AI boom will produce millions of tons of e-waste and toxic materials by 2030: report

The AI ​​boom isn’t just consuming huge amounts of energy and water: This is also creating an unprecedented tsunami of electronic waste.

According to Stanford Universityprivate investment in AI has grown from $3 billion in 2022 to $25 billion last year, with companies adopting AI tools. faster than ever. This surge is forcing data centers to constantly upgrade their equipment, discarding equipment that is still running in the race to remain competitive.

The massive use of components to power the hardware that runs AI models results in millions of tons of discarded electronic components. New research published in Nature A team of researchers from China, Israel and the UK estimates that large language models (LLMs) such as ChatGPT, Claude or LlaMa alone could generate 2.75 million tonnes (2.5 million tonnes) of e-waste annually, seriously increasing the impact on the environment. AI.

“In an optimistic scenario where LLMs become ubiquitous (i.e. everyone uses them daily, for example on social media), the results indicate that the EoS e-waste flow from designated data centers would rise to approximately 16 million tonnes (Mt) per year. over the decade from 2020 to 2030,” the study says.

The waste stream is growing at an alarming rate, with a compound annual growth rate of 110%, significantly outpacing the 2.8% growth of conventional e-waste such as screens and washing machines.

The geography of this crisis is very concentrated. According to a study by the Chinese Academy of Sciences and Reichman University, North America leads with 58% of AI-related e-waste, followed by East Asia with 25% and Europe with 14%.

In addition to the huge amount of e-waste, the AI ​​industry as a whole consumes a huge amount of resources. Last year, Decipher reported that for every 4 requests ChatGPT consumes half a liter of water. Consider that more than 220 million people visit the site every month, and you can do the math and understand why cities near AI data centers have seen water costs nearly double in less than a decade.

The study estimates that by 2030, this e-waste will contain nearly a million tons of lead, 6,000 tons of barium, and significant amounts of cadmium, antimony, and mercury, adding significant amounts of toxic elements to the environment—all at a healthy rate. -documented risks to soil, water and public health.

The researchers don’t say whether companies and governments are doing enough, but there is a financial aspect that could be helpful. Metals such as gold, silver and platinum, which are used on these discarded servers, also represent significant financial potential if they are restored. The study estimates that properly recycling these metals could add $70 billion to the economy, spurring the development of e-waste recycling infrastructure.

The study also explains that countries without access to the latest chips may produce 14% more e-waste as they are forced to use less efficient equipment.

But there are some solutions that can help solve this problem. The researchers claim that extending the life of servers through improved maintenance could reduce e-waste by 58%, and reusing certain components would further reduce waste by 21%.

Additionally, legacy AI servers could be repurposed for lighter tasks such as educational projects or basic web hosting rather than thrown away, diverting them from waste streams and maximizing their usefulness.

This is becoming a priority for environmental groups around the world. Energy analyst Alex de Vries, founder of Digiconomist, said Decipher It is important to work on solutions before the negative impacts of the AI ​​industry become too difficult to control.

“Right now the numbers are small, so you might argue, ‘Why do we need to put this issue at the top of our agenda if it’s still small?’ – said de Vries. “But this thing won’t stay small for very long.”

Edited by Andrew Hayward