OpenAI CEO Sam Altman has characterized public opposition to data centers as a reputation problem rather than a water crisis, asserting that modern facilities consume roughly the same amount of water as a large office building. Speaking on the premiere episode of Alex Heath's Sources podcast, Altman dismissed concerns about artificial intelligence water consumption as a meme that does not withstand scrutiny.
Altman's statements arrive amid widespread public backlash against data center construction. A Gallup poll conducted in May 2026 indicated that 71 percent of Americans oppose building data centers in their local area, while 70 percent expressed concerns regarding environmental impacts. Of those expressing concern, 18 percent specifically cited water usage. Altman also compared the water consumption of ChatGPT to growing California almonds, claiming that a single almond requires more water than thousands of ChatGPT queries.
Empirical Estimates and Public Opposition
Altman estimated the water usage of an average ChatGPT query at 0.32 milliliters, a figure closely aligned with Google's estimate of 0.26 milliliters for an average Gemini text prompt. To contextualize agricultural consumption, a 2019 study by U.S. Geological Survey researchers found that growing a single California almond requires approximately 3.56 liters of water. On the infrastructure side, a 2024 Virginia state government commission report concluded that most data centers use comparable or less water than an average large office building.
However, investigations present a contrasting picture of regional impact. A Business Insider investigation published in 2025 found that data centers contribute to water scarcity in the western United States. While Business Insider could not independently verify Sam Altman's comparison between ChatGPT queries and almond water usage, the underlying figures generally align with published metrics.
Altman previously addressed AI energy consumption by comparing the energy required to train an AI model to the energy needed to train a human over 20 years. In February 2026, he stated that concerns about water usage were totally fake, attributing past utility strains to outdated evaporative cooling methods.
Environmental Backlash and Stalled Investments
Environmental groups and local communities have voiced significant opposition regarding the environmental footprint of data centers, particularly concerning water and electricity consumption. In December 2025, over 230 environmental groups called for a moratorium on data center construction across the United States. Noman Bashir, a computing and climate impact fellow at MIT's climate and sustainability consortium, argued that the rapid demand for new data centers cannot be met sustainably, implying inevitable reliance on fossil fuel-based power plants.
Public resistance has translated into tangible policy roadblocks. As of July 2026, over 120 proposed moratoriums on data center development have emerged across 38 states. Furthermore, at least 48 data center projects have been blocked or stalled, representing $156 billion in capital investment, according to industry tracking data. These regulatory hurdles unfold as OpenAI projects its planned computing expenditure for 2026 to reach $50 billion.
For communities across the United States, the continued expansion of AI data centers despite public opposition could lead to increased pressure on local water and electricity resources, potentially resulting in higher utility costs for residents. For OpenAI and other technology firms, growing public and regulatory resistance poses a significant challenge to infrastructure development plans, potentially impacting their ability to scale AI models and cloud capacity.
