AI data centers could generate 13 million tons of e-waste yearly

Artificial intelligence infrastructure will generate far more electronic waste than previously calculated, according to new research that expands the scope of what gets counted as AI-related trash.

The Basel Action Network, a nonprofit focused on hazardous waste, projects AI could contribute 8.6 million to 13.1 million metric tons of e-waste each year by 2050. Between 2025 and 2050, that adds up to 395 to 617 million metric tons total—enough to fill 23 million shipping containers that would circle the globe six times if placed end to end.

The estimate dwarfs previous projections because it accounts for the full data center ecosystem, not just servers and GPUs. The analysis includes power distribution systems, cooling infrastructure, backup generators, networking equipment, and what researchers call "AI Waste Contagion"—telecommunications gear and consumer devices made obsolete by AI advances.

Why it matters

Less than 25 percent of the world's 68.3 million tons of annual e-waste gets formally recycled. The rest enters informal collection channels where burning or burial exposes workers and communities to lead, chromium, and other toxins. The United States, home to more data centers than any other nation, has not ratified the Basel Convention limiting international hazardous waste trade. American recyclers continue shipping e-waste abroad where it often ends up in dangerous "backyard recycling" operations, according to investigations cited in the report.

Previous estimates missed 87 percent of infrastructure

Earlier studies focused narrowly on compute hardware. A 2024 analysis predicted AI e-waste could reach 1.2 million to 5 million tons by 2030. Another study published in February estimated AI servers alone would generate 131,000 to 225,000 tons annually by decade's end—comparable to Denmark's total e-waste output.

But servers and accelerators represent only about 13 percent of a data center's electro-mechanical infrastructure, according to the Basel Action Network. The organization calculates 70,000 metric tons of e-waste per gigawatt of data center capacity when all supporting systems are included. With McKinsey projecting total data center capacity reaching up to 219 gigawatts by 2030, the waste volumes multiply rapidly.

A waste tsunami without planning

"AI may feel weightless, but every model depends on an enormous amount of highly specialized, cutting edge hardware," said Jim Puckett, founder and chief of strategic direction at the Basel Action Network, in a statement. "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."

The World Health Organization has documented health threats to millions of children forced to work or live near informal e-waste recycling sites. The Basel Action Network predicts global e-waste will triple to 211 million metric tons annually by 2050, with AI accounting for 15 to 20 percent of that total.

These findings were first reported by The Verge, based on a report published by the Basel Action Network.