Floating data centers are transitioning from theoretical concepts to engineered infrastructure projects, with multiple companies now designing offshore platforms capable of supporting large-scale AI computing. A series of 2026 announcements has moved these facilities beyond the purely conceptual stage, with projects ranging from 30 MW modular barges to 200 MW hyperscale floating designs. The shift reflects a fundamental challenge facing the AI industry: land-based data centers struggle to secure sufficient electrical power, freshwater, and suitable locations, while communities increasingly resist the noise and environmental impacts of conventional facilities.

Why Are Tech Companies Building Data Centers on Water?

The ocean offers solutions to multiple constraints that have become critical bottlenecks for AI infrastructure. Data center developers typically search for land that can obtain sufficient utility service from the electrical grid, a process that can take years due to interconnection delays and transmission limitations. Under the floating model, facilities can be positioned directly beside existing power generation assets, potentially eliminating dependence on lengthy grid interconnection processes.

Cooling represents another major advantage. High-density AI hardware generates extreme heat, and conventional air-cooled systems consume vast quantities of freshwater, a practice that has triggered community opposition in water-scarce regions. Floating facilities can use seawater as a heat sink without exposing sensitive equipment to corrosive saltwater. The design uses a closed internal cooling loop where coolant absorbs heat from processors and carries it to heat exchangers, while a separate seawater system transfers that thermal load to the surrounding ocean. This approach eliminates evaporative cooling entirely, removing one of the most controversial aspects of large data center development.

The noise profile also improves significantly. Because the design eliminates much of the large air-cooled chiller and fan infrastructure associated with conventional heat rejection, floating facilities are expected to have substantially lower noise profiles, addressing community resistance that has become increasingly prominent in data center development.

Which Companies Are Leading the Floating Data Center Push?

Samsung Heavy Industries and Mousterian Corporation represent the most advanced announced U.S. project. In August 2026, the companies signed an engineering contract moving their partnership into basic design, detailed engineering, and production design phases. Each purpose-built unit is designed to provide 50 megawatts of critical computing capacity, with deployments contemplated in Texas and other U.S. markets. This represents a significant scale, not a demonstration project. A 50 MW computing load could support a substantial AI deployment with high-density liquid-cooled infrastructure.

The partnership represents a fundamentally different approach to data center construction. Rather than constructing buildings sequentially on conventional sites, Samsung fabricates the floating structure and integrates much of the electrical, mechanical, and cooling infrastructure in a shipyard. Site work at the eventual mooring location can proceed simultaneously, compressing one of the longest parts of the data center development schedule.

"Signing this engineering contract transitions our partnership with Samsung Heavy Industries from intent to execution and gives us a clear path to scaling factory-built critical IT capacity for AI-class facilities. Because the facility is built in a shipyard, fabrication advances in parallel with sitework rather than after it, a schedule and scale that conventional delivery methods cannot match," said Min Suh, Chief Executive Officer of Mousterian Corporation.

Min Suh, Chief Executive Officer, Mousterian Corporation

Modern shipyards already operate as enormous industrialized manufacturing environments capable of constructing highly complex liquefied natural gas carriers, offshore production platforms, and other structures containing power generation, electrical distribution, piping, controls, and mechanical systems. Adding a floating data center effectively applies those capabilities to digital infrastructure.

Beyond Samsung and Mousterian, the floating data center ecosystem includes diverse approaches. Seatrium has designed a barge carrying modular 5 megawatt "Data-in-a-Box" units. Scotland's Mocean Energy wants to combine wave power, offshore solar, batteries, and graphics processing unit (GPU) servers in a self-powered offshore facility. Panthalassa is developing an autonomous platform that turns wave motion directly into electricity for AI computation. Startup Atomarine envisions large offshore compute campuses initially supplied by gas-fired power vessels and ultimately by marine nuclear reactors.

