How Are Undersea Data Centers Redrawing Global Power Balances?

While global debates often focus on the economic and regulatory dimensions of artificial intelligence (AI), the physical infrastructure supporting these technologies remains largely shrouded in mystery—particularly that which lies beneath the world’s seas and oceans. One major question is missing from most discussions: what role do the oceans play in the global AI landscape, given that a large portion of AI’s physical infrastructure now resides there in the form of massive data centers? Moreover, how is the strategic competition between China and the United States reshaping the undersea data center landscape?

In reality, the world’s oceans conceal an entire parallel realm of infrastructure that enables AI to function and evolve. The digital revolution we’re experiencing today doesn’t truly “float on the cloud,” as the term cloud computing suggests. Instead, it is built upon a vast cyber-physical infrastructure network—including giant data centers, energy stations, and undersea internet cables stretching tens of thousands of kilometers beneath the ocean’s surface. These form the physical backbone of the internet and smart technologies, serving as the essential foundation of modern AI systems.

While undersea cable security has drawn considerable attention in recent years, this analysis focuses on undersea data centers as the “ambitious physical infrastructure” of AI. With the intensifying strategic rivalry between China and the United States, these data centers have become both geopolitical and vital arenas in the race for AI supremacy—necessitating a deeper exploration of their strategic role and global implications.

Rising Growth

Amid a fierce global race to develop AI capabilities, attention is shifting toward the seas and oceans as a new frontier for advanced digital infrastructure. Whereas major economic powers have long competed on land and in cyberspace, the maritime domain is now emerging as a strategic element that could change the rules of the game—particularly in the realm of AI-focused data centers.

1. Drivers of Attraction

The oceans offer a unique environment for data centers. Submerging these facilities underwater allows them to take advantage of natural cooling properties provided by seawater, significantly reducing the energy required to run high-performance computing systems. This is particularly important as AI’s growing computational demands generate vast amounts of heat that are difficult to dissipate through traditional means.

Their proximity to major energy sources—such as coastal wind farms or nuclear power plants—also facilitates continuous and efficient power supply.

China provides a leading example: its Hainan Island and Shanghai coastal projects have pioneered underwater data centers that combine seawater cooling with renewable energy sources such as wind power, achieving both operational efficiency and environmental sustainability.

Beyond energy savings, there are added environmental and operational advantages. Experiments like the Hainan project have shown lower server failure rates due to the use of nitrogen-filled environments, which are less corrosive than oxygen-rich air on land. Reduced human interference also minimizes the risk of mechanical damage.

This shift toward ocean-based data centers marks a major step in the transition to low-carbon digital infrastructure, potentially setting new global standards for sustainable computing. The move carries strategic weight as well, amid growing competition over water and energy resources—key not only for technological progress but also for human survival in a climate-stressed world. Unlike land-based facilities that can consume up to 4.8 liters of freshwater per kilowatt-hour for cooling, undersea centers require none, giving them a clear ecological edge.

Undersea data centers have thus emerged in recent years as a promising innovation to overcome energy and land constraints faced by traditional facilities. Though not yet a complete solution, their sustainability and efficiency make them an increasingly strategic choice in the age of big data and AI expansion.

Interest in the concept began in 2015 when Microsoft launched Project Natick off the coast of California, testing a submerged unit with 24 servers. This experiment paved the way for further advancements, including a second phase involving a full shipping-container-sized facility housing 12 racks—or 864 servers. Following Microsoft’s lead, companies such as Beijing Highlander Digital Technology in China and Subsea Cloud in the U.S. conducted similar tests, and major players like Amazon Web Services, Google, and Meta are reportedly exploring this field as well.

Operationally, underwater data centers demonstrate remarkable reliability and efficiency. For instance, Microsoft’s submerged data center achieved a Power Usage Effectiveness (PUE) of 1.07—compared to around 1.125 in modern land-based centers. The cooler underwater environment, averaging up to 10°C lower than on land, further enhanced performance and stability.

