Interview

Zanskar Geothermal raises $115M to find hidden geothermal resources using AI-driven exploration

Apr 22, 2026 with Joel Edwards

Key Points

  • Zanskar Geothermal raises $115M to explore hidden geothermal reservoirs across the western US using AI models trained on layered geologic datasets.
  • The company owns and operates a geothermal power plant in New Mexico, executing a full-stack model of finding resources, building plants, and selling power under long-term contracts.
  • Geothermal exploration relies on AI pattern recognition across disparate data inputs rather than a single imaging technique like seismic in oil and gas, making geological judgment about model inputs the core competitive moat.

Zanskar Geothermal raises $115M to hunt blind geothermal systems in the US West

Zanskar Geothermal is a wildcatter for hidden geothermal resources, drilling for heat reservoirs that have no surface expression — no hot springs, no geysers — but sit as massive, energy-dense pockets of hot water deep underground. Joel Edwards, the company's geologist co-founder, describes the target as Yellowstone-type systems that are invisible from above, located across the western US, and accessible through well fields that pipe hot brine or steam into conventional turbines.

How the business works

Zanskar already owns and operates a geothermal power plant in New Mexico, selling electricity to an investor-owned utility that services Albuquerque and Santa Fe. The full-stack model is find the resource, build the plant, sell the power under long-term contracts. Edwards says the company has announced several sites and has more to come.

We wildcat for new geothermal resources out West to make power. The trick in geothermal is that there's not a single type of geologic data that tells you if a system is in existence other than just drilling a well. That's where AI tools really flex because they can handle the hyperdimensionality problem really well — better than humans.

Why exploration is the hard part

Unlike oil and gas, geothermal has no silver-bullet sensor. In hydrocarbons, seismic reflection imaging does most of the heavy lifting. Geothermal systems leave no single clean signal, so Zanskar layers multiple data sets and uses AI models to handle what Edwards calls the "hyperdimensionality problem" — pattern recognition across disparate geologic inputs that humans struggle to hold simultaneously. The company's core work is building what Edwards describes as the sandbox in which the models learn to find systems: choosing the right data, setting the right learning objectives. The AI is the tool; the geological judgment about what goes into the model is the moat.

Scale potential

Edwards puts tens of gigawatts of near-term potential as "super duper realistic" with current tools. Hundreds of gigawatts and beyond are being discussed in the industry, though he treats those numbers as longer-dated and technology-dependent. He draws a parallel to the Permian Basin: exploration finds more systems, while unconventional drilling techniques — horizontal wells, stimulation — extract more from each one. Both tracks are advancing simultaneously in geothermal.

Political tailwinds

Geothermal currently has bipartisan support in Washington. It is baseload power with no emissions, domestically sourced, and uses US drilling supply chains. Edwards notes it is one of the rare energy technologies that plays well on both sides of the aisle right now.

The raise

Zanskar closed $115M a few months ago, which includes a credit facility component. No valuation or lead investor was disclosed in the conversation.

Every deal, every interview. 5 minutes.

TBPN Digest delivers summaries of the latest fundraises, interviews and tech news from TBPN, every weekday.