Sam Altman has made some bombshell revelations about the power of AI at the AI Impact Summit 2026. He spoke of superintelligence or AI agents operating out of powerful data centers that could even render the most intelligent tech entrepreneurs of our generation redundant. If that wasn’t scary enough, another comment of his has caught social media by storm.
During an interview at the Express Adda event, which was part of the India AI Impact Summit 2026 in New Delhi, a question about the environmental impact and the energy usage of large-scale AI models prompted Altman to respond with comparisons to human consumption. Although Altman is particularly cautious about the dangers of AI, the energy consumed by these models is not one of them. Let’s explore what the OpenAI CEO said that put social media users on edge.
20 years of consumption before humans get smart, Sam Altman says, while defending AI model energy consumption
One question that the OpenAI CEO had to grapple with was the energy needs of the AI models he has been working on. Notably, the company’s flagship model has been incurring huge infrastructure costs, and the massive energy needs are definitely at the centre of the conundrum. However, Altman was prepared for the question and said, “People talk about how much energy it takes to train an AI model … But it also takes a lot of energy to train a human. It takes like 20 years of life and all of the food you eat during that time before you get smart.”
The aforementioned comment from Altman sparked a huge debate on social media about whether it was bordering on offensive to bring up the consumption needs of humans into the discussion. For example, one social media user on X broke down the mathematics of the energy needs of one human compared to one AI model.
The X user explained that a human consumes roughly 17,000 kWh of total food energy over 20 years; meanwhile, GPT-4, developed by OpenAI, according to his estimates, consumes 50 GWh of electricity over the same time period, which is approximately equal to “3,000 humans worth of “training energy”.
Further arguing about how newer ChatGPT models are burning through energy, he stated, “They need you to think of AI energy consumption as natural and inevitable, the same way you think about feeding a child, because the alternative framing is that they’re burning through enough electricity to rival small countries while racing to build 1-gigawatt Stargate data centers.”
Earlier, Altman had argued that the “significant fraction of the power on Earth” will be required to power these future AI models. He has already stated that the current energy infrastructure is not enough to meet the needs of the AI future he has in mind, and that “energy innovation” is paramount for his vision.
Related: Jobs at Risk? Sam Altman Claims Superintelligence Could Do a Better Job Being CEO in 2 Years
AI model vs human – the energy consumption debate rumbles on…
Some users argued about the ethical implications of his comment at the AI summit.
Meanwhile, some X users argued that these AI models are more intelligent and purposeful than a lot of human beings on Earth, and we should readily accept them.
Clearly, social media is divided on Altman’s human vs AI model energy consumption debate. The Altman defenders agree that although energy consumption for these AI systems is very high, the purpose they will serve will far outweigh the costs.
However, his detractors also have a point when they say that human beings are more than just consumption units. Nevertheless, no one knows what the future holds, and all we can hope is that AI serves humans rather than the other way around.
