Undersea Geothermal Power ⚡: First Test Results (2026)
Discover the first results of undersea geothermal power—subsea geothermal energy for Pacific island energy. See what worked, what failed, and next steps for a geothermal power plant.
Can Undersea Geothermal Power Work at Commercial Scale?
Island grids face a pretty blunt trade-off: keep burning imported fuel, which is costly and exposed to price swings, or lean on renewables that often can't produce power around the clock. Undersea geothermal power tries to split that difference - pulling usable heat from beneath the seafloor to generate electricity near coastal demand centers.
In September 2026, Seattle startup Endurance Energy tested a prototype at Axial Seamount, an active underwater volcano in the northeast Pacific. The three-week deployment didn't deliver sustained commercial power. What it did do is clear a small but meaningful technical hurdle: for about 8.7 seconds, the prototype produced more electricity than it used.
That's encouraging, but it's not a business case. The test also surfaced issues with well flow, battery monitoring, and a subsea cable. Here's what the demonstration established, what's still unclear, and how to judge what Endurance does next - especially with its proposed project in Tonga.
What the first undersea geothermal test achieved
Endurance Energy deployed its three-ton prototype, Adélie, at Axial Seamount in water about 5,000 feet (1,524 meters) deep. The site gave access to hot hydrothermal fluid. But this wasn't a "just generate power" exercise. The company also needed to test a chain of steps that usually happen on land: drilling a well, sealing and completing it, linking it to a power system, and running the whole setup remotely in a harsh environment.
The team used a subsea lander drill launched from the vessel Ocean Guardian. It drilled two holes in the seafloor. Hydrothermal fluid at roughly 500°F (260°C) vented from both. The crew cased and cemented the second hole and fitted a valve to create a well. A remotely operated vehicle, ROPOS, helped connect Adélie to it.
Adélie was designed for a rated capacity of 100 kilowatts. Conceptually, the plant takes hot incoming fluid, uses it to vaporize a working fluid, spins a turbine, and drives an electric generator. In this run, the well didn't provide enough hydrothermal flow to operate the plant as intended.
That matters because a generator's nameplate capacity tells you what it's designed to produce under suitable conditions. It doesn't tell you what the whole system can deliver reliably when the resource and hardware don't cooperate. Endurance therefore shifted part of the test to a nearby natural vent where hot fluid was already flowing. The company described placing the plant there as a one-time research and demonstration choice, not a plan for commercial installations.
What "net-positive" means - and what it does not
On September 27, Adélie's turbine reached about 26,000 revolutions per minute. The generator produced 60 watt-hours in total. For roughly 8.7 seconds, it generated more power than the equipment used to run itself. During that brief interval, the reported median net output was around 2 kilowatts, with a peak of 16.5 kilowatts.
"Net-positive" is a useful engineering checkpoint. Even a geothermal system has to spend energy on its own pumps, controls, and other equipment before it can deliver a surplus to the grid. But an 8.7-second surplus doesn't automatically translate into dependable electricity. It doesn't establish continuous output, year-round availability, maintenance intervals, or the cost per unit of power.
The honest reading is narrower: during this test, the components could work together and produce a short net surplus under the particular conditions they used. The real question is whether the system can repeat that result, run longer, and do it using a purpose-built well - not just a natural vent.
Why drilling and operating underwater is difficult
This demonstration shows why moving geothermal offshore isn't just a matter of putting a land-based power plant on the seabed. The seafloor changes what you have to do for drilling, installation, maintenance, and safety. Each complication hits reliability and cost.
The test well was about 16 meters (52 feet) deep and only 2 inches (5 centimeters) wide. Endurance said it kept the first well intentionally small and shallow to prove it could drill, case, cement, and cap a well on the seafloor. The narrow, shallow design limited how much fluid it could supply. The company's current lander drill is built to reach about 260 feet (80 meters) - about five times deeper than the test well.
That's a meaningful planned step forward, but depth by itself doesn't guarantee good flow. Commercial performance will depend on what the team finds underground, how the well is designed, how well it stays operational, and how much fluid volume and temperature actually reaches the plant. Deeper drilling may improve access to the resource, but it also brings more engineering work, more equipment needs, and higher costs.
Reliability includes the whole system
The well wasn't the only constraint. Adélie initially used a battery to run its pumps, thrusters, and controls, and to bring the turbine up to speed before the generator took over. During start-up, a battery module logged a brief current spike and then stopped reporting. The control software kept relying on the last reading, and then triggered an automatic shutdown as a safety response.
That's protective behavior, but it also highlights something people sometimes gloss over: software decisions are part of the physical system's reliability. A control system has to fail safely when sensor data goes missing or goes stale, without turning what might be a manageable event into something you can't diagnose or recover from. In a remote offshore installation, there's no quick "open the cabinet and swap the part."
A separate issue ended the restart attempt. A defect in the casing of a cable feeding the refrigerant pump let seawater in and short two electrical conductors. Endurance said it plans to use the incident to improve how it builds and tests subsea cables.
These aren't good reasons to write off the idea. Early demonstrations are supposed to surface failures like this. They do serve as a reminder that a commercial system has to perform as well as its weakest link - whether that's drilling, seals, power electronics, controls, cables, communications, or the generator itself.
