Sam Altman Defends ChatGPT's Water Use, Says It Takes 38,000 Queries To Match One Almond

As scrutiny grows over the environmental cost of the data centres powering AI, OpenAI CEO Sam Altman has pushed back against what he calls exaggerated claims about how much water ChatGPT actually consumes.

Sam Altman's almond comparison

Speaking on the Sources podcast, Altman said that roughly 38,000 ChatGPT queries use about as much water as it takes to grow a single California almond. He was upfront that he was citing the figure from memory and that it could be off, though he maintained it was in the right ballpark.

Altman also took aim at viral comparisons that liken a single ChatGPT query to hours of showering, arguing that such claims fall apart under closer examination.

The numbers behind the claim

Altman has previously put the water cost of a single ChatGPT query at around 0.32 millilitres. Separately, a 2019 study involving researchers linked to the US Geological Survey estimated that growing one California almond takes an average of 3.56 litres of water.

Doing the math on those two figures actually puts the ratio closer to 11,000 ChatGPT queries per almond, well below the 38,000 Altman cited, underscoring the uncertainty baked into these back-of-the-envelope comparisons.

Modern data centres, Altman argues, aren't as thirsty

Altman also pushed back on the assumption that today's data centres rely on the heavily water-intensive cooling methods of the past. He claimed a very large modern facility's routine water use, covering things like sinks and toilets, can be comparable to that of an ordinary office building.

Why the concern hasn't gone away

Even so, water use at data centres remains a live issue, especially in regions already dealing with water stress. Evaporative cooling systems, commonly used to keep servers from overheating, can draw enormous volumes of water, and researchers have flagged the strain this places on water-scarce areas.

Published on: Thursday, September 03, 2026, 02:44 PM IST