Skip to content
context Hardware

Study: e-waste from AI infrastructure could reach 395 to 617 million tonnes by 2050

only one source so far

Basel Action Network estimates that electronic waste associated with AI infrastructure will reach a cumulative 395 to 617 million tonnes between 2025 and 2050, representing 15 to 20 percent of global e-waste by 2050.

The nonprofit organization Basel Action Network (BAN) published a study estimating that electronic waste associated with the construction and operation of AI infrastructure will reach a cumulative 395 to 617 million tonnes between 2025 and 2050, equivalent to roughly 8.6 to 13.1 million tonnes of discarded equipment annually. According to BAN, this volume is expected to account for 15 to 20 percent of all global e-waste by 2050, when the total amount of electronic waste worldwide is expected to rise to as much as 211 million tonnes annually. The estimate assumes a rate of 70 000 tonnes of waste per gigawatt of data center capacity and is based on a projection by McKinsey that total data center capacity will reach up to 219 GW by 2030.

Unlike earlier studies, which focused only on servers and accelerators, BAN also included power supply and distribution systems, cooling, backup power supplies, networking equipment and so-called "AI Waste Contagion" in its calculations — meaning telecommunications infrastructure and personal devices that, according to the organization, will become obsolete and be replaced sooner as a result of AI development. According to BAN, the narrower scope of earlier estimates overlooked about 87 percent of the electromechanical infrastructure in data centers. The founder of the organization, Jim Puckett, said in a press release that unless companies and governments begin planning for this waste, the current expansion of AI could become an even greater toxic crisis than the one the world already faces today.

The study notes that of the 68.3 million tonnes of e-waste the world produces annually, less than a quarter is formally collected and recycled. The rest often ends up in informal processing, where burning or burying equipment exposes workers and the surrounding environment to toxic substances such as lead and chromium; according to the World Health Organization (WHO), millions of children working or living near such operations also face this exposure. The United States, which has the most data centers in the world, has not yet ratified the Basel Convention restricting international trade in hazardous waste, and according to earlier investigations, American recycling companies continue to export e-waste abroad.

Earlier estimates were significantly lower: a study from 2024 predicted 1.2 to 5 million tonnes of AI e-waste by 2030, while a study from February 2025 estimated 131 000 to 225 000 tonnes annually from AI servers alone by the end of the decade — an amount comparable to the total e-waste of a country the size of Denmark.

What changed

Why it matters

The study shows that the actual waste impact of AI infrastructure could be orders of magnitude greater than previous estimates focused only on servers and GPU suggested, because they did not include cooling, power supply, networks and the associated obsolescence of devices. This matters for companies operating or planning data centers, as well as for regulators, who, according to BAN, must start accounting for this volume of waste now, rather than years later. For the general public, this is currently more of an environmental warning than a change requiring specific personal action.

Relevant practical impact

What this means

01

For a business

According to the study, companies building or operating AI data centers face a growing risk of future hardware disposal and recycling costs, as well as regulatory scrutiny, because planning to date has accounted only for servers, not the entire infrastructure.

Risks and compliance
What to decide When planning AI infrastructure, account for the life cycle and recycling of associated equipment (cooling, power supply, networks), not just servers and accelerators.
More business impacts →
AI datová centra Basel Action Network ekologická regulace e-waste hardware infrastruktura recyklace

Check the original

Event sources

only one source so far · 1 publisher, 1 independent. We count feeds from the same owner only once.

1
The Verge AI independent context · first detected The AI data center e-waste problem is huge — and getting bigger