Artificial intelligence infrastructure could generate enough electronic waste by 2050 to fill a line of shipping containers stretching around Earth six times. This finding comes from a report by the Basel Action Network arguing that existing estimates drastically understate the problem.
Previous estimates focused chiefly on servers and accelerators like GPUs. These components account for just 13 percent of a datacenter's equipment by weight. Once power, networking, cooling, and other infrastructure are included, the total could be 40 to 60 times higher than the most widely cited academic projections.
The report identifies five equipment categories. These comprise networking, power distribution, backup power systems, servers plus accelerators, and cooling. Together, they add up to about 7,000 metric tons for a reference 100 megawatt AI bit barn. Much of this hardware will require replacement during facility upgrades.
The non-profit organization Basel Action Network takes its name from the Basel Convention. That international treaty controls movements of hazardous waste and seeks to prevent its transfer from developed to developing countries. The new 46-page paper is titled How Big Is the AI Waste Wave?
The model starts with industry buildout plans and calculates potential waste. It estimates that AI infrastructure will generate between 395 million and 617 million tonnes of e-waste from 2025 to 2050. That volume would fill between 15 million and 23 million shipping containers.
Placing 20 million of those containers end to end creates a line about 244,000 kilometers long. That distance equals roughly six times Earth's circumference. The estimate assumes copper, steel, cooling and power distribution equipment, networking gear, and computing hardware will be decommissioned within years rather than decades due to rapid technological obsolescence.
Facility upgrades drive much of this turnover. When a facility replaces racks drawing between 5 and 15 kilowatts with the 50 to 140 kilowatt racks required for AI workloads, much of its power infrastructure must be ripped out. A single 100 megawatt facility contains about 2,700 tonnes of copper in its cabling and busbars alone.
GPUs and other AI accelerators are replaced every two to three years. Traditional general-purpose servers last five to seven years. Owners typically replace entire servers when upgrading to the latest Nvidia GPUs because original equipment manufacturers ship complete, preconfigured systems.
Other equipment operates on different replacement cycles. Networking infrastructure lasts three to four years. Power distribution lasts eight years. Backup power and cooling systems both operate on five-year cycles.
No hyperscaler, government, or international body has published a plan for handling AI-driven e-waste on this scale. Existing infrastructure cannot safely process current volumes. Jim Puckett, founder and chief of strategic direction at the Basel Action Network, noted that the environmental debate around AI has focused on electricity, carbon, and water while overlooking the hardware.
This publication is the first in a four-part series. The second report will examine reuse, refurbishment, and repurposing. The third will consider potential toxicity including PFAS contamination, and the fourth will explore risk mitigation.
