Sam Altman is pushing back against what he calls exaggerated claims about artificial intelligence's water consumption, offering an unusual comparison to put the concern in perspective. Speaking on the Sources podcast, the OpenAI CEO said that producing a single California-grown almond requires roughly the same amount of water as 38,000 ChatGPT queries, suggesting that AI's environmental impact has been wildly overstated.

Why Is AI's Water Usage Suddenly a Major Concern?

As artificial intelligence systems become more powerful and widespread, data centers that power them require enormous amounts of electricity and cooling water. These facilities generate significant heat from powerful computers, and cooling systems need water to keep machines operating safely. With demand for AI growing rapidly, companies are building new data centers at an accelerating pace, raising concerns about their impact on local water supplies and the environment.

A May Gallup survey found that 71% of Americans opposed having a data center built near their home, with environmental impact and water use cited as key reasons for concern. The debate extends beyond individual queries; it encompasses the broader resource demands of the rapidly expanding AI industry and where that water comes from.

What Does Altman's Almond Comparison Actually Reveal?

Altman's comparison is meant to make individual AI query water use seem trivial by comparison to everyday consumption. He noted that people casually eating almonds do not typically worry about the water footprint of their snack, so why should they worry about ChatGPT queries.

However, the math tells a more nuanced story. Research estimates that producing a California almond requires approximately 3.56 liters of water, or 3,560 milliliters. In June, Altman himself stated that an average ChatGPT query uses around 0.32 milliliters of water. Using these two figures, one almond's water footprint would be equivalent to roughly 11,000 ChatGPT queries, not 38,000.

"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," said Sam Altman, CEO of OpenAI.

Sam Altman, CEO of OpenAI

Altman later acknowledged that he was recalling the 38,000 figure from memory and that it was only approximate. The discrepancy highlights a critical point: the exact assumptions behind water use estimates matter significantly, and small differences in methodology can produce vastly different conclusions.

How Do Modern Data Centers Actually Use Water?

Altman argued that modern data centers operate very differently from older facilities that relied heavily on evaporative cooling systems. He claimed that a very large modern data center can use an amount of water comparable to an office building for routine activities such as sinks and toilets, though water consumption varies considerably between facilities and cooling systems.

• Cooling Technology: The type of cooling system used significantly impacts water consumption, with some facilities using more water-intensive methods than others.
• Local Climate: Geographic location affects how much cooling is needed, with warmer regions requiring more water for temperature control.
• Operational Efficiency: How well a facility is managed and maintained influences overall water use, with some data centers operating more efficiently than others.

A 2024 report from a Virginia state government commission found that water consumption differed significantly between facilities, although many data centers used amounts comparable to or lower than those of large office buildings. Meanwhile, technology companies are working to reduce water consumption. Amazon reported that water used by its data centers fell by 2% in 2025 compared with the previous year, although the company still used more than 9 billion liters during the year.

What's Altman's Broader Argument About AI Hype?

Altman criticized what he described as exaggerated claims about AI and water consumption, saying the perception around data center water use had become a "robust meme" that was difficult to correct. He pointed to online claims suggesting that running a single ChatGPT query uses as much water as running a shower for six hours, calling such assertions unfounded.

"I saw this thing going around about the water usage of ChatGPT. And it was like every time you run a single ChatGPT query, it's like, you know, you run your shower for like six hours, and the water never comes back. I don't think it holds up to any scrutiny," said Altman.

Sam Altman, CEO of OpenAI

This criticism fits into a broader pattern of Altman's public messaging. At the G20 Innovation Ministerial in Chapel Hill, North Carolina, Altman argued for a balanced approach to AI regulation and risk assessment, cautioning against both excessive doomerism about AI risks and blind optimism that ignores legitimate concerns.

What's the Bigger Picture Beyond Individual Queries?

While Altman's comparison highlights that individual ChatGPT queries use minimal water, it does not address the full scope of the issue. An individual query may use a tiny amount of water, but billions of queries combined with the construction and operation of large data centers can add up to significant resource consumption. The real question is not simply "How much water does one ChatGPT query use?" but rather "How much water does the rapidly growing AI industry require, and where does that water come from?".

Researchers have warned that water-intensive cooling systems can put additional pressure on water supplies in already dry regions, a concern that extends beyond the per-query calculation. As AI demand continues to accelerate and more data centers are built worldwide, understanding the cumulative environmental impact becomes increasingly important for policymakers and communities.

Altman's comments come as OpenAI prepares to release Astra, the company's first AI model designated as having critical cyber capability, which underwent a voluntary review by the Trump administration before its expected public release. The company is also developing increasingly powerful models, signaling the rapid pace of AI development and the growing need for thoughtful discussion about both the benefits and resource demands of advanced AI systems.