Drogue raises $20 million to build floating data centres powered by the ocean

A startup emerging from stealth

A young company called Drogue is working on a way to run data centres at sea, using heat drawn from the ocean itself as the power source. According to TechCrunch, which reports the company had kept a low profile until now, the Washington, D.C. startup was founded only six months ago and has already raised $20 million.

The funding came from a range of backers, including venture firms with links to the automotive, defence and climate technology worlds. Co-founder and chief executive Sam Sipe said the first prototype should be operating on the ocean in less than two years.

How the ocean thermal system works

The technology relies on a process known as ocean thermal energy conversion, or OTEC, an idea first developed in the 1950s. It works like a heat pump in reverse: instead of using electricity to move heat, it uses a temperature difference to make electricity.

Warm water from the ocean surface is used to cool the servers on board. That warmed water then boils ammonia, which turns to gas at a very low temperature. The gas drives a turbine that generates electricity, and cold water pumped up from about 1,000 metres down condenses the ammonia back into liquid. The cycle then repeats in a sealed loop.

Sipe told TechCrunch that the deep water sits at roughly 4 degrees Celsius, and that the temperature gap between surface and depth is far too large for the system to upset. He said modern manufacturing techniques for pipes and heat exchangers have improved since the concept was tested and abandoned in the 1980s.

Why data centres belong at sea

Earlier attempts at ocean thermal power struggled partly because sending electricity to shore needs long, costly cables, and manufacturing at sea means shipping materials back and forth. Drogue argues data centres avoid both problems: servers are compact, and moving data off a vessel is cheap.

Sipe said the data centre industry works on real estate timelines of eight to ten years, including negotiations with utilities and grid operators. He argued that AI demand has grown tenfold in three years, which he described as incompatible with that pace. Each vessel is expected to stay at sea for about a decade.

TechCrunch notes the idea echoes plans for orbital data centres: place computing next to a cheap power source. Unlike space-based schemes, though, a ship can simply sail to its position.

Prototype plans and regulatory progress

Within its first six months, Drogue has designed a multi-megawatt prototype that will be fitted to a conventional oceangoing ship and sent to sea in early 2028. Later versions are expected to use floating structures that look more like offshore oil platforms.

The company has received approval in principle from the American Bureau of Shipping, the body that sets and checks safety standards for ships. This is a preliminary sign-off rather than full certification. Drogue is now buying its first ship to begin building the prototype.

Sipe, a former Navy pilot who flew E-2D Hawkeye aircraft from carriers, said his time at sea shaped his view of the ocean as an overlooked resource. He interviewed people from the early OTEC projects and concluded the concept works, but that earlier teams could not raise enough capital to build at the right scale.

What it means for AI infrastructure

The oceans absorb an enormous amount of energy each year as the planet warms, far more than all of human civilisation consumes annually, according to figures cited by TechCrunch. Tapping a small share of that could, in theory, power computing without competing for grid electricity.

The plan remains unproven at scale, and the first real test will come when the prototype reaches the ocean in 2028. If it works, floating data centres could offer developers a faster route to capacity than waiting for land, grid connections and construction permits.