AI-related hardware could turn into 8.6 to 13.1 million tonnes of electronic waste every year by 2030, environmental group Basel Action Network (BAN) estimates in a report. Most of that may come not from data centers but from computers, smartphones and other devices in homes and offices.
The reason is how quickly hardware gets replaced. AI data centers run on servers, accelerators, networking gear, and cooling and power systems, and some of that equipment has to be swapped out within just a few years.
Still, the report’s authors tie most of the waste to what happens outside data centers. As AI demands grow, they expect ordinary PCs, smartphones and telecom equipment to become obsolete faster, too. New AI features, for instance, could push businesses and consumers to upgrade their electronics sooner.
That faster replacement cycle explains the 13.1 million tonne upper bound. By BAN’s calculations, data center infrastructure will account for around 2.8 million tonnes of discarded equipment in 2030, while upgrades of other devices will add another 5.8 to 10.3 million tonnes.
Follow Mediasat on Google News: the latest in TV, telecom and tech
Discarded electronics can contain lead, cadmium, mercury and other hazardous substances. Yet only about a fifth of the world’s e-waste is currently recycled or disposed of properly, the group notes.
One earlier estimate was far more modest. It put generative AI’s e-waste at up to 2.5 million tonnes a year by 2030, but that study focused mainly on server hardware.
Data centers are drawing environmentalists’ attention beyond Earth, too. In July, a coalition of environmental groups urged the US Federal Communications Commission (FCC) to pause licensing of orbital data centers.
