Sam Altman Pushes Back on AI Water Fears, Urges Clean Energy Shift

​Water Claims ‘Have No Connection to Reality’
Sam Altman is pushing back hard against growing criticism over artificial intelligence’s environmental footprint. Speaking at an AI summit in New Delhi, the OpenAI chief dismissed viral claims that ChatGPT uses gallons of water per query, calling them “completely untrue” and detached from reality. While data centers have historically relied on water for cooling, he argued that per-query water panic has been wildly overstated.

Energy Use Is Rising — and That’s the Real Issue
Altman did acknowledge a more serious concern: total energy consumption. As AI adoption accelerates globally, overall electricity demand tied to data centers is climbing fast. His solution is not to slow AI development, but to accelerate investment in cleaner power sources, including nuclear, wind, and solar.

Humans Use Energy Too
One of Altman’s more controversial arguments compared AI energy use to human energy consumption. Training an AI model requires significant power, he noted — but so does raising and educating a human over two decades. Inference, the process of generating answers from an already trained model, is far less energy-intensive. Measured that way, he suggested, AI may already be competitive with humans on an energy-efficiency basis for certain tasks.

Backlash Builds as Data Centers Expand
The debate comes as governments and tech companies pour billions into new data centers to support AI workloads. Electricity use from data centers has already reached levels comparable to entire developed nations, and projections show demand continuing to surge. Some communities are pushing back over grid strain and rising costs, with at least one proposed billion-dollar data center recently voted down after public opposition.

AI’s Growth Collides With Climate Goals
​Altman’s defense underscores a bigger tension: AI’s explosive growth versus environmental and infrastructure limits. As countries race to build computing capacity, they must also expand energy production without derailing climate targets. The question isn’t whether AI will consume power — it will — but whether the energy system can scale fast enough, and cleanly enough, to support it.