Startup Drogue raises $20M to power floating data centers with ocean heat

A six-month-old Washington, D.C. startup called Drogue has raised $20 million to build data centers that float on the ocean and run on the ocean's own heat. The company says its first multi-megawatt prototype will be installed on a traditional oceangoing ship and sent out to sea in early 2028.

The round, which Drogue disclosed exclusively to TechCrunch, includes investments from XYZ Venture Capital, Toyota Ventures, Tamarack Global, Dream, FDVC, Garuda Ventures, Flexcap, Boost VC, and Slow Ventures, among others. For an AI industry desperate for power on timelines that traditional utilities can't meet, that's a meaningful bet on a technology the energy world largely abandoned in the 1980s.

Why does the ocean matter for AI data centers?

The short version: speed. Co-founder and CEO Sam Sipe argues that data center construction and AI demand are on a collision course because they run on completely different clocks.

"The data center industry is a real estate game, and so they're playing on 8- to 10-year timelines, including interoperability with utilities and [independent system operators]" that manage regional power grids, he said. "AI demand has increased 10x in just the last three years. It's completely incompatible."

That mismatch is the market opening. If a floating vessel can generate its own power at sea, a data center developer doesn't need a grid interconnection, a land parcel, or a decade of permitting. The ship simply sails out and starts crunching.

Drogue's answer to the power problem starts with a number that dwarfs everything humans do. As the planet warms, the oceans absorb an additional 5.8 million terawatt-hours of energy every year. All of human civilization uses about 180,000 terawatt-hours annually. The resource, in other words, is enormous relative to demand.

How does the technology actually work?

Drogue's system is essentially a heat pump running in reverse. A normal heat pump uses electricity to move heat from one place to another; Drogue's setup uses a temperature difference to generate electricity instead. The approach is called ocean thermal energy conversion, or OTEC, and it was first developed in the 1950s.

The mechanism relies on a simple fact of the sea: surface water is warm because the sun heats it, while water far below is cold, averaging about 39°F (4°C). That vertical temperature gap is the fuel.

Here's the full cycle, step by step:

• The system draws in warm surface water and uses it to help cool the data centers on board.
• The now-warmer water boils ammonia, a chemical that turns to gas at -28°F.
• The ammonia gas drives a turbine, generating electricity.
• The gas then passes through a condenser, where cold water pumped up from 1,000 meters deep flows over the fins and absorbs the heat, turning the ammonia back into liquid.
• The warmed water is pumped back down to 1,000 meters, where it dissipates into the ocean, and the ammonia loops back to the start of the sealed cycle.

Sipe isn't worried about running out of thermal headroom. "There's basically no amount of heat we can pull out of that that will collapse that temperature differential," he said.

Why now, if the idea dates to the 1950s?

OTEC was tested in the 1980s and then tossed aside. Sipe's explanation is that the physics worked but the hardware and the economics didn't.

"A lot has changed since the '80s, specifically manufacturing techniques around the pipe and the heat exchangers," he said.

Before building anything, Sipe tracked down the engineers who had worked on those early efforts and interviewed them. "I just talked to everyone, basically interviewed them and figured out what went wrong, what went right, what would you do next. And everyone was like, dude, this stuff works. It's just a matter of being able to raise enough capital to build something big enough, because there's a sweet spot," he said.

Two conditions had to line up for the idea to make commercial sense. First, the equipment had to be buildable at the right scale. Second, there had to be a customer who could use the power while still at sea — because sending electricity to shore would require long, expensive cables, and doing manufacturing on board would mean costly shipping of materials and finished goods.

Data centers fit that slot almost perfectly. Servers are compact relative to other industrial equipment, and moving the output — data — off the platform is incredibly cheap. Each Drogue vessel would stay at sea for about a decade, according to Sipe.

Who is behind this, and how far along is it?

Sipe's path to the ocean ran through the U.S. Navy, where he spent years flying E-2D Hawkeyes off aircraft carriers. That experience gave him a direct feel for how much energy the sea holds, and a different reaction than most people have to it.

"Most people think, oh, ocean — hard waves, storms, corrosion, biofouling. And I just see this majestic place and resource that we just ignore," he said.

Drogue has moved quickly since its founding. In six months, the team has designed a multi-megawatt prototype destined for a conventional ship, with future versions planned for floating structures that resemble offshore oil platforms. The company has also secured "approval in principle" — a preliminary sign-off — from the American Bureau of Shipping, the body that sets and checks safety standards for ships. It is now in the process of buying its first ship to begin building the seagoing prototype.

How does this compare to other wild power ideas?

In structure, Drogue's pitch mirrors Elon Musk's proposal for orbital data centers: find a cheap, abundant source of power and put the computing as close to it as possible. The difference is the delivery cost. Orbital data centers need expensive rocket launches. Drogue's just need a ship to sail them out — something humanity has done for millennia.

The stakes for the broader market are straightforward. AI power demand is growing faster than land-based grid capacity, and developers are now weighing everything from nuclear restarts to gas turbines to fill the gap. A floating, self-powered data center that can deploy in roughly two years instead of eight to ten would remove the grid from the equation entirely.

Drogue still has to prove the 1980s-era technology works at commercial scale, survive the corrosion, storms, and biofouling that skeptics point to, and then build a fleet. The first real test comes in early 2028, when the prototype ship leaves port.