NEWS
Hidden
Geothermal Goldmines
Discovery of the first blind geothermal system in more than three decades is setting the stage for geothermal’s future in the U.S.
Written by Nicole Imeson
FOR THREE DECADES, the geothermal industry overlooked “blind” sites, assuming the easy energy had vanished. But beneath a dusty patch of Nevada desert, sat a hydrothermal reservoir with the potential to power 100,000 homes. Zanskar, a Western U.S. startup, shattered the myth of energy scarcity by discovering Big Blind, the first major blind U.S. geothermal system identified for commercial use in 30 years.
Zanskar bypassed traditional exploration by developing a machine learning model to synthesize dozens of seemingly unrelated data sets. Their algorithm generated high-probability predictions for both reservoir location and depth, creating an iterative mapping process that improved with every data point.
“We scraped together regional geologic data sets from satellite, airborne, and ground-based survey campaigns,” explained Joel Edwards, co-founder and chief technology officer at Zanskar. “We fed those into a model space with confirmed locations of geothermal sites. The model trained itself to infer relationships across the different types of geologic data.”
The Big Blind geothermal system in the western Nevada desert is the first blind geothermal system identified in the U.S. in more than 30 years. Video: Zanskar
Big Blind, Pumpernickel, and Lightning Dock
Zanskar announced the discovery of Big Blind in December 2025 after validating high-probability AI targets through physical drilling. Located just 823 meters (2,700 feet) deep with temperatures reaching 121 °C (250 °F), sites with similar properties like Big Blind can generate more than 100 megawatts (MW). To confirm these reservoirs, teams drilled test wells to measure thermal spikes against the Earth’s natural background heat. By comparing incremental depth temperatures to a baseline below the reach of seasonal weather, they successfully identified hot water surging through hidden faults. Just months earlier, in September 2025, Zanskar used the same process to validate Pumpernickel, another greenfield site sitting at 762 meters (2,500 feet) with 137 °C (279 °F) water.
These discoveries relied on a Bayesian evidential learning (BEL) model. “The trick lies in how you manage the spatial aspects and biases of the datasets,” explained Carl Hoiland, co-founder and CEO at Zanskar. “We built out different data layers, none of which acts as a silver bullet, but altogether, they reveal a specific signature.”
This predictive success grew from operational roots planted in May 2024, when Zanskar acquired Lightning Dock, a struggling geothermal plant in New Mexico. Farmers discovered the resource in the 1940s, but the 2013 power plant underperformed because the previous owners tapped only a shallow zone. Zanskar’s team identified a deeper reservoir by applying a modern physics simulator to analyze the field’s physical possibilities.
“When we entered the field, we treated the resource probabilistically instead of deterministically. This gave us a full suite of possibilities rather than relying on just one or two expert opinions,” Hoiland explained. This dual focus on greenfield exploration and active operations allowed Zanskar to de-risk projects by tailoring drilling strategies to the specific mechanical needs of a long-term power plant.
Geothermal Power Generation
Both Big Blind and Pumpernickel power generation will use binary cycle technology to convert moderate temperature resources into electricity. Unlike traditional dry steam or flash plants that pipe geothermal steam directly into turbines, these facilities operate as a closed loop.
Geothermal water flows through a heat exchanger, transferring thermal energy to a working fluid such as isobutane, which possesses a lower boiling point than water. The vaporized isobutane spins the turbine before cooling and recirculating. This separation shields mechanical components from corrosive minerals and salts found in underground brine, while reinjecting the cooled water back into the reservoir preserves the resource for decades.
The 24/7 baseload energy supply offers a holy grail for data centers requiring constant uptime. However, modern AI workloads introduce massive, erratic power spikes that challenge traditional steady-state power grids. To meet this demand, geothermal sites include load following designs, allowing the plant to ramp electricity production by 15–30 percent per minute.
By modulating the flow of the working fluid through the heat exchanger, engineers quickly adjust the mechanical spin of the turbine to chase AI power demands in real time. When local demand drops, the plant diverts excess electricity to the utility grid, maximizing revenue without throttling heat production.
A Zanskar drilling rig in the Nevada desert. Photo: Zanskar
Zanskar is powering the 15 MW Lightning Dock geothermal plant in New Mexico with a single well. Photo: Zanskar
Conventional vs. Unconventional
Historically, geothermal exploration prioritized conventional reservoirs, naturally occurring systems of heat, water, and permeable rock. However, high failure rates during early search efforts soured investors, shifting the industry’s focus toward unconventional enhanced geothermal systems (EGS). In EGS, engineers drill deep into solid, dry rock and inject high-pressure water to fracture the stone, a process known as fracking, to create an artificial network for heat exchange.
While the industry turned to brute-force engineering, Zanskar pivoted back to conventional resources by targeting “blind” systems. Geothermal experts now believe the vast majority of naturally occurring hydrothermal reservoirs show no visible evidence, like hot springs or geysers, at the surface.
“It’s a common error to underestimate the size of underground resources that we have not yet discovered. There is a perception in the market that conventional geothermal is tapped out, but we’ve barely scratched the surface. We think these systems can be a lot bigger than expected,” Edwards explained.
Zanskar’s AI driven approach complements the industry’s shift toward EGS by offering a parallel path to rapid scaling. While EGS provides greater geographic flexibility by creating reservoirs closer to existing infrastructure, Zanskar’s model identifies hidden, natural pathways that offer higher immediate efficiency. By uncovering these shallower, high-energy systems, they avoid the technical complexities of fracking and the high costs of extreme-depth drilling, while adding to the diverse portfolio of clean energy solutions for the grid.
“While we focus on discovering new systems and de-risking existing systems, we’re increasingly optimistic that these same technologies will help us extract more sustainably over many decades. The only way to test and prove out those ideas is to be an operator of those assets,” Hoiland explained.
By shifting from a strategy of “drilling where it’s obvious” to using AI for “geological prospecting,” Zanskar uncovered a subterranean goldmine that sat hidden for decades. The realization that we have barely scratched the surface of conventional geothermal suggests that we aren’t facing a scarcity of clean energy but stand at the threshold of a new age of global energy abundance.
Nicole Imeson is an engineer and writer in Calgary, Alta.
© 2026 The American Society of Mechanical Engineers. All rights reserved.