AI’s power surge sparks sustainability concerns across tech industry

AI’s energy consumption is emerging as a significant environmental concern, with new research indicating it now represents 20% of global data center power demand and is set to surpass Bitcoin’s energy use by the end of 2024. This surge in energy consumption threatens tech companies’ climate goals and highlights the growing tension between rapid AI advancement and environmental sustainability in an industry that’s already consuming as much electricity as entire countries.

The big picture: New research published in the journal Joule reveals AI’s rapidly escalating energy footprint, which could double by year-end to comprise nearly half of all global data center electricity consumption.

• The study by Alex de Vries-Gao, founder of Digiconomist, calculates that AI will consume up to 82 terrawatt-hours of electricity this year—roughly equivalent to Switzerland’s annual electricity usage.
• If production capacity for AI hardware doubles as projected, the energy demand could increase at a similar rate, potentially representing almost 50% of all data center demand by December.

Why this matters: Tech giants’ ambitious climate goals are being undermined by their own AI investments, creating a significant contradiction in corporate sustainability strategies.

• Google‘s greenhouse gas emissions have increased by 48% since 2019, complicating its goal of reaching net zero by 2030.
• The rapid escalation of AI energy use represents what de Vries-Gao calls “a much bigger threat” than Bitcoin mining, which until now has been the poster child for excessive computational energy consumption.

Behind the numbers: Data centers already consume a substantial portion of global electricity, with demand growing rapidly.

• According to the International Energy Agency, data centers accounted for 1.5% of global energy use in 2024—around 415 terrawatt-hours, slightly less than Saudi Arabia’s yearly energy demand.
• The IEA predicts data center electricity consumption will more than double to over 900 TWh by 2030.

Counterpoints: Some researchers caution against drawing definitive conclusions about AI’s energy impact given the limited transparency from tech companies.

• Sasha Luccioni, an AI and energy researcher, emphasizes that disclosure from tech giants is crucial to accurately calculating AI’s true energy footprint.
• The rapid pace of technological development and deployment makes precise forecasting challenging.

Reading between the lines: The financial resources being directed toward AI dwarf previous computational investments, suggesting the energy problem could accelerate even faster than predicted.

• De Vries-Gao notes that “the money that bitcoin miners had to get to where they are today is peanuts compared to the money that Google and Microsoft and all these big tech companies are pouring in [to AI].”