Artificial intelligence eats megawatts: Google's data centres consume record amounts of electricity
AI is cool. But there's a caveat: all these chats, image generators, and search engines are eating up electricity like there's no tomorrow. And Google has just confirmed this.
What we know
In its new Sustainability Report, the company frankly admitted that in 2024, its data centres consumed 30.8 million megawatt-hours of electricity. To understand the scale: this is twice as much as in 2020 (14.4 million MWh) and 7 times more than in 2014 (4 million MWh). The reason: the boom in artificial intelligence and cloud services.
Google's total electricity consumption grew by 27% in just one year, and in 2023 it was +17%. That is, the growth is like a startup on steroids.
At the same time, the company is trying not to become an environmental monster. According to them, thanks to the optimisation of data centres and the transition to clean energy, CO₂ emissions decreased by 12% over the year. The power usage efficiency (PUE) indicator has approached the theoretical limit of 1.09 (the ideal is 1.0, but it is practically unattainable).
How Google is trying to save the planet
Google boasts that since 2018, it has been fully covering its electricity consumption with renewable sources. In 2024, this helped to avoid 8.2 million tonnes of CO₂.
The company also signed 60 new contracts for 8 gigawatts of green energy. Among them are a geothermal power plant in Nevada, solar farms in South Carolina and Oklahoma, and - attention - future mini-nuclear reactors. That's right, Google is already thinking about the atom.
Video: Why Google is betting on advanced geothermal electricity
The plan for the future is even more interesting. The company is testing so-called "carbon-intelligent computing". The idea is to transfer tasks to regions or times of day when electricity is cleaner.
AI is beautiful. But now every chat request or image generation is another small drop in the global balance of electricity consumption. And while Google swears it's doing everything it can for green energy, the numbers show a simple truth: the AI era has proven to be far more energy-hungry than we ever imagined.
Source: Google
