapore is moving toward commercial scale with infrastructure company Keppel building a four-story floating data center scheduled to open in 2028, which will use seawater cooling to reduce reliance on treated freshwater. South Korea's Ulsan region began planning an underwater data center in 2025 that could house more than 100,000 servers and use 30% less power than land-based facilities.
In North America, DeepGreen Western Passage has proposed a submersible AI data center in Maine's Bay of Fundy, powered by tidal turbines designed to harness the area's strong tidal currents. Portugal's SIN01 AI data center in Sines already operates using seawater from the Atlantic Ocean to cool its servers before returning the water to the ocean.
How to Evaluate Ocean Data Center Sustainability
- Energy Efficiency Gains: Ocean-based facilities reduce cooling electricity by 30% or more by using seawater instead of refrigerated freshwater, directly lowering grid demand and carbon emissions from power generation.
- Renewable Integration: Coastal locations enable direct connection to offshore wind farms and solar installations, reducing reliance on fossil fuel power plants and improving the carbon footprint of AI infrastructure.
- Land and Water Conservation: Floating and underwater data centers eliminate demand for terrestrial real estate and treated freshwater supplies, addressing two critical constraints in densely populated regions like Singapore and coastal Japan.
What Are the Environmental Risks?
The promise of ocean-based data centers comes with significant environmental concerns. The primary issue involves thermal pollution from cooling water discharge. When seawater is used to cool servers, it must be returned to the ocean warmer than when it entered. Current facilities show temperature increases of less than 1 degree Celsius (1.8 degrees Fahrenheit) in surrounding seawater, but this seemingly small change poses risks to marine ecosystems. Many marine species depend on stable water temperatures for breeding, feeding, and migration patterns. If multiple underwater data centers operate in the same region, their combined thermal discharge could create localized thermal pollution and alter marine food webs.
The timing of this expansion is particularly concerning. UNESCO estimates that approximately 60% of marine ecosystems are already degraded or used unsustainably. The ocean is warming due to climate change, and coral reefs, mangrove ecosystems, and seagrass beds are already under stress. Adding waste heat from data centers could accelerate these declines, creating cascading effects throughout marine ecosystems.
What Practical Challenges Remain Unsolved?
Beyond environmental concerns, underwater data centers face significant operational hurdles. Maintenance is perhaps the most critical challenge. If a single server fails in a sealed underwater module, technicians cannot repair or replace it on site. The entire data center module may need to be brought to the surface for repairs, a process that is expensive, time-consuming, and disruptive to operations. This limitation makes underwater facilities less attractive to hyperscalers, which require rapid equipment upgrades and replacements to stay competitive in AI development.
Regulatory uncertainty also remains. Underwater data centers require environmental permits and must navigate complex maritime and environmental regulations that vary by country and region. These approval processes can delay projects significantly, as Microsoft's experience demonstrated. The lack of established regulatory frameworks for ocean-based data centers means that companies must work with governments to develop new rules, adding time and uncertainty to project timelines.
Whether underwater data centers become a sustainable solution to AI's infrastructure crisis depends on solving these economic, technical, and environmental problems. The technology shows promise for reducing energy consumption and freshwater demand, but the marine ecosystem risks and operational challenges suggest that ocean-based facilities will likely complement, rather than replace, land-based data centers in the near term.