Nations Have Turned to Maritime Data Centers
Rising compute demand for generative AI has prompted the exploration of offshore, water-cooled infrastructure.
Updated on Sept. 21, 2026 in Data Centers
Live Poll
Should nations prioritize building underwater data centers to meet growing AI computing power demands?
Countries and technology firms are increasingly developing floating and underwater data centers to support the escalating energy and cooling requirements of artificial intelligence. These facilities utilize maritime environments for natural cooling, marking a shift from traditional land-based site development.
Why it matters
The rapid expansion of generative AI is driving an urgent need for scalable computing power that can bypass the land and power limitations of existing facilities. By leveraging seawater, these projects aim to achieve more efficient thermal management.
These maritime facilities leverage seawater for natural cooling, a significant shift from the industrial air-conditioning systems used in land-based data centers. Modular designs allow for scalable deployment, though maintenance stability and communication latency remain key engineering variables.
The players
Microsoft
A global software and cloud infrastructure giant that operates a vast network of terrestrial data centers and conducted early research into submerged server hardware.
The details
Maritime data centers use the surrounding seawater as a thermal sink to dissipate the intense heat generated by high-density server racks, significantly reducing the energy needed for active cooling. These units are built with modular architectures—interchangeable and repeatable hardware blocks—that allow for flexible deployment in offshore environments. Despite these advantages, the technology must overcome significant challenges in hardware maintenance, communication stability across undersea cables, and potential ecological impact on local aquatic environments.
Timeline
2018: Microsoft began initial experiments with underwater data centers.
2028: Singapore plans to complete a four-module floating data center.
The Tech Race
The transition to maritime infrastructure follows the precedent set by early research projects like Microsoft's Project Natick. Countries such as China, Japan, and South Korea are now scaling these concepts to national-level initiatives to address the surging energy needs of their domestic AI sectors.
These projects are primarily focused on national-level infrastructure and will not impact individual consumers or small-scale workflows in the near term. The success of these deployments will dictate how quickly and efficiently AI services can scale globally by offloading cooling demand from the power grid.
The takeaway
Maritime data centers represent a shift toward specialized, climate-resilient hardware hosting for the AI era. Readers should watch for the completion of Singapore's 2028 facility as a benchmark for the feasibility of large-scale offshore computing.
What happens next
Singapore expects to complete its four-module floating data center by 2028.
Further reading
For more on the changing landscape of server infrastructure, see Data Centers.
Live Poll
Should nations prioritize building underwater data centers to meet growing AI computing power demands?
