OpenAI CEO Sam Altman addressed rising concerns about artificial intelligence’s environmental impact during the Express Adda event hosted by The Indian Express in India. While affirming that AI’s total energy consumption is a legitimate topic for discussion, Altman reframed the conversation by drawing a direct parallel to the enormous resources required to “train” a human being.
Altman strongly refuted sensational claims circulating online, such as assertions that a single ChatGPT query consumes “17 gallons of water” or the energy equivalent of “1.5 iPhone battery charges.” Altman labeled these figures “completely untrue, totally insane, no connection to reality.” He explained that older water-usage concerns came from evaporative cooling in data centers, but modern facilities have largely eliminated that issue.
Altman also acknowledged that overall energy demand from widespread AI usage is a fair point of worry. He pressed for rapid investment in cleaner sources, suggesting a “move towards nuclear or wind and solar very quickly.”
Altman reframes the comparison to humans vs AI
The core of Altman’s defense lay in challenging what he described as an unfair apples-to-oranges comparison. Critics frequently contrast the massive electricity needed to train large AI models with the tiny amount a human brain uses for a single inference task.
“But it also takes a lot of energy to train a human,” Altman countered. “It takes like 20 years of life and all of the food you eat during that time before you get smart. And not only that, it took the very widespread evolution of the 100 billion people that have ever lived” who built cumulative knowledge in survival, science, mathematics, and more.
He proposed a more equitable benchmark, i.e., compare the energy cost of one trained AI answering a question versus one trained human performing the identical task. By that measure, Altman argued, “AI has probably already caught up on an energy efficiency basis.”
Altmans also spoke about data centers in space
During the same interaction at the India AI Impact Summit 2026, Altman was asked about proposals to place data centers in orbit—ideas championed by rivals like Elon Musk (via SpaceX) and explored by Google. He dismissed the concept as impractical in the current landscape.
“I honestly think the idea of putting data centres in space is ridiculous,” Altman said, citing prohibitive launch costs compared to terrestrial power generation and the near-impossibility of repairing hardware like broken GPUs in orbit. “If you just do the rough math of launch costs relative to the cost of power we can do on Earth, just say nothing of how you’re gonna fix a broken GPU in space, we are not there yet.”
He acknowledged that space could play a role in the future for certain applications but insisted “orbital data centres are not something that’s going to matter at scale this decade.”