Drogue raises $20M to power floating data centers with ocean thermal energy

Six-month-old startup banks $20 million

Drogue, a Washington, D.C. startup founded just six months ago, has raised $20 million to build data centers that float at sea and draw both their power and their cooling from the ocean, according to TechCrunch. The round drew participation from XYZ Venture Capital, Toyota Ventures, Tamarack Global, Dream, FDVC, Garuda Ventures, Flexcap, Boost VC and Slow Ventures, among others.

The company has already designed a multi-megawatt prototype that will be installed on a conventional oceangoing ship and sent to sea in early 2028. Later generations of the platform are planned for floating structures that more closely resemble offshore oil rigs.

Turning ocean heat into electricity

Drogue's system is based on ocean thermal energy conversion, or OTEC, a technique first developed in the 1950s. It exploits the temperature gap between sun-warmed surface water and the near-freezing water found about 1,000 meters down, where temperatures average roughly 4°C.

As TechCrunch describes it, warm surface water is first run through the data center to cool the servers. The now-warmer water is then used to boil ammonia, a fluid that vaporizes at -28°F, and the resulting gas spins a turbine to generate electricity. After the turbine, the ammonia passes through a condenser chilled by cold water pumped up from depth, returning it to liquid form. The spent deep water is pumped back down to 1,000 meters, where its heat dissipates, and the ammonia loops back to the start of the sealed cycle.

The basic concept was tested in the 1980s and then abandoned. Co-founder and CEO Sam Sipe, a former U.S. Navy pilot who flew E-2D Hawkeyes off aircraft carriers, told TechCrunch he interviewed veterans of those early programs and found consistent answers: the physics worked, but nobody had financed a plant large enough to reach the scale where OTEC becomes economical. He argues that modern manufacturing of pipes and heat exchangers has since changed the calculus, and that deep-water temperatures are stable enough that the temperature differential will not collapse.

Why data centers are the right customer

Sipe's core argument is about timelines. The data center industry, he says, is essentially a real estate business built on 8-to-10-year horizons, including the slow work of interconnecting with utilities and regional grid operators. AI demand has grown roughly tenfold in the past three years, a pace he considers fundamentally incompatible with that model.

Ocean platforms also answer OTEC's other historical weakness: moving power to land would require long, expensive undersea cables. Data centers instead consume the electricity where it is generated. Servers are compact relative to other industrial loads, and the product travels back to shore almost for free. Each vessel would stay at sea for about a decade before returning.

Drogue has secured approval in principle, a preliminary safety sign-off, from the American Bureau of Shipping, and is now purchasing its first ship to begin building the prototype. The resource itself is vast: TechCrunch cites figures that warming oceans absorb roughly 5.8 million terawatt-hours of additional energy every year, against about 180,000 terawatt-hours of annual consumption by all of human civilization.

Why it matters

Power availability, grid interconnection and cooling have become the binding constraints on AI infrastructure. A floating platform that generates its own electricity while sitting beside an effectively unlimited cold-water supply attacks all three at once, and it could let operators add compute on shipping timelines rather than utility timelines.

The obstacles are equally real. Waves, corrosion and biofouling are standing challenges in marine engineering, and the 1980s attempts faded for reasons that capital alone may not fix. The concept also invites the comparison TechCrunch draws to proposals for orbital data centers — chase cheap energy and place the servers right beside it — except a ship can simply sail out rather than ride a rocket. Whether Drogue's 2028 prototype survives the open ocean will say a lot about whether the sea becomes the next frontier for compute.