Another strategic benefit is reduced latency, improving connectivity for coastal populations—over half of the global population. With modular, pre-fabricated designs, these units can be deployed within just 90 days and operate autonomously for up to five years without maintenance. Microsoft’s model involves retrieving the units every five years to replace servers—allowing four operational cycles over a 20-year lifespan before full recycling.

Ongoing Challenges

While undersea data centers represent an innovative, eco-friendly approach to boosting efficiency and reducing carbon footprints, they still face significant challenges limiting widespread deployment in the near term.

Economically and technically, they face high design costs to withstand harsh marine conditions, as well as the risk of unforeseen failures that could lead to data loss due to prolonged unmanned operation—raising questions about their cost-effectiveness compared to land-based centers.

Environmentally, large-scale deployment of submerged structures raises concerns about potential harm to marine ecosystems, including thermal pollution, chemical emissions from metallic components, and construction noise. These risks could grow as the technology scales globally.

From a security perspective, undersea data centers’ direct integration with global energy and communication networks makes them vulnerable to complex security threats—including cyberattacks targeting control systems and physical sabotage at maritime chokepoints such as the Bab el-Mandeb Strait. This overlap between cybersecurity and maritime security calls for dual-protection strategies that inevitably raise operational and insurance costs.

Finally, there are regulatory and legal hurdles concerning maritime sovereignty and water-space rights. Issues like data ownership, jurisdiction, and liability for environmental damage or operational failures remain unresolved in many international contexts, complicating efforts to establish clear governance frameworks.

In short, undersea data centers embody a multi-dimensional fragility, and their future depends on how well they can adapt to a complex web of regulatory, environmental, and security challenges at the intersection of global investment, sustainability, and geopolitics.

The Geopolitical Race

In the past, sovereignty was defined by land and territorial waters. Today, nations are claiming digital sovereignty extending into the ocean depths. Data centers have become the core powerhouses of the AI era, pulling them into the center of the global technological and geopolitical rivalry, especially between the U.S. and China. These centers are now seen as strategic assets, akin to oil pipelines or maritime trade routes—key elements in a multi-front, largely undeclared technological war.

Just as the technological landscape has split between American and Chinese systems—such as in fiber-optic cables and satellite internet—the competition over data centers is intensifying. China’s investments in marine data facilities and the expansion of American firms into Indian and Middle Eastern ports signal a silent but high-stakes battle.

1. China’s Ambitions

China has taken the lead in deploying underwater data centers to support its AI ambitions, leveraging ocean resources to meet the immense energy needs of high-performance computing. A prime example is the Hainan Undersea Data Center, launched in December 2022 by the China National Offshore Oil Corporation. Located 30 meters below the surface, its nitrogen-filled, sealed containers use seawater for cooling—achieving a 90% reduction in cooling energy costs compared to land-based centers, with server failure rates below 1% due to the stable, oxygen-free environment.

In June 2025, China announced another ambitious $223 million project to build a wind-powered underwater server farm near Shanghai, supported by the Shanghai government, Hailanion, and Alibaba. The facility will host around 800 AI-optimized servers and align with China’s carbon neutrality goal for 2060.

China’s advancements in undersea data infrastructure also bolster its global influence. By offering efficient, sustainable models, it appeals to developing nations in the Global South, particularly in Southeast Asia and Africa. Beijing has explored partnerships with countries like Malaysia and Indonesia, potentially integrating these facilities with local 5G and AI ecosystems. Such collaborations could reshape international standards for data infrastructure—challenging Western norms and fostering a parallel tech ecosystem led by China, helping it circumvent Western technological isolation.

2. The United States

American companies still dominate the global cloud computing market, and by extension, the data center industry. The U.S. alone accounts for roughly 40% of the global market, hosting over 5,300 large-scale data centers run by firms like AWS, Microsoft Azure, Google Cloud, Equinix, and IBM Cloud.

