“People are often I wonder how much energy the chatgpt inquiry uses, “Altman himself, the general director of Opeli, wrote aside in Long blog post last week. Altman wrote that the average inquiry uses 0.34 watt-hour of energy: “About what the oven would use in just over one second, or a high-performance bulb would use in a few minutes.”
For a company with 800 million vigorous users (i development), The question of how much energy all these searches apply is becoming more and more pressing. But experts say that the number of Altman does not mean much without a major public context on the part of OpenAI about how these calculations have occurred – including the definition of the “average” query, regardless of whether it covers image generation and whether Altman includes additional energy consumption, as in the case of AI models training and cooling OPENAI servers.
As a result, Sasha Luccioni, a climate leader in AI Company Hulging Face, does not place too much altman inventory. “He could get it out of his ass,” he says. (OpenAI did not answer the request for more information about how this number occurred).
When AI takes over our lives, he also promises to transform our energy systems, recharging carbon dioxide emissions when we try to fight climate change. Now a up-to-date and growing group of research is trying to apply demanding numbers on how much coal we actually emit with all our artificial intelligence.
This effort is complicated by the fact that the main players, such as OpenAi, reveal diminutive information about the environment. The analysis submitted for review this week by Luccioni and three other authors analyzes the need for greater environmental transparency in AI models. In the up-to-date analysis of Luccioni, she and her colleagues apply data from OpenRouterThe leader plaque of the huge language model movement (LLM) to state that 84 percent of the LLM application in May 2025 was for models with zero environmental disclosure. This means that consumers mostly choose models with completely unknown environmental influence.
“It believes that you can buy a car and know how many miles for a gallon we use, but we use all these AI tools every day and we have absolutely no performance indicators, emission factors, nothing,” says Luccioni. “This is not ordered, it is not regulatory. Considering where we are with a climate crisis, it should be at the top of the regulatory authorities program around the world.”
As a result of the lack of transparency, Luccioni says that society is exposed to estimates that do not make sense, but are treated as the Gospel. You may have heard, for example, that the average CHATGPT demand requires 10 times more energy than the average Google search. Luccioni and her colleagues follow this claim to be public that John Hennessy, chairman of Alphabet, home company Google, produced in 2023.
The claim submitted by a member of the management board from one company (Google) about the product of another company to which there is no relationship (OpenAI), is at most uncertain – but, as Luccioni’s analysis stated, this number was repeated in the press and principles reports. (When I wrote this song, I have a pitch with this exact statistics.)