A New Mexico geothermal plant on the verge of retirement just ran a year its owner calls one of the U.S.'s most productive

A small geothermal power plant in southwestern New Mexico, on track to be retired two years ago, has just completed a full year of operations that its new owner says makes it one of the most productive pumped geothermal sites in the United States. The comeback at Lightning Dock runs on a deliberately unglamorous recipe: a new well, modern drilling tools, AI-assisted targeting, and a year of patient operating data. The result, if Zanskar's numbers hold up and the recipe travels, would matter for the thousands of aging U.S. geothermal wells sitting on reserves the industry has already paid to find. (WIRED)

Geothermal provides less than 1 percent of U.S. electricity, a slice limited less by physics than by geography: the most productive reservoirs sit in the West, and the cost of drilling and proving a new well has historically kept development slow. Most wells fade with age as their reservoir cools and pressure drops, which is why reviving an old well is rarely as simple as turning the pumps back on. "Pumped" geothermal, the kind Lightning Dock runs, cycles water down into hot rock and back up to drive a turbine; the bottleneck is keeping enough hot water flowing. Lightning Dock, originally drilled in the 1980s and acquired by Salt Lake City-based Zanskar in 2024, was fading faster than most: the company says its heat loss ran between five and 10 times the average for U.S. wells. (WIRED, Canary Media)

The fix was not a single miracle well. Zanskar's engineers drilled a new production well at the site, leaned on AI-assisted targeting to pick the trajectory, then ran the plant continuously through a full year to build a dataset on reservoir behavior under real operating conditions. "AI-assisted targeting," in this context, means using reservoir models and the site's own well history to pick where the new well should land rather than relying on a geologist's read of the subsurface. The company attributes the result to a combination of new drilling technology, machine-assisted targeting, and, in the words of co-founder Joel Edwards, "a little bit of luck." (Zanskar Substack, ThinkGeoEnergy)

The productivity claim is Zanskar's, not an independent measurement. WIRED obtained the year of operating data exclusively from the company, and Zanskar has not published third-party validation. The site may well be unusually productive, but the comparison to other U.S. wells is a Zanskar-relative judgment. (WIRED)

The broader interest is in the recipe. There are thousands of aging U.S. geothermal wells sitting on reserves the industry has already paid to find. If a similar playbook works on even a fraction of them, the operating map of U.S. geothermal could be wider than the reserves map suggests. Trade-press confirmation of the repowering arrived this summer. (ThinkGeoEnergy)

The honest caveat is in Zanskar's own telling. Shallow wells, the right reservoir geology, a startup willing to spend on a new well, and a team that already knew the site: that stack may not transfer to every aging asset. Wells with deeper, hotter, or more depleted reservoirs may need a different recipe or none at all. (WIRED, Canary Media)

Zanskar's broader pitch is that geothermal is in an "investment renaissance," and Lightning Dock is the proof point: legacy sites deserve a second look before anyone drills a new field. Whether the recipe generalizes is the open question. The next test will be whether Zanskar, or any peer, can repeat the result at a second site with different geology, and whether the year of operating data survives independent review. (Zanskar Substack)