Microsoft experimented with this idea through Project Natick.
The company deployed a sealed data center module off the coast of Scotland in 2018 and retrieved it in 2020.
The experiment was designed to investigate whether subsea data centers could provide benefits in reliability, cooling and deployment.
Microsoft reported that the underwater servers experienced a failure rate one-eighth that of its land-based comparison group during the experiment. The company also reported that cold seawater could be used for cooling without drawing on freshwater resources.
The project also demonstrated how quickly a factory-built underwater module could be deployed.
Microsoft's published figures say the system could move from factory production to operation in less than 90 days.
But Project Natick was a research project, not a commercial rollout.
Microsoft ultimately did not continue with a commercial underwater data center business.
That history is important.
It demonstrates that subsea computing can work technically, but it also shows that technical feasibility alone does not guarantee commercial adoption.
Samsung's floating approach is somewhat different because the facilities are designed to remain accessible rather than being permanently submerged.
AI May Make These Experiments More Relevant
One reason floating and underwater data centers are receiving renewed attention is the rapid increase in AI computing.
The computing requirements of modern AI systems are changing the physical design of data centers.
The IEA notes that the power density of AI servers has increased dramatically and expects further increases in the coming years.
That means cooling systems have to become more capable.
A traditional air-cooled facility may become increasingly inefficient as server racks become more powerful.
Liquid cooling is therefore becoming more important across the data center industry.
Samsung's floating platform is entering this environment at an interesting moment.
The company is not trying to solve a problem that exists only because of AI.
It is trying to solve a problem that AI is making substantially larger.
The Energy Question Remains
There is one issue that floating infrastructure cannot magically solve:
electricity.
A data center still needs enormous amounts of power whether it sits on land or water.
Moving the servers offshore does not reduce the electricity required by the GPUs.
The IEA expects global data center electricity consumption to roughly double between 2024 and 2030 in its base case.
Samsung's concept therefore includes options for different power arrangements.
Its 50 MW design has been presented with the possibility of onboard generation, while the first nearshore projects are expected to rely on shore-based electricity. Samsung has also discussed renewable energy and other power-generation possibilities for future offshore configurations.
This means the floating data center could eventually become part of a larger energy system.
Imagine a coastal computing platform located near offshore wind generation.
In principle, the electricity could be supplied through nearby infrastructure rather than requiring the facility to occupy scarce urban land.
But whether that configuration is economically competitive will depend on local energy prices, transmission capacity, weather, regulation and the cost of the floating facility itself.
Could This Reduce Pressure on Power Grids?
Potentially, but not automatically.
A data center that connects to an already-congested coastal grid can still create significant demand.
However, a floating platform with its own generation capacity could eventually provide more flexibility than a conventional facility.
Samsung's design work includes consideration of onboard power generation, and the company says one advantage of the floating platform is the potential to reduce reliance on onshore electricity.
That could become particularly interesting in regions where land is available but grid connections are limited.
Instead of waiting years for a new terrestrial data center and associated infrastructure, developers could potentially manufacture a standardized platform and deploy it near an appropriate energy source.
That is still a future possibility rather than a proven large-scale model.
But it is one of the reasons the shipping industry is paying attention.
A New Business for Shipbuilders
For Samsung Heavy Industries, the project is about more than data centers.
It is also about finding a new application for shipbuilding expertise.
The global shipping industry is highly competitive and cyclical.
Building infrastructure for the rapidly expanding digital economy could offer shipbuilders another market.
A floating data center is essentially a specialized offshore asset.
It requires structural engineering.
Electrical systems.
Cooling.
Fire protection.
Power management.
Environmental protection.
Marine stability.
Transportation.
And long-term maintenance.
These are areas where a company with decades of offshore engineering experience can potentially have an advantage.
Samsung Heavy Industries has explicitly described FDCs as a new business opportunity that combines its offshore construction capabilities with the digital infrastructure industry.