Sam Altman, the CEO of OpenAI, has challenged criticisms about the energy consumption of artificial intelligence systems like ChatGPT. Speaking at a major AI summit, Altman argued that comparing the energy cost of training an AI model to a single human query is fundamentally flawed, as it ignores the immense energy and time invested in human education and evolution.
Altman made the comments in an interview with The Indian Express, where he directly addressed growing scrutiny over the environmental impact of large-scale AI operations. He also labelled online claims about ChatGPT's water usage as "totally fake."
Defending AI's Energy Footprint
Altman's core argument centred on the long-term investment in human capability. "It takes a lot of energy to train a human," he stated, prompting laughter from the audience. "It takes, like, 20 years of life, and all of the food you eat during that time before you get smart." He extended this perspective back thousands of years, referencing the cumulative experience of humanity.
He elaborated that the process encompasses "the very widespread evolution of the 100 billion people that have ever lived and learned not to get eaten by predators and learned how to, like, figure out science or whatever."
Addressing Water and Power Consumption Claims
The OpenAI leader was particularly dismissive of viral claims regarding water usage. "Water is totally fake," Altman said when asked about concerns. He explained that while older data centres used evaporative cooling, this practice has been discontinued, rendering such estimates obsolete.
On electricity, Altman provided a concrete figure, stating on social media platform X in June that the average ChatGPT query uses about 0.34 watt-hours. He compared this to the energy needed to power a lightbulb for a couple of minutes or an oven for just over one second.
Advocacy for Alternative Energy and Industry Context
Despite his defence of AI's relative efficiency, Altman acknowledged the industry's overall growing energy demand is a valid concern. He cited this as the reason he and other AI executives are championing alternative energy sources like solar, wind, and nuclear power.
Unlike some peers, such as xAI's Elon Musk, Altman is sceptical about the near-term feasibility of space-based data centres to reduce energy consumption. Outside of OpenAI, Altman is a major investor in nuclear energy, having served as chairman of startup Oklo and backed fusion company Helion.
The debate occurs against a backdrop of increasing attention on data centre energy use. Consulting firm McKinsey & Company estimated last year that data centres could account for 14% of total US power demand by 2050. In the US, the issue has reached the political level, with President Donald Trump recently stating he was working with tech companies on a commitment to prevent citizens from paying higher bills due to nearby data centres.