Big Tech's AI Data Centres Now Emit as Much Carbon as a Third of France

Microsoft, Amazon and Google collectively pumped out 119 million metric tonnes of carbon dioxide last year. The main culprit: building more data centres to feed AI demand.

AI2Day NewsdeskAI-assistedPublished Updated Editor: Lee Brown2 min read
Aerial view from directly above a vast data centre campus at dusk, rows of grey rectangular server buildings with white cooling units on rooftops, surrounded by
Illustration made with AI. Not a photograph of the events described.
Share

Key points

  • Microsoft, Amazon and Google combined emitted 119 million metric tonnes of carbon dioxide equivalent in the financial year ending March 2026.
  • That total equals roughly one third of France's entire annual national carbon output.
  • The three companies' collective emissions rose by nearly one fifth in a single year.
  • All three companies still say they intend to reach net zero carbon output, meaning they plan to remove as much carbon as they produce.

Three of the biggest names in technology are producing far more planet-warming carbon than they did a year ago. The reason is almost entirely AI.

Microsoft, Amazon and Google produced a combined 119 million metric tonnes of carbon dioxide equivalent in the financial year ending March 2026. France emits roughly three times that amount. These three companies alone account for one third of France's total, up nearly one fifth on the prior year. That's a steep jump, driven by one thing: the furious pace of data centre construction.

A data centre is a large building packed with computer servers and specialised chips that run AI systems. Training and running AI models demands enormous electricity. The more AI capacity a company builds, the more power it needs, and the more carbon its operations produce.

All three companies told reporters, first covered by The Guardian AI, that they remain committed to net zero goals. Net zero means removing from the atmosphere as much carbon as you put in, through renewable power, energy efficiency and carbon offset programmes.

The gap between that ambition and the current numbers isn't narrowing. It's widening.

We've tracked this tension since our first report on a Scottish AI megasite that overstated its green-energy plans, published 13 July 2026. The pattern there and here is the same: big construction commitments, climate pledges that can't yet keep pace.

For ordinary people, this connects to something concrete. Every AI chatbot query, every AI-generated image, every automated customer service call runs on hardware inside one of these buildings. The tools millions use daily carry a physical carbon cost, even when they feel invisible and free.

Will big tech actually hit its net zero targets?

Right now, genuinely unclear. All three companies have set net zero or carbon-neutral targets for future dates, but each has seen emissions rise sharply as AI investment accelerates. Hitting those targets would require either a dramatic shift to renewables for new data centres or significant gains in chip efficiency. Both, probably. Independent analysts are watching the gap between the pledges and the annual reports closely, and so are we.

© 2026 AI2Day