Google's geothermal experiments are engineering templates for the energy transition

Google is pioneering the integration of engineered geothermal systems (EGS) into its next-generation data centers to address the growing thermal and power demands driven by AI-scale computing. As AI workloads increase, traditional cooling methods like air cooling are becoming insufficient, especially with emerging high-performance chips such as Nvidia’s GB200, which generate significantly higher thermal loads. Google's approach involves leveraging subsurface heat as a stable, low-carbon energy source that can be engineered for dispatchability and scaled to meet the real-time power and thermal needs of hyperscale compute infrastructure. This initiative aims not only to provide near-constant carbon-free energy (CFE) for Google’s operations but also to serve as a scalable blueprint for the broader energy transition. Google’s geothermal efforts include two major projects: an enhanced geothermal system in Nevada developed with startup Fervo Energy, which employs advanced techniques like horizontal drilling and fiber-optic monitoring; and a corporate geothermal power purchase agreement in Taiwan with Baseload Capital, designed to deliver 10 MW of reliable power