Drogue Raises $20 Million to Power Floating Data Centers With Ocean Heat
Drogue, a startup building a system that harvests heat from the ocean to power data centers on board floating vessels, has raised $20 million in funding, according to TechCrunch.
The company was founded in Washington, DC, about six months before November 13, 2026. Its co-founder and CEO is Sam Sipe. Investors include XYZ Venture Capital, Toyota Ventures, Tamarack Global, Dream, FDVC, Garuda Ventures, Flexcap, Boost VC and Slow Ventures.
The core technology is ocean thermal energy conversion (OTEC), which generates electricity from the temperature difference between warm surface seawater and colder water at depth. It works much like a heat engine running on any gap between hot and cold. The concept is not new, but it has rarely been paired with a computing load on a moving or floating platform. Drogue's design puts the power generation and the data center on the same vessel, so the electricity is used where it is made.
The timeline has three stages. Drogue has designed a multi-megawatt prototype for a conventional ocean-going ship, with a planned sea trip in early 2028. The company expects its first prototype data center to be operating at sea less than two years after November 2026. Later versions are planned for floating structures resembling offshore oil platforms. These are targets, not operating facts. The prototype has not yet been built, deployed or tested at sea, and the schedule depends on engineering work still ahead.
For infrastructure engineers, the operating envelope matters more than the headline. OTEC needs a usable thermal gradient, which is found mainly in tropical and subtropical waters, so any ocean-hosted site will be limited by geography. Conversion efficiency at multi-megawatt scale is a central open question. Connectivity is another. A data center at sea must reach the wider network over subsea cable or satellite links, and the latency and bandwidth of those links will determine which workloads suit the platform. Training jobs that tolerate interruption differ sharply from inference serving, which runs against tight latency budgets.
Maintenance and reliability are the other unresolved items. Hardware on a ship faces corrosion, constant motion and limited access, and a failure at a remote site carries a logistics cost that a land campus does not. The conventional-vessel prototype is, in effect, a test of whether those constraints can be engineered around before the company commits to purpose-built platforms.
The broader context here is that data center developers are seeking power and cooling wherever they can secure them, and seawater has long been an obvious cooling medium for coastal sites. Drogue's proposal goes further by treating the ocean as an energy source as well as a heat sink. Whether that combination works in practice will be answered by the engineering record that a 2028 sea trial would begin to create.
In my view, the most useful thing to watch over the next two years is not the funding figure but whether the prototype reaches the water on schedule, and what it reports about delivered power, uptime and the cost of running compute offshore. If those numbers hold, operators gain a new option for sites where grid capacity is scarce. If they do not, the investor list will have funded a demonstration rather than a product. Either outcome will be informative, and this publication will have a firmer basis for judging the claims once measured results exist.
