The passage begins by discussing how energy consumption of AI queries is a topic of growing interest for companies due to the increasing number of weekly active users on platforms like Hugging Face, OpenAI, and others. Sam Altman, the CEO of OpenAI, made a informal aside in his blog post to hint that these queries use about 0.34 watt-hours of energy. He compares this to a microwave oven operating for about a second or a high-efficiency light bulb operating for about a couple of minutes.
Altman’s figure is derived from statistical averaging of the energy usage of over a million weekly active queries, with the average query estimated to power a microwave oven in a fraction of a second. He acknowledges that this is a rough estimate and suggests that more detailed context would be necessary to refine it.
Altman emphasizes that this metric alone does not provide enough context for public understanding. He points out that the concept of "average" uses in this context is not universally defined, and the use of such metrics can vary significantly, including whether they include image generation, which is particularly notable for platforms with collaborative AI development projects.
One expert, Sasha Luccioni, a climate leadership finalist at Hugging Face, has expressed skepticism toward Altman’s number. Luccioni comments that her assertion, while Steamy, certainly has a ring of水分. Altman has not responded to external requests for more information.
Altman acknowledges the importance of telling more about the methods and definitions used to calculate his number, including details about how he arrived at the figure, whether image generation is included, and what metrics were considered for additional energy use, such as training AI models or managing AI server cooling.
As AI systems increasingly take over tasks in our daily lives, there is a growing intuition that they could be reshaping energy systems in a way that directly contributes to our environment. This reshape is not yet understood, especially since major players like OpenAI and Hugging Face do not disclose environmental transparency regarding their AI products.
A growing field of research is addressing how to quantify the carbon emissions associated with AI usage. The field is complex, despite the fact that companies like OpenAI have sole responsibility for not disclosing their environmental impact.
The edge of AI’s influence on our lives, particularly as AI transforms energy systems, poses a significant challenge for carbon emissions. The efforts behind creating hard numbers on this issue highlight the need for stronger environmental transparency in AI development.
The analysis submitted for peer review by Sasha Luccioni and three co-authors examines the need for more environmental transparency in AI models. Using data from OpenRouter, the team discovers that only 84 percent of large language models (LLMs) in May 2025 used LLMs with zero environmental disclosure. This finding underscores the importance of such transparency.
Luccioni compares this 84% figure to other statistics on AI-related carbon emissions, such as the claim that a ChatGPT request consumes 10 times more energy than a Google search. She argues that while these numbers are not groundless and are frequently cited in press reports, they produce estimates that seem nonsensical or even.getElementByIdifying.
Luccioni’s perspective is humanity-driven, as she sees the broader carbon footprint of AI-driven innovations as something that deserves attention. As a result, she and her colleagues argue that society must prioritize regulatory action to*)
Summary:
In “People are often curious about how much energy a ChatGPT query uses,” Sam Altman introduced insights about the energy consumption of AI queries. He suggested that an average query uses around 0.34 watt-hours, equivalent to a microwave oven’s performance or a high-efficiency light bulb’s use over a couple of minutes. Experts, however, expressed skepticism, finding the metrics inconsistent and lacking context on their sources. OpenAI emphasized the importance of defining these terms and mentioned the need for more environmental transparency in AI development. The growing field of research aims to create quantifiable carbon footprints for AI usage, despite companies like OpenAI clashing with experts on definitions and public perception. This transformation of energy systems by AI serves as a significant challenge for environmental sustainability.
