AI needs more energy than ever
Google has secured a massive 150-megawatt geothermal energy agreement to power its data centers, sending a clear signal that artificial intelligence energy demands are transforming how major tech companies think about electricity.
The deal positions Google at the forefront of a trend that is redefining the industry: as the AI race accelerates, clean and abundant energy has become a competitive advantage as important as chips and algorithms.
Why geothermal
Unlike solar or wind energy, which depend on weather and time of day, geothermal energy provides constant electricity 24 hours a day. This characteristic makes it ideal for data centers that need to operate without interruption to train and run AI models.
AI data centers consume significantly more energy than traditional data centers. Training a large language model can consume as much electricity as a small city for weeks. With the proliferation of services like Gemini, Google needs to secure energy sources that can scale without relying on fossil fuels.
150 megawatts in perspective
To put it in context, 150 megawatts can power approximately 100,000 homes. It is enough energy to operate a large-scale data center dedicated to training AI models. The agreement represents one of the largest geothermal energy purchases by a technology company to date.
The big tech energy race
Google is not alone in this race. Microsoft has signed deals to reactivate nuclear reactors, Amazon has invested in solar energy at massive scale, and Meta is exploring nuclear energy options for its data centers. All face the same problem: AI requires unprecedented amounts of electricity.
Analysts estimate that global data center energy demand could double by 2028, driven primarily by artificial intelligence. Companies that secure reliable and clean energy sources will have a significant advantage over their competitors in the race to deploy AI at scale.
Environmental and economic impact
The bet on geothermal energy allows Google to advance toward its climate commitments while meeting the growing demand for computation. However, critics point out that no amount of clean energy compensates for total consumption if it continues growing exponentially.
For users, this means that the AI services we use daily, from intelligent searches to virtual assistants, carry a real and growing energy cost. The sustainability of the AI revolution will depend on the industry ability to power it without destroying the planet in the process.
