Researchers are warning that generative AI could drive a massive increase in e-waste – up to five million tonnes per year by 2030 – worsening the global toxic trash crisis.
The explosive growth of generative artificial intelligence, which creates content like text, images, audio and synthetic data, is expected to add millions of metric tonnes of electronic waste annually by the end of the decade, a study in Nature Computational Science has said.
This rise in e-waste is due to the rapid expansion of AI applications and data centres, which demand frequent upgrades of high-performance computing hardware.
Short life cycles for advanced processors and storage equipment mean devices are replaced often to meet rising demand, resulting in a surge of discarded electronics.
If left unchecked, researchers warn that e-waste could spiral, further contributing to environmental pollution and resource depletion worldwide.
Resource-intensive
Generative AI models, such as large language models, are highly resource-intensive, requiring powerful servers, processors and storage solutions to operate effectively.
As big-tech companies race to develop more sophisticated models and hardware, e-waste from discarded equipment is piling up.
At the current adoption rate, e-waste from generative AI could reach between 1.2 and 5 million metric tonnes annually by 2030 – a thousand-fold increase over today’s levels.
Researchers estimate that this jump in waste is largely tied to applications like ChatGPT, which run on hardware with an expected lifespan of only two to five years.
AI-related e-waste often contains hazardous materials like lead, chromium and mercury, which pose severe health and environmental risks if not properly managed. Globally, just over 12 percent of e-waste is recycled.
Towards a circular economy
To address the rising tide of e-waste, researchers recommend moving towards a circular economy by extending hardware life, remanufacturing components and recycling materials from old devices.
Implementing these practices could reduce AI-related e-waste by up to 86 percent.
Asaf Tzachor, co-author of the Nature report, told the MIT Technology Review that extending the lifespan of technology by using equipment for longer is one of the most effective ways to reduce e-waste.
“Refurbishing and reusing components can also play a significant role, as can designing hardware in ways that make it easier to recycle and upgrade,” Tzachor said.
“For companies and manufacturers, taking responsibility for the environmental and social impacts of their products is crucial. This way, we can make sure that the technology we rely on doesn’t come at the expense of human and planetary health.”
This shift would require grassroots e-waste collection and recycling initiatives to keep valuable metals – like gold, copper, and silver – out of landfills. By refurbishing older devices and designing easily recyclable hardware, tech companies can help curb AI’s environmental impact.
Barriers to reducing e-waste
Reducing e-waste generated by artificial intelligence is not without its challenges.
Data security is a major barrier, as companies often destroy used devices to protect sensitive information. Secure data erasure technology could allow for safe reuse without compromising privacy.
Recycling also remains expensive due to the cost of safely handling hazardous materials, even though recycled metals hold significant economic value.
The Global E-Waste Monitor estimates that only 22 percent of electronic trash is formally recycled, with much of it ending up in informal recycling systems in lower-income countries, where safe processing methods are usually unavailable.
This looming crisis calls for sustainable AI development. According to the Nature report, as AI technologies advance, manufacturers and companies need to take responsibility for the social and environmental impacts of their products.
Researchers say industry standards for sustainable hardware use and cross-border cooperation in waste management will be key to addressing the issue.