The ocean helps - and adds operational burden
Endurance points to two possible advantages of working at depth. Cold seawater can provide cooling, and the surrounding pressure can help keep the system and well pressurized. Those factors may make a more compact plant design plausible.
But the same environment makes installation and servicing harder. Equipment has to be transported offshore, lowered to depth, connected, monitored, and then retrieved or repaired using specialized vessels and subsea tools. Stanford energy science and engineering professor Roland Horne also cautioned that subsea geothermal is likely to be more expensive than land-based geothermal, even if the technology works technically.
So the core engineering trade-off is straightforward: a favorable thermal resource and a compact system don't automatically cancel out the costs of operating at depth. Reliability has to be shown over time, not inferred from a successful moment during a test run.
The commercial case for island grids
The business case is clearest where electricity depends heavily on imported fuel. Tonga - Endurance's first planned commercial customer - relies primarily on imported diesel. When fuel prices rise, electricity generation costs can track upward, and the country's exposure to global supply disruptions increases.
Geothermal power has one quality that fits that need: it can provide continuous, weather-independent generation. Solar and wind can reduce fuel use, but their output changes with time of day and weather. A steady geothermal source could complement those renewables, cut fuel burned, or provide stable power where other low-carbon options struggle. But those are potential roles - the prototype hasn't proven them yet.
Endurance says it's building a 2-megawatt commercial generator - about the size of a shipping container - and discussed deploying up to six units offshore Tonga in 2027. A 2 MW unit is twenty times the prototype's stated 100 kW design capacity. That's a big jump. Commercial hardware has to operate consistently, and adding multiple units raises new questions: installation logistics, shared infrastructure, grid connections, and maintenance planning.
What determines whether the price works
A project's economics depend on more than the generator's headline capacity. A serious comparison should include:
- Resource performance: How much usable heat and fluid each well can produce, and how consistently?
- Drilling and installation: What do the wells, vessels, subsea connections, and offshore work cost?
- Availability: How often does the plant generate power, and how long do repairs take?
- Operations and maintenance: Which components need replacement, and can they be serviced without fully recovering the installation?
- Grid value: How much imported fuel does the system actually displace, and how much backup is still required?
Land-based geothermal can already be cost-effective, but Horne described the sector as near the edge of price competitiveness. If underwater construction and maintenance push costs higher, even a technically successful system could still be a bad commercial deal. On the other hand, avoided fuel imports might matter more in places where delivered diesel is expensive or volatile. The answer will vary with local conditions, the quality of the resource, and what alternatives cost.
That's why a customer announcement isn't proof. The only evidence that really counts is operating data: sustained net output, measured downtime, realistic maintenance needs, and total project cost versus the fuel and power options the grid already uses.
What to watch in the next deployments
Endurance says its next trial is planned for winter 2026-2027. The refurbished and upgraded plant will be renamed Gentoo, continuing the company's penguin-inspired naming. The stated goal is higher power generation. The key metric won't just be a higher brief peak - it'll be how long the system can run reliably and how much net energy it produces after subtracting its own energy use.
If you want a practical next-test scorecard, it should include:
- A drilled-well run: Show the plant can operate from a well rather than relying on a natural vent.
- Sustained net output: Report how long it ran, average and peak output, and the energy used by pumps, controls, and other plant equipment.
- Fault recovery: Explain how the system handles missing sensor data, cable faults, and safe restarts without breaking protection logic.
- Repeatability: Make sure the results aren't a one-off tied to unusually favorable site conditions.
- Operating cost and maintenance: Track what equipment and vessel time it takes to keep the installation running.
Environmental monitoring needs to be part of the operating plan, not an afterthought. For the Axial Seamount test, Endurance said it chose a site it considered less disruptive, sampled water, recorded organisms on the seafloor, compared observations with independent scientific sensors, and sent water, rock, and mineral samples for laboratory analysis. Those steps describe monitoring activity - but they don't, by themselves, establish what a commercial-scale project would do over the long term.
A larger installation also demands an assessment that matches its scale and location. That means looking at construction impacts, changes to local habitat, discharge and water conditions, and ongoing effects - not just documenting baseline conditions before equipment arrives. Independent review is especially valuable because both the developer and local communities need evidence they can evaluate.
Endurance's Tonga plan will be a much more consequential test than a short prototype run. If the company moves forward, the project should be judged using a straightforward framework: Does it deliver dependable power? Does it lower the grid's total cost and reduce fuel exposure? Can the equipment be maintained safely and predictably? And are the environmental impacts understood and acceptable?
Conclusion: a technical milestone, not a power solution yet
The Axial Seamount trial showed that a subsea geothermal system can briefly reach net-positive generation. It also exposed how far that is from dependable commercial power: the well supplied too little fluid, battery monitoring triggered a shutdown, and a cable fault ended the run.
These results matter because they make the next engineering questions concrete. Endurance now has to demonstrate sustained output from a drilled well, reliable control and cable systems, maintainable operations, and economics that beat the alternatives in the markets it's targeting.
For island grids, the upside is real. Local, continuous generation could reduce dependence on imported fuel. But potential isn't the same thing as a working project. Watch the next trial for run time and net energy, then judge any Tonga deployment on measured cost, reliability, and environmental evidence - not on the novelty of putting a power plant on the seafloor.