How to Evaluate Floating Data Center Feasibility for AI Infrastructure

• Power Proximity: Assess whether the facility can be positioned near existing generation assets, reducing dependence on transmission infrastructure and lengthy interconnection processes that typically delay land-based projects by years.
• Cooling Efficiency: Evaluate whether the design uses closed-loop cooling systems with seawater heat rejection, eliminating freshwater consumption and reducing the environmental and community opposition that plague conventional facilities.
• Construction Timeline: Determine whether shipyard fabrication can proceed in parallel with site work, compressing development schedules compared to sequential conventional construction methods that extend projects by months or years.
• Regulatory Classification: Confirm that the facility has engaged classification societies like the American Bureau of Shipping (ABS) to establish technical baselines and ensure compliance with maritime and electrical standards.
• Equipment Validation: Verify that AI server infrastructure has undergone operational verification in marine environments, where vibration, vessel inclination, salt-laden air, and rapid humidity changes can affect equipment reliability and lifespan.

Samsung has also been building a broader development ecosystem around floating data centers. In June 2026, the shipbuilder signed agreements with Greece-based Capital and Lloyd's Register covering project development, investment sourcing, and regulatory requirements, while Lloyd's Register Advisory is working with Samsung on North American market analysis, infrastructure assessments, and commercial feasibility.

Samsung entered a joint development project with Supermicro to validate AI server infrastructure for offshore conditions. Samsung will develop positioning-control and salt- and humidity-protection technologies while Supermicro conducts operational verification of AI server infrastructure in river and marine environments.

What Role Do Floating Data Centers Play in Broader AI Infrastructure Trends?

The emergence of floating data centers reflects a broader shift in how the AI industry approaches infrastructure challenges. U.S. AI data center expansion in 2026 is increasingly focused on grid connectivity, high-density computing, advanced cooling, power availability, land, and equipment capacity. The Federal Energy Regulatory Commission (FERC) launched proceedings in June 2026 to examine grid connections for large loads such as data centers, recognizing that today's large loads can be significantly larger and more concentrated than traditional demand and may change their electricity consumption rapidly.

The U.S. Department of Energy announced the COOLERCHIPS 1.5 program in August 2026, supporting development and testing of advanced cooling systems for high-power AI data centers. Project teams are working toward cooling systems capable of handling heat loads of 1 megawatt per rack while testing energy use, cooling capacity, and efficiency. This development highlights how AI infrastructure is changing the engineering requirements of U.S. data centers, with cooling increasingly designed around rack-level AI workloads rather than treating facilities simply as collections of conventional server rooms.

Companies like Oracle, Crusoe, and CoreWeave are pursuing vertically integrated approaches that link AI data center expansion directly with power-generation investments. In September 2026, Oracle announced investments in wind projects expected to deliver more than 1.7 gigawatts of carbon-free electricity in Texas to supply its Abilene data center. Crusoe announced a $3.9 billion Series F financing round to expand its vertically integrated AI infrastructure platform covering energy generation, data center development, infrastructure manufacturing, and AI cloud services, with more than $140 billion in total contracted value.

The infrastructure supporting these facilities is also experiencing unprecedented demand. Trench Group, a Berlin-based manufacturer of transformer bushings and instrument transformers essential to every major high-voltage substation connecting AI data centers to the grid, has seen its order backlog swell to more than 2.1 billion euros (approximately $2.40 billion), roughly twice annual revenue. The company's revenue more than doubled since private equity firm Triton Partners acquired it from Siemens Energy just two years ago. Lead times for high-voltage bushings now run up to two years, with the constraint driven not by assembly-line capacity but by the availability of specialty insulator inputs manufactured by a very small number of globally qualified suppliers.

Floating data centers represent one solution among several emerging approaches to address the fundamental mismatch between AI infrastructure demand and the constraints of land-based development. Whether they become a dominant model or remain a niche solution will depend on regulatory approval, cost competitiveness, and the ability to scale manufacturing and mooring infrastructure. What is clear is that the AI industry's infrastructure requirements have become so demanding that companies are now exploring solutions that would have seemed impractical just a few years ago.