Innovation

A closed loop drilled 2 and a half miles under Bavaria sends power to the grid without touching underground water, but the 8.2 megawatts everyone quotes come from 4 loops and only 1 is built

By Hugo Rojas · November 17, 2026 · 8:50 AM · 5 min read
Drilling rig at Geretsried's sealed underground loop geothermal site in Bavaria

A wellhead stands on a gravel pad south of Munich with an insulated line running off it.

No steam leaves the site. Nothing is pumped out of the ground and nothing is pumped back into it.

The fluid in that pipe has never touched the rock it was heated by.

It is one loop of four, carrying six lateral pairs instead of twelve, and the reporting says it makes about what the company expected from a loop that size.

Electricity first reached the grid from this site on the fourth of December, 2025.

The number attached to the project is 8.2 megawatts, and it describes a build that does not exist yet.

Which leaves the question the pipe answers. What actually sets the output?

Why 1 loop is just a surface area

A conventional geothermal plant produces hot water and sends it back down. This one produces nothing.

The working fluid runs inside a sealed circuit and is warmed only through the steel wall around it, by conduction out of the rock, the way a radiator works in reverse.

That makes the physics unusually plain. Output is the contact area between pipe and rock, multiplied by the temperature difference across the wall, multiplied by how fast the fluid moves.

Temperature is fixed by the geology and flow is limited by friction, so the only term an operator can buy is contact area, which means drilled length.

Halve the laterals and you have halved the heat exchanger. There is no reservoir to make up the difference, and no exploration risk either, which is the trade the design makes.

13,895 feet and 6 lateral pairs

The hardware is two wells about 25 miles south of Munich.

Each was drilled to 13,895 feet by the trade account of the work, while the developer’s own project page puts the depth nearer 14,760, and the two figures have never been reconciled publicly.

The target is a band of limestone and dolomite where the rock holds heat but almost no moving water. Measured temperatures there ran between 302 and 338 degrees.

From each vertical well the crews kicked sideways and drilled laterals that curve around and meet their opposite number, turning two separate holes into one continuous circuit.

Six lateral pairs were completed. The wellbores were built to carry as many as twelve.

That gap is not a detail. It is half the heat exchanger.

107 planned days against 2 years

The drilling record is where the difficulty shows.

The program initially allowed 107 days. The work ran from July of 2023 to October of 2025.

Downhole tools failed under shock and vibration, pipe stuck in the hole repeatedly, and an offset well took a kick. By the fourth lateral pair the crews concluded the drilling fluid system itself was not adequate.

Connecting two wells thousands of feet apart into one sealed circuit is the part that has no equivalent in oilfield practice, because an ordinary well only has to arrive somewhere useful rather than meet another well.

None of that is a scandal. It is what a first of its kind costs, and it is the number a second project will be priced against.

The drilling was always going to be the whole bill.

What half a megawatt means against 8.2

Here is the figure the announcements leave out.

The completed loop is producing about half a megawatt, in the way that a drilling deal or a water problem turns into a finished plant.

The build behind that number is documented step by step, from the failed downhole tools to the fluid system that had to be rethought, and the drilling account is where the sequence sits.

Read against the design, half a megawatt is not a failure. It is one loop of four, carrying six lateral pairs instead of twelve, and the reporting says it makes about what the company expected from a loop that size.

The cost driver is public too. Cement on the two vertical wells let fluid pass between them, the fix was to run one rig at a time instead of two, that doubled the time and the money for the first loop, and the follow up reports the second loop timeline being revised.

The headline capacity is a plan. The meter reads something smaller.

What the first 1 of its kind settles

Strip the capacity numbers away and something real is left standing.

A sealed circuit in hot dry rock delivered electricity to a commercial grid, which nobody had done before, and it did so in a place where there was no usable underground water to find.

That removes the two risks that have stalled deep geothermal elsewhere. Nothing is injected, so the pathway that triggers tremors is absent, and nothing has to be discovered, so a dry hole is not a possibility.

What it does not remove is the cost of drilling two wells and sewing them together underground, which is the entire economic question.

A closed loop turns geothermal into a construction problem rather than a prospecting one.

Somebody still has to drill it cheaper.

RE+
Hugo Rojas
Hugo Rojas

Hugo is an engineer with strong technical expertise. Multilingual from an early age, his writing combines technical clarity with a strong interest in science and energy.

Hugo_writer
Hugo Rojas

Hugo is an engineer with strong technical expertise. Multilingual from an early age, his writing combines technical clarity with a strong interest in science and energy.