Sam Altman, the chief executive of OpenAI, has pointed out that the water required to grow a single California almond can equal the amount used for 38,000 average ChatGPT queries.
The statement came during a recent interview on the Sources podcast. Altman addressed growing public concerns about the water footprint of large-scale AI infrastructure, including data centers that power models like ChatGPT.
He compared the intensive water needs of almond cultivation in California with the far smaller consumption per AI interaction. Altman said one almond requires roughly the same volume of water as 38,000 queries.
Altman estimated that an average ChatGPT query uses about 0.32 milliliters of water. The figure appears drawn from his recollection and could be subject to slight variation depending on the exact method of calculation.
Sam Altman was born in 1985 in Chicago, Illinois. He holds a degree in computer science from Stanford University and spent several years at Y Combinator before founding OpenAI in 2015, initially alongside Elon Musk. He stepped back from the board in 2019 to focus on building the organization and later returned as CEO. Altman has no prior criminal record or notable legal history related to environmental issues.
The episode took place on the Sources podcast. Altman responded directly to questions about AI data centers and their environmental impact.
Altman opened by noting that fears about data centers being major water guzzlers have been overstated. He cited a 2024 report from a Virginia state government committee, which examined data centers in an area with heavy construction activity.
The report found that while some individual facilities use more water than others, the majority consume amounts comparable to or less than those of large office buildings.
Altman said modern data centers have moved away from older evaporative cooling systems that rely heavily on water, and he pointed to the similar usage patterns in other large facilities.
He then turned to the specific comparison with California almonds, describing the production process and its water demands.
Altman said:
“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 the estimate came from memory and could be off, but the point served as a reminder of scale.
The remark quickly spread on social media, with users sharing clips and debating whether AI infrastructure truly has a smaller water impact than agriculture.
Critics noted that California almonds require significant irrigation in a region facing drought, while AI data centers draw power from the grid, sometimes with associated water use for cooling and electricity generation.
Supporters of Altman pointed out that efficiency improvements in data centers and the fact that one query produces a complete response make the comparison imperfect.
Altman’s comments align with his broader efforts to clarify that AI’s resource demands are often misunderstood, and he has referenced similar water estimates for queries in earlier public remarks.
The exchange did not include any new technical details about OpenAI’s infrastructure but served as a defense of the technology amid rising scrutiny over its environmental footprint.
No additional named individuals or official statements from regulatory bodies accompanied the podcast appearance.
The story remains based on Altman’s recollection and the widely shared clip from the Sources podcast.
