Sam Altman Dismisses AI Water Usage Concerns as Overblown

OpenAI Chief Executive Sam Altman has publicly dismissed mounting concerns regarding artificial intelligence data centers excessive water consumption, characterizing the controversy as an unfounded narrative that withstands neither technical scrutiny nor empirical data. Addressing the issue during the premiere episode of the Sources podcast with The Verge Alex Heath, Altman stated that modern computational facilities utilize water volumes comparable to those of standard commercial office buildings. He acknowledged the persistence of the water usage concerns but emphasized that rigorous analysis contradicts the perception of AI infrastructure as resource guzzlers. The debate has intensified amid widespread infrastructure expansion across the United States. Recent polling indicates substantial public resistance to localized data center development. A May Gallup survey revealed that seventy-one percent of Americans oppose the construction of a data center within their communities, with environmental impact cited as a primary driver. Approximately seventy percent expressed specific environmental worries, while eighteen percent identified water scarcity as a direct concern. This apprehension is particularly pronounced in the Western United States and Virginia, regions experiencing rapid data center growth and documented strain on local water resources. A 2024 report by a Virginia state commission confirmed that while water consumption varies significantly between facilities, the majority of modern data centers operate at or below the water usage levels of large office complexes. Industry stakeholders maintain that technological advancements and regional adaptation are mitigating environmental impact. The Data Center Coalition noted that developers increasingly deploy closed-loop cooling systems and waterless alternatives, optimizing resource allocation based on local hydrological conditions. While some older or less optimized facilities rely on evaporative cooling, which draws heavily on regional supplies, sector leaders argue that next-generation infrastructure prioritizes efficiency. To contextualize the per-query environmental footprint, Altman compared AI processing to agricultural water consumption. He estimated that a single ChatGPT query utilizes approximately 0.32 milliliters of water, whereas producing a single almond in California requires roughly 3.56 liters. While independent verification of the exact comparative metrics varies, broader hydrological studies consistently indicate that agricultural irrigation and food production far exceed the water intensity of individual computational requests. Despite asserting that the data on water consumption is clear, Altman recognized that public perception remains the primary hurdle for industry adoption. He argued that technological reassurance must be paired with tangible utility. The path forward, he suggested, relies on the industry delivering demonstrable value and simplifying user experience to reframe the narrative around artificial intelligence. As computational demand continues to scale, the sector faces mounting pressure to transparently communicate sustainability metrics while advancing cooling technologies that align with regional environmental constraints.