Zanskar thinks 1 TW of geothermal power is being overlooked

Beneath our feet lies an untapped reservoir of energy, prompting experts at the Department of Energy to estimate that geothermal power could provide up to 60 gigawatts — nearly 10% of the United States' electricity needs — by the year 2050. However, Carl Hoiland, co-founder and CEO of Zanskar, contends that this figure significantly underrepresents the true potential of conventional geothermal energy. The Department of Energy's projections primarily hinge on advancements in enhanced geothermal systems, which utilize fracking techniques to access heat from deep underground. While companies like Fervo and Sage Geosystems are exploring this promising avenue, conventional geothermal energy, which harnesses naturally fractured hotspots, has seen stagnation, contributing only 4 gigawatts to the U.S. grid, with minimal growth over the past decade. Hoiland argues that outdated assumptions have hindered the progress of conventional geothermal. "They underestimated the number of undiscovered systems, perhaps by a factor of ten or more," he stated. With the advent of modern drilling techniques, he suggests the potential output could increase significantly. "What if the numbers soared from tens of gigawatts to a terawatt-scale opportunity?" To capitalize on this potential, Zanskar is leveraging artificial intelligence to rejuvenate the stagnant sector. The startup has successfully revitalized a struggling power plant in New Mexico and identified two new geothermal sites with a combined capacity of over 100 megawatts. This innovative approach has led to a substantial funding boost, with Zanskar securing $115 million in a Series C round led by Spring Lane Capital, alongside contributions from a variety of investors. Many geothermal sites have been overlooked due to reliance on surface indicators like hot springs or volcanic activity. According to Hoiland, a staggering 95% of geothermal systems do not exhibit such surface signs. "We keep discovering them by chance, which highlights AI’s potential in this field," he explained. Zanskar employs supervised machine learning to analyze diverse data sets, including past accidental geothermal discoveries. Once a promising site is identified, a validation team is dispatched to confirm the find. The company also utilizes Bayesian evidential learning (BEL) techniques to devise development plans based on existing data and probabilities derived from their models. The results so far have been encouraging. Following their recent funding, Zanskar explored three sites and achieved success in all instances, raising the question of what might happen if they expanded their efforts to ten sites. Hoiland believes they have enough prospects lined up to support at least a gigawatt of generating capacity, focusing initially on the U.S. West, which holds significant potential. The goal is to confirm at least ten sites to attract project finance investors, who offer more affordable capital than traditional venture capitalists. While Hoiland acknowledges that challenges remain in locating geothermal resources, he remains hopeful about Zanskar's trajectory. "We are now aware that this is the future of exploration," he remarked. "This could transform the geothermal landscape in a remarkably short time."