Electronic waste generated by the artificial intelligence boom has been significantly underestimated. A new report from the nonprofit Basel Action Network warns that discarded hardware could fill 23 million shipping containers by 2050. These 40-foot containers would circle the world six times if lined up in a row.
Previous studies missed a large portion of the infrastructure required to run modern data centers. The new paper factors in all the supporting hardware needed for servers. Jim Puckett is the founder and chief of strategic direction at the Basel Action Network. He issued a warning about the physical reality behind digital tools in a press release.
AI may feel weightless, but every model depends on an enormous amount of highly specialized, cutting edge hardware, Puckett says. If companies and governments do not begin planning for this new waste tsunami, today’s AI buildout could become an even more cataclysmic toxic waste crisis than we are already experiencing.
Globally, less than a quarter of the 68.3 million tons of e-waste created each year is formally collected and recycled. Most discarded electronics end up in informal waste collection sites. Workers burn or bury this equipment, exposing local environments to toxic materials like lead and chromium.
The United States operates far more data centers than any other nation. However, the country has not ratified the Basel Convention, which limits the international trade of hazardous wastes. Investigations show that United States recyclers still ship e-waste abroad. Much of that material ends up in backyard recycling operations.
The World Health Organization reports that millions of children forced to work or live nearby face severe health threats from informal recycling. Total global e-waste will triple to reach up to 211 million metric tons a year by 2050. Artificial intelligence will account for about 15 to 20 percent of that trash.
Past estimates only focused on servers and graphics processing units. The new report expands its count to include power supply and distribution, cooling systems, backup power, and networking equipment. It also counts what the organization calls AI Waste Contagion. This category includes telecommunications infrastructure and personal devices that will become obsolete faster because of AI advances.
The Basel Action Network calculates 70,000 metric tons of e-waste per gigawatt of data center capacity. McKinsey projects total data center capacity will reach up to 219 gigawatts by 2030. Considering equipment retired between 2025 and 2050, the total accumulation will reach between 395 and 617 million metric tons by the middle of the century.
This total breaks down to roughly 8.6 million to 13.1 million metric tons of retired equipment each year. Past studies omitted about 87 percent of a data center electro-mechanical infrastructure by focusing solely on servers and accelerators. A separate 2024 study predicted AI e-waste would reach 1.2 million to 5 million tons by 2030. Another study from February found that AI servers alone could create up to 225,000 tons annually by the end of the decade.
The Basel Action Network published its findings today.
