Endurance Energy Proves Subsea Geothermal Can Run Net-Positive

The turbine spun at 26,000 rpm. The vent ran at 281 °C. For 8.7 seconds, the plant on the seafloor made more electricity than it consumed.

That is the entire result from Endurance Energy's Adélie test: about 60 watt-hours over a 29-second run, a median of 2 kilowatts net and a peak of 16.5 kilowatts, measured nearly a mile beneath the Pacific.

The duration invites ridicule. It is also the first time a geothermal power plant has been landed on the deep seabed, connected to hydrothermal fluid and run net-positive under remote control.

Endurance is a Seattle company founded in January 2025. It has raised $63 million, including a $54 million Series A led by Founders Fund, and it plans to sell its first commercial power in Tonga in 2027. The pitch is baseload electricity from subsea volcanoes, built faster than nuclear and without the land constraints that slow conventional geothermal.

The test moved that pitch from concept to engineering problem. The engineering problem is not small.

What 8.7 seconds actually bought

Adélie is a 100-kilowatt organic Rankine cycle generator: a three-ton package that boils a working fluid with heat from hydrothermal water, then spins a turbine with the vapour. Endurance lowered it to the Axial Seamount, an active volcano on the Juan de Fuca Ridge roughly 300 miles off Oregon, in about 1,500 metres of water.

Engineers in Seattle ran the plant over the National Science Foundation's Regional Cabled Array, a seafloor fibre network operated by the University of Washington. A robot connected the unit to a natural vent. The turbine reached operating speed. A battery protection circuit then tripped and cut the run.

Ten minutes later, a power cable feeding the refrigerant pump failed during a restart attempt. That ended the dive.

The run was short, but it answered the narrow question the company set out to answer. A plant can be landed, connected, controlled from shore and produce net power at depth. The failures were in auxiliary systems, not in the geothermal loop.

Endurance recovered, serviced and redeployed the unit in under 16 hours. That number matters as much as the 8.7 seconds. Subsea equipment that cannot be repaired quickly becomes a liability, and the company's fifth deployment in just over a year was its first full rehearsal of the repair cycle.

The well is the hard part, and the well underdelivered

The milestone came from a natural vent, not from the well Endurance drilled. That distinction is the most important technical caveat in the release.

The lander drill bored two holes less than two feet apart at about 5,000 feet below the surface. Both hit hydrothermal fluid. The team cased and cemented the second hole, capped it with a valve and completed a 52-foot well. After cementing, it flowed at 260 °C.

It did not flow fast enough to run the plant. Endurance moved Adélie to a nearby vent at 281 °C to finish the test.

The binding constraint is flow. Temperature is abundant; a 16-metre well in fractured volcanic rock is a straw. The company's current drill can reach 80 metres, and deeper wells should intercept more permeable fractures. Whether they do, at a cost that supports commercial generation, remains unproven.

The site also imposes costs that onshore geothermal avoids: saltwater corrosion, high pressure, mineral precipitation inside heat exchangers, and the price of sending a ship and a remotely operated vehicle to fix anything that fails. Endurance calls these lessons, which is the right framing for a test article.

One more detail from the release deserves attention. The well was completed with a valve and flowed on its own. A subsea geothermal field would need several such wells feeding one plant, or one very good well, to run continuously. The drilling record so far is two holes, both live, one unusable for the plant. That is a small sample.

Tonga is the real test case

Endurance has a memorandum of understanding with the Tongan government to supply Tongatapu, the main island, and is targeting first power in 2027 with Leopard, a 2-megawatt commercial generator.

Tonga is a sensible first market. Its electricity is among the most expensive in the world, at roughly $0.35 per kilowatt-hour in 2021, and most of it comes from imported diesel. Fuel price shocks pass straight through to consumers, and the country has spent years trying to displace them with solar, batteries and efficiency programs.

Two megawatts is small in absolute terms. At a 90% capacity factor it would produce about 15.8 gigawatt-hours a year, roughly a quarter of Tonga's electricity sales, which run around 60 GWh annually. For a single generator, that is material.

The economics will be decided by the cable. Endurance has not published the route, the length or the cost of the subsea link to Tongatapu. The company says it will choose sites by optimising cable cost against resource quality and onshore demand. Some prospects are a few dozen miles offshore; others are hundreds.

That optimisation is the business model. A subsea field with high-grade heat close to a diesel-dependent grid can compete. A field far from shore, or one that needs a long high-voltage cable and a new converter station, may not.

It is adding so much technological challenge to an energy production technology.— Emily Pope, senior fellow, Center for Climate and Energy Solutions

Pope also argues that remote islands could meet their needs more easily with solar and batteries. That is the competitive threat in its simplest form: a mature alternative that a utility can finance today, against a novel baseload source that has run for seconds.

The addressable market is large if the model works. Andrew Redd, the founder and chief executive, estimates about 6 terawatts could be developed around the Ring of Fire within five to ten years. That is a company estimate, not an independent one, and it assumes costs fall along a curve that has not yet started.

Terrestrial geothermal just set a harder benchmark

Endurance is not competing with diesel alone. It is competing with geothermal on land, which has just had its own commercial breakthrough.

Fervo Energy declared commercial operation at Cape Station in Utah on 30 September, the first greenfield enhanced geothermal project to reach contractual commercial operations. Fervo drills deep horizontal wells and fractures hot rock, borrowing the oil industry's playbook. Its costs are falling as it repeats the process.

The cost trend is not uniformly friendly. IRENA's 2025 cost report put geothermal LCOE up 53% year on year, even as utility-scale battery costs fell about 30%. Enhanced geothermal well costs at some US projects have fallen by more than half, and the IEA projects next-generation costs could fall up to 80% as drilling matures.

Qualitative comparison. Fervo's Cape Station reached commercial operation on 30 September 2026; Endurance's Adélie ran net-positive for 8.7 seconds on 27 September.

The table is unkind to the subsea case on transmission and repair, and generous on heat grade and siting. Endurance's answer is that the ocean offers high-temperature reservoirs close to coastal demand, without the land assembly and permitting that delay onshore projects.

What would make this investable

Three numbers will decide whether Endurance becomes a power producer or a science project.

The first is run length. Adélie ran for seconds. Gentoo, the upgraded unit planned for this winter, needs to run for days, then weeks, without a support vessel alongside.

The second is installed cost per kilowatt for a full 2-megawatt unit, including the well, the plant and the cable. Published offshore geothermal projections have used figures around $4,000 per kilowatt, an overnight cost that would make the technology competitive with new firm capacity if it holds. No commercial subsea unit has been built, so the number remains a projection.

The third is availability. A subsea plant that needs a ship every few months will not beat diesel. Endurance's 16-hour turnaround is a good sign; a full maintenance campaign that takes weeks is not.

The bet, stated plainly

Subsea geothermal is a bet that the ocean's heat is easier to reach than the land's politics. The physics is real. The engineering is now demonstrated at the smallest possible scale.

What remains is the unglamorous part: deeper wells, longer runs, cheaper cables and a repair record. Endurance has two years and one commercial project to show that those can be bought, not invented.

What happens if Gentoo clears a week of continuous operation?

Sources