Washington has adopted a geopolitical strategy that uses export controls as a tool to maintain dominance in AI and cloud computing while limiting competitors—chiefly China. These restrictions not only block access to advanced chips and sub-7nm manufacturing equipment, but also indirectly target the infrastructure supporting undersea data centers, considered essential for advanced AI applications.

Since 2022, the U.S. Department of Commerce has imposed stringent export bans on major Chinese tech firms such as Huawei, SMIC, Tencent Holdings, and Alibaba Group, cutting off access to high-performance computing equipment and advanced GPUs from NVIDIA and AMD—components crucial for both terrestrial and undersea data centers.

Moreover, undersea data centers rely heavily on advanced technologies for submarine engineering, submerged cooling systems, and marine sensors—all classified by the U.S. as dual-use technologies (civilian and military). As such, Washington restricts exports of these technologies to ensure U.S. and Western control, preventing China from building integrated undersea data ecosystems that could provide a strategic AI advantage.

Microsoft’s successful Project Natick since 2015—proving the viability of underwater data centers—has heightened U.S. concerns over similar Chinese efforts. Washington now monitors any attempt by China to develop or import related technologies. Companies like Beijing Highlander, Digital Technology, and Subsea Cloud have all drawn attention from U.S. regulators amid fears they could evolve into massive undersea AI hubs capable of supporting military applications such as underwater surveillance or autonomous fleet control.

Thus, U.S. export controls act as more than economic tools—they form part of a “technological containment” strategy aimed at slowing or disrupting China’s potential use of undersea data centers as an AI power platform.

Meanwhile, the U.S. has rallied allies through the “Chip 4” alliance (Japan, South Korea, Taiwan) to secure semiconductor supply chains. The CHIPS Act of 2022, worth $52 billion, supports domestic semiconductor production and strengthens companies like TSMC. Yet, America’s allies maintain deep economic ties with China, complicating full decoupling—Japan cooperates with China on 5G infrastructure, and Samsung continues supplying memory chips to Chinese data centers.

Global Data Centers by Country (March 2024)

Future Outlook

1. An Alternative Undersea Network
   China may compensate for U.S. restrictions by expanding cooperation with Russia, North Korea, and Southeast Asian states seeking independence from Western infrastructure. Moscow could provide cybersecurity or technical support, while Beijing offers funding and engineering. This could result in a parallel undersea data network outside U.S. control—a direct challenge to Western digital dominance.
2. Militarization of Undersea Data Centers
   Through initiatives like “Made in China 2025”, Beijing aims to build a domestic semiconductor industry. If it achieves mass production of mid-range chips (10–14 nm), it could deploy them in military-oriented undersea data centers supporting defense AI or maritime surveillance systems.
   Integrated with “smart underwater sensor networks,” such facilities could monitor U.S. naval movements in the Pacific or operate long-endurance unmanned submarines, turning data centers into strategic military assets—heightening tensions in the South China Sea and Taiwan Strait.
3. Fragmentation of Global Digital Infrastructure
   These trends may prompt the U.S. to strengthen tech-security partnerships in the Indo-Pacific, expanding beyond AUKUS (Australia–UK–U.S.) to include India, Japan, and South Korea in developing secure, jointly operated undersea data centers. In response, China could extend its Digital Silk Road to include undersea data centers in Gulf and African nations, where it enjoys economic leverage—leading to global fragmentation of the digital-marine infrastructure into rival blocs.

Conclusion

Undersea data centers have become a reality, offering innovative solutions for cooling, energy efficiency, and space limitations—along with promises of sustainability. Yet, they face hurdles in maintenance, expansion, and security. As they increasingly support AI technologies, they have evolved into a new geopolitical battleground between the U.S. and China.

This rivalry deepens the global tech conflict and is reshaping the geopolitical balance in the age of artificial intelligence.