“When I circled the moon and looked back at Earth, my outlook on life and my viewpoint of Earth changed… Earth is a spaceship, just like Apollo – and just like Apollo, the crew must learn to live and work together. We must learn to manage the resources of this world with new imagination.”
Jim Lovell, Apollo 8
Drogue is determined to change how the world generates and uses electricity. The ocean is the largest, untapped energy source on the planet, and we’ve set out to build an entirely new economy on it. We envision a future where it’s obvious for energy intensive activities to exist offshore with colocated power generation. AI data centers are only the best place to start.
Terrestrial data center expansion was already constrained before AI; now it's completely deadlocked. Companies are going to great extents to take over existing data center projects for model training and inference, CEOs are complaining about closets full of stranded GPUs, diesel and natural gas generation (required to build grid-connected data centers) have a 9 year backlog, and regulatory frameworks for grid interconnection are fragmented, resulting in multi-year delays across the United States and in many international markets. The largest constraint on building new data centers isn’t construction, land, water, or compute, it’s electricity.
Truthfully, data centers are the canary in the coal mine, because of the industry’s growth rate and high demand for AI. This along with other load growth has put a massive strain on the electric grid, and on electricity production. It’s bad enough that the White House declared a national energy emergency—we’re no longer talking about replacing natural gas, and we’re turning old coal plants and nuclear reactor facilities back online. Data center and AI-driven tech companies have walked back net zero aspirations as the industry is now projected to triple their energy consumption by the end of the decade. This will represent up to 12% of all of the power we generate in the United States—we are seeking the age of abundance, but energy is by far the largest constraint on AI.
Investing in renewable energy sources that do not require extensive land use, finite fuel sources, or exotic raw materials—and that can scale to megawatts quickly—offers a clear path out of this supply problem. Taking loads off the grid entirely is also paramount. Otherwise wholesale energy prices will continue to rise and the economics of AI, electric cars, and every industry will suffer. Every single economic process is constrained by readily available, cost-effective electricity.
70% of the Earth is water, and the ocean is by far the largest solar energy collector on the planet. More than 90% excess heat trapped by the atmosphere finds its way into the ocean, as its capacity to absorb this energy is immense. This represents a massive opportunity and there is enough ocean thermal energy to power all of human activity, assuming it can be converted into electricity. This climate system represents one of many available ocean energy resources, both renewable and non-renewable.
Getting non-renewable energy out of the ocean is challenging and expensive but the oil & gas industry is booming, though it’s finite resource and companies have had to go further and deeper to get the same amount of energy to shore. Offshore wind has grown but has faced seafloor use issues, regulatory pressures, and starts to become uneconomic at certain depths. Sometimes the wind doesn’t blow or it blows too hard, and they have to run long sub-sea cables back to shore. Solar, tidal, and wave energy have the same headwinds but aren’t as energy dense. Nuclear power at sea has been in use by the Navy for 70 years, and is promising but takes a long time to deploy, mired in regulatory bottlenecks.
Another solution is to use the stable thermal gradient (thermocline) of the ocean to make unlimited, clean, dense, baseload power. This technology has been around for a long time, has already been deployed at grid-scale, and has been continuously subject to ongoing techno-economic analysis, with levelized cost of electricity (LCOE) figures that make it viable for commercial deployment at a much larger scale. There are even few land based ocean thermal plants that have been operating continuously for over 10 years each producing around 150kW each.
Ocean thermal power works in a large swaths of the ocean but is most ideal in areas where the thermocline is the steepest. In simplest terms, it uses this temperature differential to operate a Rankine cycle which generates mechanical energy that produces electrical energy with a generator.
A funny word, but an ambitious company. Drogue is seeking funding to design and build a prototype 2.5+ MW ocean thermal energy platform. This is about as much power as 2,000 US households, and is widely regarded as the minimum size needed to demonstrate the technology over time. This prototype will also have a small data center and ground station onboard to utilize the energy.
The two main moats we hope to achieve are density and speed to power. Density, or how much power is produced per platform, will make it easier to work with data center customers and scale in a way that is bankable and insurable. Speed to power is the biggest buzzword in the energy industry right now, but it’s Drogue’s biggest opportunity. The components of this system are common and won’t be bottlenecked by supply chain issues or long backlogs, vertical integration over time would assure that we can make our own components to scale with demand, and deploy quickly. Put a different way, this is truly additive and is not taking someone else’s kWh off the grid.
Drogue envisions itself as the default energy provider for any enterprise that is constrained by the amount of energy they can pull from the grid. We will be able to provide power, cooling, water, and real estate on the high seas to a number of tenants and will meet the needs of many customers, not just data centers and AI companies. This means that in the near term the company does not plan to run its own cloud offering and instead partner with companies who do. This is also very similar to the terrestrial data center business today and what the majority of hyper-scaler cloud providers and AI companies will expect.
Drogue is differentiated because it is focused on building and owning the lowest possible levels of infrastructure in the new energy focused economy. It’s not a data center company, but the first most economically viable use case happens to be data centers for AI inference.