Google reported a 13 percent rise in greenhouse gas emissions for 2023 driven by the energy appetite of artificial intelligence and scarce availability of renewable energy in Asia and certain U.S. regions.
Google’s total emissions reached 14,314,800 metric tons, a 48 percent increase over 2019 levels, which is the baseline for its emissions reduction targets, according to the company’s 2024 Environmental Report published July 2:
- Scope 1 emissions, generated directly from Google’s own operations, decreased 13 percent.
- Scope 2, related to energy purchases, climbed 37 percent.
- Scope 3, including purchased goods such as data center construction materials and representing 75 percent of the total, grew 8 percent.
Google’s net-zero goal calls for a 50 percent reduction across Scope 1, 2 and 3 by 2030. Its disclosure of a sizable increase mirrors Microsoft’s revelation in mid-May, when that company posted a 29.1 percent increase in emissions since 2020, the year it declared a “carbon negative” goal.
Both Google and Microsoft said voracious data center expansion has made meeting corporate climate goals challenging.
An ‘inflection point’ for AI
The amount of energy consumed by Google’s data centers in 2023 was more than 24 terawatt-hours — approximately 7-10 percent of global electricity consumed by data centers, the company reported. That was a 17 percent increase from 2022 and roughly the same as the 2021 to 2022 increase, said Google CSO Kate Brandt. AI is largely responsible for the rise. As of 2022, data centers accounted for approximately 1-1.3 percent of worldwide electricity demand.
“Because AI is becoming so deeply integrated into our products, the distinction between what is an AI workload and what is not an AI workload is kind of not meaningful,” she said.
Google is betting big on AI for its cloud services business and as the foundation for future search revenue. It’s also a big focus for the sustainability team, which has set a goal of using AI to help cities, individuals and companies reduce their carbon footprints by 1 gigaton by 2030. The company cites research estimating that AI could help mitigate 5-10 percent of global emissions by the end of the decade.
With that in mind, Google is prioritizing three strategies to manage AI’s environmental impact:
- Model optimization, via software that decreases the resources and time required to “train” AI to make decisions.
- Efficient infrastructure, including next-generation computing hardware like Nvidia’s Blackwell technology, which the company said will train AI models using 75 percent less power than previous versions. Water is also a big focus: Google’s data centers consumed 6.1 billion gallons in 2023, 17 percent more than in 2022.
- Emissions reductions, through the company’s investments in renewable energy: It has signed contracts for 25 terawatt-hours. Some of that capacity is delivered directly, some comes in the form of credits. The next phase of investments will center on adding clean power in Asia and accelerating the availability of technologies such as enhanced geothermal through new corporate partnerships.
Reality check for ‘carbon-free energy’ strategy
Google signed its first corporate power purchase agreement for renewable energy in 2010, pioneering a strategy that has become widespread among large companies with ambitious emissions reduction targets.
“We set a 100 percent renewable energy goal in 2012, without a full roadmap for how to get there,” said Brandt, who celebrates her ninth anniversary with Google at the end of July. “It pushed us to innovate, it pushed us to partner, and it pushed us to work h
Read More
