WangDou logo
WangDou

Six months old and $20M raised, this startup wants to power floating data centers with ocean temperature differences

2026-11-15·WangDou AI Express·AI Compute / Energy / Data Centers

While everyone else worries about the power bill, someone moved the servers out to sea, where the surface is warm and the depths are cold.

Three key facts

Drogue is a Washington, D.C. startup founded six months ago that had kept a low profile until now. It is building a system that harvests heat from the ocean to power data centers aboard floating vessels. Founder and CEO Sam Sipe said the first multi-megawatt prototype should be at sea and running within two years. The company told TechCrunch exclusively it has already raised $20 million, from investors including XYZ Venture Capital, Toyota Ventures, Tamarack Global, Dream, FDVC, Garuda Ventures, Flexcap, Boost VC and Slow Ventures.

Sipe's entry point is a timing mismatch. The data center industry is a real estate game running on 8- to 10-year cycles involving utilities and independent system operators, while AI demand has grown tenfold in three years. The technology works like a heat pump run in reverse, using a temperature difference to generate electricity, a concept known as ocean thermal energy conversion, first developed in the 1950s. Surface water is warmed by the sun while deep water averages about 4 degrees Celsius, and that gap stays stable.

The process: warm surface water cools the data centers, then the now-warmer water boils ammonia, which turns to gas at minus 28 degrees Fahrenheit. The vapor drives a turbine to generate power, then passes a condenser where cold water drawn from 1,000 meters deep absorbs the heat, returning the ammonia to the start of a sealed loop. The basic idea was tested in the 1980s and abandoned; Sipe argues manufacturing techniques for pipes and heat exchangers have advanced considerably since.

Wang Dou's Take

Start with one number. The oceans absorb an extra 5.8 million terawatt-hours of energy every year, while all of human civilization uses about 180,000 terawatt-hours. The ocean's surplus is more than 30 times global annual consumption. That figure does not prove the plan works. It proves that running out of resources was never the problem.

The hard part has always been the engineering ledger. A technology tested in the 1980s and thrown away is being revived four decades later because pipe and heat exchanger manufacturing improved. That is an honest answer, and it also shows the company is nowhere near being able to say what a kilowatt-hour will cost.

But Sipe's best line is the mismatch itself: demand up tenfold in three years against an 8-to-10-year property cycle. That is the real selling point. Not cheaper power, but faster power. When the grid queue runs five years, sailing a ship out is the fastest option available.

Six months, $20 million, a prototype at sea in 2028. Industrial capital like Toyota Ventures is most likely not betting on ocean thermal conversion itself, but on the idea that compute will go wherever the power is.

Source: TechCrunch

Comments

Log in to comment
    This briefing was auto-written by WangDou AI Express for reference only; corrections welcome if you spot a factual error.
    指挥舱👽