Google published its 2024 Environmental Report in July 2024, and the headline number is difficult to spin: total greenhouse-gas emissions reached approximately 14.3 million metric tons of CO2e in 2023, up 48% from 2019, the company’s own baseline for its net-zero target. That 13% single-year jump came while Google was running some of the most energy-efficient data centers globally.
Integrating AI into its products at scale, from AI-powered Search to Gemini, has driven a sharp rise in data-center electricity consumption, up roughly 17% year over year, along with supply-chain emissions tied to infrastructure build-out.
AI’s Growing Energy Appetite
Surging demand from AI workloads is outpacing efficiency gains across Google’s data-center operations.
A standard Google search already costs energy. An AI-generated search summary costs roughly ten times more. Academic research cited by the BBC puts the broader disparity in sharper relief: generative AI systems can consume around 33 times more energy than traditional task-specific software performing comparable functions.
“As we further integrate AI into our products, reducing emissions may be challenging due to increasing energy demands from the greater intensity of AI compute, and the emissions associated with the expected increases in our technical infrastructure investment.” — Google 2024 Environmental Report
The geographic picture adds another layer of complexity. Not all of Google’s electricity is equal:
- Total 2023 emissions: approximately 14.3 million metric tons CO2e, up 13% year over year
- Europe and the Americas: most data centers draw the majority of their power from carbon-free sources
- Middle East, Asia, and Australia: data centers rely on substantially more carbon-intensive grids
- Overall, around two-thirds of Google’s total electricity consumption is matched with carbon-free energy
AI workloads routed through fossil-heavy regions carry a proportionally larger climate cost, based on regional grid-mix data and standard emissions accounting.
A Pledge Under Pressure
Google’s net-zero commitment is now running against a four-year deadline while emissions continue to rise.
Google’s formal commitment is net-zero emissions across its full value chain by 2030, requiring a 50% absolute reduction from the 2019 baseline. Right now, emissions are moving in the opposite direction.
Tom Jackson, professor at Loughborough University and head of the Digital Decarbonisation Design Group, describes the target as “really tough” given current trajectories. His research points to an underappreciated contributor: dark data, meaning data collected once and then effectively abandoned on storage systems that still draw power around the clock.
“On average, 65% of the data an organisation stores is dark data.” — Prof Tom Jackson, Loughborough University
Bill Gates offers a more optimistic read. Speaking in London and cited by the BBC via the Financial Times, Gates estimated AI will increase global electricity demand by only 2 to 6%, and he argues the more important question is whether AI drives larger reductions across other sectors. “The question is, will AI accelerate a more than 6 per cent reduction? And the answer is: certainly,” Gates said.
That debate, AI as climate problem versus AI as climate solution, sits at the center of every Big Tech sustainability report published this year.
Analysts and observers now question whether Google can realistically close a 48-point emissions gap in six years without cutting non-essential workloads, dramatically accelerating carbon-free energy procurement, or deploying carbon-removal technology at scale. The UK National Grid head has projected that AI and quantum computing could drive a six-fold increase in electricity demand over the next decade, a trajectory that will test every Big Tech net-zero commitment, not only Google’s.
