New analysis suggests artificial intelligence systems are rapidly becoming one of the world’s largest drivers of electricity demand, with AI-related computing potentially overtaking bitcoin mining energy consumption by the end of the year.
By yourNEWS Media Newsroom
Artificial intelligence systems may already account for as much as 20% of worldwide data center electricity consumption, and that share could climb dramatically by the end of 2026 as AI infrastructure expands at a historic pace, according to new research published in the journal Joule.
The findings, highlighted in a report by Wired, estimate that AI could consume nearly half of all global data center electricity by the end of this year, excluding energy used for bitcoin mining.
The analysis was conducted by Alex de Vries-Gao, founder of Digiconomist, a research organization focused on the environmental impact of emerging technologies.
De Vries-Gao previously examined the energy demands of cryptocurrency mining but said the rapid expansion of AI systems such as ChatGPT and other large language models has created a much larger and faster-growing energy challenge.
“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],” de Vries-Gao said. “This is just escalating a lot faster, and it’s a much bigger threat.”
The research estimates that AI systems could consume up to 82 terawatt-hours of electricity this year if hardware production remains stable — roughly equivalent to the annual electricity use of Switzerland.
If production of AI hardware doubles as projected by analysts, AI electricity consumption could rise at a similar rate, potentially accounting for nearly half of all data center energy demand worldwide.
The report comes amid mounting concern over the environmental and infrastructure impact of the AI boom, which has triggered an unprecedented expansion of global data center construction.
According to a recent report from the International Energy Agency, data centers accounted for approximately 1.5% of total global electricity consumption in 2024, using roughly 415 terawatt-hours of power annually — slightly below the yearly electricity demand of Saudi Arabia.
The IEA projected that data center electricity consumption could exceed 900 terawatt-hours annually by the end of the decade.
Researchers noted that electricity demand from data centers has grown four times faster than overall global electricity demand in recent years, driven largely by investments tied to AI infrastructure.
De Vries-Gao approached the problem by analyzing the semiconductor supply chain rather than attempting to calculate energy use from individual AI searches or software queries.
The research focused heavily on TSMC, the Taiwan Semiconductor Manufacturing Company, which dominates production of advanced AI chips used by companies including Nvidia, Google, and AMD.
Using production estimates, earnings call data, device specifications, and publicly available electricity profiles for AI hardware, the analysis attempted to estimate how much global energy demand is now tied specifically to artificial intelligence systems.
Researchers acknowledged major uncertainties remain because technology companies generally do not publicly disclose detailed electricity consumption data for their AI systems.
Sasha Luccioni cautioned that accurately calculating AI energy demand remains difficult because so much operational information is kept private by major technology firms.
“It’s because we don’t have the information that [researchers] have to do this,” Luccioni said. “That’s why the error bar is so huge.”
The rapid growth of AI infrastructure is already complicating climate goals established by major technology companies.
Google disclosed in its 2024 environmental report that its greenhouse gas emissions have risen 48% since 2019, driven in large part by increased energy demand tied to AI expansion.
“As we further integrate AI into our products, reducing emissions may be challenging due to increasing energy demands from the greater intensity of AI compute,” the company stated in the report.
Google previously disclosed in a 2022 research paper that machine learning systems accounted for between 10% and 15% of the company’s total electricity use between 2019 and 2021. Researchers noted, however, that the disclosure came before the large-scale deployment of newer AI systems such as Gemini.
Environmental concerns tied to AI infrastructure increasingly extend beyond electricity usage itself.
Large data centers require enormous cooling systems, water consumption, transmission infrastructure, backup diesel generators, and expanding semiconductor manufacturing operations, all of which carry additional environmental and resource impacts.
Critics argue the pace of AI infrastructure deployment is far outstripping public understanding of its long-term effects on power grids, emissions targets, water systems, and local communities.
The issue has also begun attracting political attention. Alex Jones discussed the research in a video post, warning that expanding AI infrastructure could dramatically reshape cities, energy systems, and industrial development.
Watch below:
De Vries-Gao said greater transparency from major technology firms would significantly improve researchers’ ability to measure AI’s environmental footprint.
“You really have to deep-dive into the semiconductor supply chain to be able to make any sensible statement about the energy demand of AI,” he said. “If these big tech companies were just publishing the same information that Google was publishing three years ago, we would have a pretty good indicator of AI’s energy use.”
Original article
Comments