By 2027, yearly power demands of AI could surpass those of small countries

By 2027, the electricity powering artificial intelligence (AI)-related activities could exceed the yearly power demands of countries such as the Netherlands, Argentina and Sweden, new research finds.

Wide adoption of AI could have a large energy footprint as generative AI technologies such as ChatGPT that create content based on the data they were trained on have been seeing rapid growth since 2022, noted the research published as a commentary in the journal Joule.

Training these tools is an energy intensive process as it requires feeding the models a large amount of data, said author Alex de Vries, a researcher at Vrije Universiteit Amsterdam in the Netherlands.

To train its textual content-generating AI tool Hugging Face, a New York-based AI company reported that it consumed 433 megawatt-hours (MWh), enough to power 40 average American homes for a year, said de Vries, who also founded Digiconomist, a research company dedicated to exposing the unintended consequences of digital trends.

Further, these AI tools again take up a significant amount of computing power, and thus energy, every time they generate a text or image based on prompts.

For example, ChatGPT could cost 564 MWh of electricity a day to run, de Vries' analysis showed. He also estimated that if Google, which processes up to 9 billion searches a day currently, were to power its search engines with AI, it would need roughly 30 TWh of power a year, which is almost the entire yearly electricity consumption of Ireland.

Google has been incorporating generative AI in the company's email service and is testing out powering its searches with AI.

Thus, by 2027, worldwide AI-related electricity consumption could increase by 85 to 134 TWh annually, said de Vries, based on the projection of the rate of AI-server production, which is expected grow rapidly in the near future.

Also, even as companies globally are working on making AI hardware and software more efficient and less energy intensive, de Vries said that this only increases their demand.

"The result of making these tools more efficient and accessible can be that we just allow more applications of it and more people to use it," said de Vries in the commentary.

Thus, "looking at the growing demand for AI service, it's very likely that energy consumption related to AI will significantly increase in the coming years.

"The potential growth highlights that we need to be very mindful about what we use AI for. It's energy intensive, so we don't want to put it in all kinds of things where we don't actually need it," said de Vries.