How Google is Lowering Data Centre Emissions Amid AI Growth

The AI revolution is forcing technology companies to confront a difficult truth: the same models that promise to solve climate problems are themselves driving demand for electricity. 

Data centres worldwide are consuming ever-larger amounts of power to train and run increasingly sophisticated algorithms, raising questions about whether the industry can maintain its environmental commitments while chasing the AI frontier.

Now, Google has managed to reduce emissions from its data centres by 12% during 2024 despite electricity consumption jumping 27%, according to its 10th annual environmental report.

The company achieved this through what it describes as a combination of clean energy procurement, hardware innovation and infrastructure efficiency improvements.

“We’re proud to release our 10th annual Environmental Report, which details how we’re working to address the increased energy demands of AI to enable this positive impact, while also showcasing how AI can be used to build a more energy-efficient and resilient world,” it says.

Google’s record year for clean energy contracts

Google signed contracts for over 8 gigawatts of additional clean energy generation in 2024 – the largest annual total in its history and double the volume contracted the previous year. 

The giant has now signed more than 170 agreements since 2010 to purchase over 22 gigawatts of clean energy generation, roughly equivalent to Portugal’s total renewable energy capacity in 2024.

Google brought 2.5 gigawatts of new clean energy online during 2024 from projects contracted over recent years, some dating back to 2019. 

It also maintained what it calls a 100% renewable energy match on a global basis for the eighth consecutive year. 

More significantly, its carbon-free energy percentage across data centres and offices increased from 64% to 66% on an hourly matching basis, meaning facilities matched their electricity use with clean power generation on the same grid during the same hour.

Nine out of 20 grid regions with Google-owned and operated data centres achieved at least 80% carbon-free energy. 

The company estimates that once operational, the new clean energy agreements signed in 2024 could generate nearly four times more electricity than its incremental load growth from 2023 to 2024.

Google’s clean energy purchases additionally avoided more than 8.2 million metric tons of carbon dioxide equivalent emissions in 2024 alone. 

From 2011 to 2024, the company estimates its clean energy purchasing cumulatively avoided more than 44 million metric tons, roughly equivalent to the emissions from the combined annual electricity use of every home in New York State.

“Transparency, accuracy and rigor are the foundation of sustainability reporting,” says Luke Elder, Lead Sustainability Reporter at Google.

“As the volume and complexity of data and strategies grow, we’re innovating our processes to meet rising expectations without compromising on these core commitments.

“This year, we integrated generative AI into our workflow to support the development of Google’s 2025 Environmental Report. 

“We’ve learned a tremendous amount from this first implementation and are already building more ways to leverage AI for future reports.”

Google’s nuclear and geothermal breakthroughs to power AI

The company has moved beyond traditional renewable sources by signing the world’s first corporate agreement to purchase nuclear energy from small modular reactors, partnering with Kairos Power. 

The deal will bring up to 500 megawatts of clean energy to US grids by 2035, with the first reactor expected to become operational by 2030 and additional deployments through 2035.

Small modular reactors represent a shift from conventional nuclear plants, offering smaller-scale, factory-built units that can be assembled on site. 

Kairos uses a molten salt cooling system and ceramic pebble fuel, allowing the reactor to operate at low pressure and simplifying the design.

Google also expanded its partnership with Fervo Energy for enhanced geothermal projects in Nevada, contracting for a 115 megawatt project through a new utility rate structure. 

This expansion will increase the amount of enhanced geothermal generation enabled by Google by almost 25 times compared to its initial pilot project, which became operational in 2023.

Combining data centres, energy and AI innovation for sustainability 

Google’s data centres now deliver over six times more computing power per unit of electricity than they did five years ago. 

The average annual power usage effectiveness for the company’s global fleet dropped to 1.09 in 2024, below 1.10 for the first time in six years. 

This metric compares non-computing overhead energy to energy used for information technology equipment, with a figure closer to 1.0 indicating nearly all energy goes to computing.

The company’s seventh-generation Tensor Processing Unit, called Ironwood, operates nearly 30 times more efficiently than the first Cloud TPU from 2018. 

These custom-built processors are designed specifically for AI and ML applications.

Google’s sixth-generation TPU, Trillium, delivers 67 % more energy efficiency than the previous generation and 14 times more compute per watt compared to first-generation Cloud TPUs.

How Google’s AI products enable emission reductions

Five of Google’s products enabled others to collectively reduce an estimated 26 million metric tons of greenhouse gas emissions in 2024, according to the report. 

This figure is only a subset of the company’s efforts but exceeds Google’s total emissions of 11.5 million metric tons during the year.

Nest thermostats helped customers save more than 25 billion kilowatt-hours of energy in 2024, enabling approximately 8.4 million metric tons of emissions reductions. 

Fuel-efficient routing in Google Maps enabled over 2.7 million metric tons of emissions reductions, equivalent to taking approximately 630,000 petrol-powered cars off the road for a year.

“AI isn’t just a tool – it’s a catalyst,” Google says.

“It’s helping people make smarter decisions faster, and its potential to help manage emissions in key sectors – like transportation and energy – is transformational.”