AI’s E-Waste Crisis: New Research Predicts a 2,000-Fold Surge to 5 Million Tonnes by 2030

The explosion of artificial intelligence (AI) over the past few years has brought many benefits to humanity. However, humans are also facing a major concern that has not yet received due attention: what will we do with the huge amount of electronic waste we have created?

“The e-waste generated by AI, especially large language models (LLMs), could increase significantly and potentially reach 2.5 million tons per year by 2030 if no waste reduction measures are taken,” said Asaf Tzachor, a climate and sustainability researcher at Reichmen University in Israel.

Mr. Tzachor’s research team discovered that the amount of electronic waste from AI servers could reach 5 million tons by the end of this decade, about 2,000 times the amount of waste generated by AI in 2023, according to IFLScience on November 14.

The United Nations considers all types of e-waste in general a major problem, saying that “the 62 million tonnes of e-waste generated in 2022 could fill 1.55 million 40-tonne trucks, enough to line up around the equator.”

Tackling the e-waste problem has many benefits. First, it prevents millions of people, including children, from dying from exposure to toxic chemicals and pollution from e-waste. At the same time, e-waste is also a valuable resource and mineral that should be recycled.

“A tonne of iPhones contains more gold and silver than a tonne of ore from a gold or silver mine,” said Lisa McLean, a member of research group Circular Australia.

According to research, the adoption of circular economy strategies reduces e-waste by 16% and in the best-case scenario this can be as high as 86%. The circular economy is an economic model in which design, production and service activities aim to extend the life of materials and eliminate negative impacts on the environment.

The study makes clear that the e-waste crisis is global in nature, which is why it is important to focus on cross-border e-waste management.

There is no one-size-fits-all solution, but we should aim to extend the life of existing hardware and reuse or refurbish devices and components, Tzachor said.

“It is much easier and more cost-effective to address the e-waste challenges created by AI now before they get out of control,” Tzachor added.

The study was published in the journal Nature Computational Science.