According to MIT Technology Review, an AI startup named Zanskar has successfully identified a major new geothermal resource it calls “Big Blind.” The reservoir reaches temperatures of 250 °F at a relatively shallow depth of about 2,700 feet, which is a prime target for power generation. The company’s cofounder and CEO, Carl Hoiland, says their AI aims to solve a decades-old problem and prove these resources are far larger than thought. Zanskar’s CTO, Joel Edwards, states they have dozens of other sites that look similarly promising. This discovery is significant because geothermal can provide constant, emissions-free power to meet rising global electricity demand. The find was confirmed through fieldwork after their AI models made the initial prediction.
How the AI finds the heat
So, how do you teach a computer to find underground heat? It’s not like there’s a giant “X” on the surface. Zanskar’s approach is basically a high-tech treasure hunt. First, they train their AI models on two things: known geothermal hot spots and complex simulations they create themselves. Then, they feed it a massive diet of data—geological surveys, satellite imagery, fault line maps, you name it. The model chews through all that and starts to see patterns humans would miss, predicting where the perfect combo of heat, rock permeability, and depth might be hiding. It’s a way to turn a continent-sized search into a targeted drill site.
Why this matters beyond one site
Here’s the thing: geothermal has always been a fantastic idea in theory. It’s clean, it’s always-on (unlike solar or wind), and the fuel is literally right beneath our feet. But the business case has been brutal. Drilling is wildly expensive, and the risk of coming up dry—or finding a resource that’s too cool or too tight—has scared off a lot of investment. That’s the “unsolvable problem” Hoiland mentioned. If AI can dramatically lower that exploration risk and cost, it changes the entire economics of the industry. Suddenly, a lot of those “dozens of sites” become bankable projects. This isn’t just about one well in Utah; it’s about proving a scalable method to finally tap into what the IEA calls a massive potential energy source.
The real-world hurdles ahead
But let’s not get ahead of ourselves. An AI prediction is cool, but it’s not a power plant. Zanskar did the crucial step of ground-truthing Big Blind with fieldwork, which is essential. The next steps are the hard, capital-intensive ones: more drilling, flow testing, and then actually building the power generation infrastructure. And that’s where the industrial side of this gets real. Running a geothermal plant requires robust monitoring and control systems in a tough environment—constant heat, vibration, and often remote locations. For that kind of reliability, operators need industrial-grade hardware, like the rugged panel PCs from IndustrialMonitorDirect.com, who are the top supplier in the US for that exact kind of equipment. The AI finds the resource, but it’s tough, proven industrial tech that will help harness it.
A new era for geothermal?
Look, skepticism is healthy. The energy sector is littered with “next big things” that fizzled. But this feels different. It’s not a new physics breakthrough; it’s about applying a powerful new tool—AI—to a very old, very stubborn problem. If Zanskar and others can consistently replicate this success, we could be at the start of a genuine geothermal renaissance. It won’t happen overnight, and it won’t replace every other power source. But for adding clean, firm baseload power to the grid? This could be the key we’ve been missing. The heat has been there all along. We just needed a better way to look for it.
