OpenAI CEO Sam Altman says 38,000 ChatGPT queries use as much water as the production of one almond — says data centers use no more water than an office building

OpenAI's CEO, Sam Altman, has dismissed concerns over the water usage of AI data centers, stating that the issue has been exaggerated and is no longer relevant given advancements in cooling technology. During an appearance on the Sources Podcast, Altman explained that producing a single almond in California uses as much water as 38,000 ChatGPT queries, though he acknowledged that the exact number might be inaccurate.

He noted that this comparison was based on full accounting of water usage, not just what occurs within individual data centers.

Altman pointed out that modern large data centers now employ cooling methods that consume water at a rate comparable to an office building. He emphasized that the water-intensive nature of almonds is a concern, but the water used by data centers has significantly reduced since their reliance on evaporative cooling. In comparison, one ChatGPT response uses between 1 to 50 milliliters of water, which is far less than the estimated 85 to 5,500 chatbot queries per almond using old cooling methods.

While new technologies like Nvidia’s liquid cooling system and Microsoft’s closed-loop systems have further reduced water consumption in data centers, the increased adoption of AI agents, which utilize more computational resources than traditional chatbots, could potentially lead to higher water usage. Despite this, data centers in the U.S. have faced challenges with water consumption, with some sites using as much as 29 million gallons over 15 months.

Nonetheless, Altman highlighted that advanced cooling techniques, such as Nvidia’s and Microsoft’s, have reduced data center water usage by up to 100%.

Amazon has also claimed that its data centers consume only 0.075% of the water Americans use for their lawns and gardens. The podcast episode can be viewed in full for further details.