Sam Altman is pushing back against what he calls a persistent myth: that artificial intelligence data centers are water-guzzling environmental disasters. The OpenAI CEO said concerns about AI's water consumption don't hold up to scrutiny, comparing the water used by ChatGPT queries to the far more water-intensive process of growing California almonds.

Why Are People So Worried About AI and Water?

Public concern about data centers and water consumption is widespread and deeply rooted. A Gallup report published in May found that 71% of Americans said they opposed the construction of a data center in their area, with 70% saying they worried specifically about the environmental impact. When asked what environmental concerns mattered most, half of respondents cited the effect on resources, with 18% naming water specifically.

This anxiety isn't entirely unfounded. A Business Insider investigation published last year found that data centers were contributing to the water crisis in the West, in part because of water-intensive evaporative cooling techniques. The concern has become so prominent that it shapes public policy and community resistance to new data center projects.

What Does the Data Actually Show?

Altman's argument rests on a straightforward comparison: modern data centers use about as much water as an average office building. A 2024 report from a state government commission in Virginia, a major hub for data center development, found that while some facilities used significantly more water than others, most data centers used the same amount or less than an average large office building.

To illustrate his point, Altman drew an unusual parallel. He said growing a single almond in California requires more water than is used in thousands of ChatGPT queries. According to available research, a single almond requires an average of 3.56 liters, or roughly a gallon of water to grow, while the average ChatGPT query uses about 0.32 milliliters of water. By comparison, Google's Gemini app text prompt uses about 0.26 milliliters.

"That has been a robust meme and difficult to disprove, but I don't think holds up to any scrutiny," said Altman when discussing water consumption concerns during an interview on the Sources podcast with Alex Heath.

Sam Altman, CEO at OpenAI

How Is the Industry Responding to Water Concerns?

The data center industry acknowledges the challenge and says it's taking steps to address it. The Data Center Coalition, an industry trade group, says data center developers use various cooling methods, including some closed-loop or waterless systems, and that they prioritize efficient water practices that account for local water availability.

However, Altman acknowledged that the industry has work to do in changing public perception. He emphasized that the path forward isn't just about the facts, but about demonstrating value to consumers.

"I think the right way to get people to like something is to deliver them value," Altman said, adding that "the industry has got work to do in terms of how we make these products easy to use and easy for people to get a lot of value out of."

Sam Altman, CEO at OpenAI

Steps to Understanding AI's Real Environmental Impact

• Compare data center water use to other industries: Modern data centers use roughly the same amount of water as large office buildings, according to Virginia state government research, making direct comparisons to agriculture and other sectors essential for context.
• Examine cooling technology differences: Data centers employ various cooling methods, from water-intensive evaporative cooling to closed-loop and waterless systems, so water impact varies significantly by facility and location.
• Evaluate per-query consumption: A single ChatGPT query uses approximately 0.32 milliliters of water, while growing one California almond requires 3.56 liters, illustrating how AI's water footprint compares to everyday agricultural products.
• Consider local water availability: The Data Center Coalition emphasizes that efficient water practices must account for regional water scarcity, meaning environmental impact depends heavily on where facilities are built.

The disconnect between Altman's data-driven argument and public opinion reveals a broader challenge facing the AI industry. Even when the numbers suggest concerns are overblown, 71% of Americans still oppose new data center construction in their communities. This gap between technical reality and public perception suggests that facts alone may not be enough to shift attitudes. The industry's ability to build trust and demonstrate tangible benefits to local communities could ultimately matter more than statistical comparisons to almonds or office buildings.