This article is brought to you by Datawrapper, a data visualization tool for creating charts, maps, and tables. Learn more.

All Blog Categories

How much energy are you using on AI?

Portrait of Mirko Lorenz
Mirko Lorenz

Hi, this is Mirko, co-founder of Datawrapper. This week I looked at how much energy is used for one ChatGPT request. Given that we all use AI chatbots more and more, should we worry about creating unsustainable demand in the future?

You may have heard that a single query to ChatGPT can use up to 10 times as much energy as a Google search. This comparison is surprisingly sticky — but is it reason to worry? Should it even make us consider not using AI chatbots at all?

In practical terms, the energy requirements of ChatGPT use are negligible compared to other household appliances. And this applies to your overall personal footprint as well: If you want to reduce your energy usage there are many other lifestyle changes that have a bigger effect, like going vegan, avoiding flights, or using a bike instead of car.

But what if everyone uses AI? Globally, AI-driven data centers are expected to more than double their electricity demand by 2030, reaching nearly 1,000 terawatt-hours (TWh) — about as much as the entire country of Japan.

However, even large-scale AI operations still represent only a fraction of national or global electricity use — often less than 0.2% of U.S. demand. Based on a recent energy report, even a fast growth of AI data centers will not surpass other energy-hungry areas of the economy.

Arriving at these figures is far from straightforward. As Andy Masley highlights, calculating the per-request impact of ChatGPT requires careful estimation, diverse data sources, and a lot of comparison work. The size of the AI model, the efficiency of the data center, its cooling methods, and even its local climate all have an impact. There is no single, universally agreed-upon number, but the consensus is clear: The per-use energy impact of AI chatbots is small, and the bigger picture is complex and evolving.

The efficiency and analytical powers of AI tools also represent a new way to use the information we have about climate change, hunger, or pollution. In the best case, AI could be a tool to solve the undeniably big problems we're already facing.


If you're concerned about your energy usage, you can keep using ChatGPT for now — but remember to ask the right questions. See you next week!

Further reading

Andy Masley: Using ChatGPT is not bad for the environment

Andy Masley: Replies to criticisms of my posts on ChatGPT & the environment

Hannah Ritchie: What’s the impact of artificial intelligence on energy demand?

United Nations Environment Programme: AI has an environmental problem, here is what the world can do about it

Portrait of Mirko Lorenz

Mirko Lorenz (he/him, @mirkolorenz@vis.social, LinkedIn) is a co-founder of Datawrapper and a member of the advisory board. Before that, he served as CEO and later co-CEO of the company for almost 10 years since its inception in 2012. Mirko lives in Cologne.

Sign up to our newsletters to get notified about everything new on our blog.