Altman: 38,000 ChatGPT queries use the water of one California almond

OpenAI CEO Sam Altman is pushing back on the narrative that AI data centers are draining the water supply, and he brought a produce-aisle analogy to make the point. Speaking on the Sources podcast with Alex Heath, Altman argued the water consumption issue tied to AI has been blown out of proportion, and claimed that producing a single almond in California is equivalent to the water used for 38,000 ChatGPT queries.

What Altman actually said

His exact line: “For every 38,000 ChatGPT queries, that is the same amount of water that is used in the production of a single almond in California.” He added that “people that are scarfing down 12 almonds at a time don’t feel like they’re doing something horrible from a water perspective for the most part.”

Altman 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. He also argued that public perception of data center water consumption has developed into a robust meme that is difficult to correct.

Does the math check out?

In June, Altman had said that a single ChatGPT query uses 0.32 millilitres of water. 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. Either way, the per-query footprint is small — the debate is really about scale.

Do you want to see how to make more plays? Do you want to find gains yourself?

Unusual Whales helps you find market opportunities through our market tide, historical options flow, GEX, and much, much more.

Create a free account here to start conquering the market with Unusual Whales.

Why the water debate matters for markets

Water and power are the two hard constraints on the AI build-out. Training AI models in older facilities can use far more water, with estimates suggesting that GPT-3 used up almost 185,000 gallons of water with older systems, and the rising popularity of AI agents — which chew through far more tokens than chatbots — means data centers could still use more water than almond farming in aggregate.

That is the piece Altman’s almond comparison sidesteps. A per-query number looks trivial; a hyperscaler campus running 24/7 does not.

Options market and stocks to watch

Data center water and power exposure runs through much of the AI trade. Watch for:

MSFT — Microsoft’s Azure buildout and OpenAI partnership put it at the center of any AI-resource narrative shift.

NVDA — Nvidia demand is tied to how aggressively hyperscalers can permit and cool new sites.

GOOGL — Alphabet has faced its own scrutiny over data center water disclosures.

AMZN — AWS capacity plans hinge on siting decisions in water-stressed regions.

META — Meta’s AI capex ramp keeps it exposed to any regulatory response on data center resource use.

Watch for headline risk around state-level water regulation and permitting, plus how management teams frame efficiency on the next round of earnings calls. For more, see other news on the AI infrastructure trade.

Want more market intelligence? Create your free Unusual Whales account for options flow, market tide, GEX, and the full toolkit.