“People are often Curious about the amount of energy used by a Chatgpt consultation ” Sam Altmanthe CEO of Openaiwrote aside in a Publication of the Long Block Last week. The average consultation, Altman wrote, uses 0.34 Watts-Hours of Energy: “About what an oven would use in just over a second, or a high efficiency bulb used in a couple of minutes.”
For a company with 800 million active weekly users (and farming), the question of what energy all these searches use is to become more and more pressing. But experts say that Altman’s figure does not mean much without much more public context of Openai on how he reached this calculation, including the definition of what is an “average” consultation, whether or not Image generation, and whether or not Altman includes additional energy use, such as AI models training and refrigeration of Openai servers.
As a result, Sasha Lucioni, the climate leadership of the AI company that embraces the face, does not put too much stock in the number of Altman. “I could have taken it out of the ass,” he says. (Openai did not respond to a request for more information on how it reached this number.)
As AI takes care of our lives, it is also promising to transform our energy systems, overcoming carbon emissions properly as we try to combat climate change. Now, a growing new research body is trying to put hard numbers on the amount of carbon we really emit with all our use of AI.
This effort is complicated by the fact that the main players such as Openai disseminate little environmental information. An analysis presented for peer review this week by Lucioni and three other authors take into account the need for more environmental transparency in AI models. In the new Lucioni analysis, she and her colleagues use data from Open -rodA Grandience traffic classification (LLM), to find that 84 percent of the use of LLM in May 2025 was for models with zero environmental dissemination. This means that consumers are overwhelming models with completely unknown environmental impacts.
“I blow my mind that you can buy a car and know how many kilometers per Galó consumes, but we use all these tools of the AI every day and that we have absolutely any metric of efficiency, emission factors, nothing,” says Lucioni. “He is not obligatory, he is not regulatory. Given where we are with the climate crisis, he should be at the top of the regulators’ agenda everywhere.”
As a result of this lack of transparency, Lucioni says, the public is exposed to estimates that are meaningless, but are taken as a gospel. You may have heard, for example, that the average application of Chatgpt takes 10 times more energy than the average Google search. Lucioni and his colleagues track this statement to a public observation, which John Hennessy, the President of Alphabet, the Google’s parent company, did in 2023.
A claim made by a company member member (Google) on the product of another company to which it has no relationship (Openai) is faint at best: however, the analysis of Lucioni Find, this figure has been repeated again and again in press and policies reports. (As I wrote this piece, I had a playing field with this exact statistics.)
