Google Just Bet 396 MW on Geothermal to Power Its AI

Google has made a substantial wager of 396 megawatts on geothermal energy to power its artificial intelligence (AI) operations. The growing reliance on AI for various tasks has led to an increase in global electricity consumption by data centers. To meet this demand, the International Energy Agency predicts a potential rise in consumption to nearly 945 terawatt-hours.

While renewable energy sources like solar and wind remain vital, geothermal electricity has emerged as a promising alternative. This energy source harnesses heat from beneath the Earth's surface, providing a consistent and reliable power supply, independent of factors such as sunlight or wind. However, locating and accessing sufficient underground heat at an affordable cost has traditionally been a challenge.

Advanced drilling techniques and geothermal power generation methods are gradually overcoming these obstacles. The process involves extracting heat from deep underground and converting it into usable electric energy through a series of steps. Hot underground fluids are brought to the surface through wells, either as hot water or by circulating a sealed loop.

The heat then generates steam or vaporizes a secondary working fluid, which spins a turbine connected to a generator, producing electricity. The remaining steam or liquid is condensed and returned to nature for reuse.

Three conventional geothermal power generation technologies exist: dry steam, flash steam, and binary-cycle plants. Each technology varies in the temperature and characteristics of the geothermal resources they can utilize. Binary-cycle technology is particularly noteworthy as it expands the range of thermal resources that can be converted into electrical output, even from lower-temperature geothermal sources. However, lower-temperature heat generally results in reduced thermodynamic conversion efficiency.