The growing popularity of generative AI is expected to lead to rapid growth in e-waste, according to a study published in Nature Computational Science.

The researchers behind the study calculated that e-waste could reach a total of 1.2 million to 5.0 million tons by 2030, or about 1,000 times more e-waste than was produced in 2023 .

“We found that e-waste generated by generative AI, particularly large language models, could increase significantly, potentially reaching up to 2.5 million tonnes per year by 2030 if no waste reduction measures are taken. waste is not implemented,” said Asaf Tzachor, a sustainable development expert. development at Reichman University in Israel and co-author of the study.

The study also suggests solutions to reduce e-waste: strategies to extend, reuse and recycle generative AI hardware could reduce the creation of e-waste by 16% to 86%, they estimate.

“This represents a tremendous opportunity to reduce the waste stream if these practices are widely adopted. It is clear from this study that the nature of the e-waste crisis is global, which is why it is important to focus on cross-border e-waste management. ” said Saurabh Gupta, founder of Earth5R, a sustainability organization based in India. Gupta was not involved in the study.

What is electronic waste?
Whenever we throw away an “obsolete” or broken electronic device, it is considered electronic waste. This can include computers, smartphones, chargers and cables, electronic toys, cars and larger server systems.

E-waste accounts for 70% of the total toxic waste produced worldwide each year, but only 12.5% ​​of e-waste is recycled. This live counter from The World Counts shows how quickly e-waste is increasing.

“Reducing e-waste is important because improper disposal results in the release of hazardous materials, like lead and mercury, that harm ecosystems and human health,” Gupta told DW by email.

Researchers in the study published October 28, 2024 focused on e-waste produced from generative AI algorithms – types of AI that generate text, images, videos or music from massive data sets.

It is clear from previous research that AI has high energy requirements: calculations from research firm SemiAnalysis suggest that AI could enable data centers to use 4.5% of global energy production from here 2030.

But Tzachor said it was less clear how much e-waste generated by generative AI programs, such as ChatGPT. This includes all IT resources needed for training and using AI in data centers.

And because generative AI depends on rapid improvements in hardware infrastructure and chip technologies, there is some evidence that it leads to more e-waste as hardware is updated or replaced.

“It is much easier and more cost-effective to address the e-waste challenges posed by AI now, before they spiral out of control,” Tzachor said.

How did researchers calculate the growth of AI e-waste?
Researchers created a model to quantify the scale of e-waste from data centers that support the use of generative AI models, such as large language models.

They found that e-waste could reach 5 million tonnes per year in a scenario with high AI growth.

But their estimates of AI-related e-waste were potentially low, Tzachor said, due to the rapidly changing AI business landscape.

“Factors such as geopolitical restrictions on semiconductor imports and rapid server turnover may intensify the generation of e-waste associated with generative AI,” Tzachor told DW by email.

Additionally, the study only included e-waste created by generative AI systems, particularly large language models, and not other forms of AI.

“E-waste from the broader AI ecosystem is significant. The study predicts that this figure will increase with increasing adoption of AI, creating a combined environmental challenge due to multiple forms of AI,” Gupta said.

Reducing e-waste requires global strategies
The study estimates that implementing circular economy strategies could reduce e-waste production by 16%, or even up to 86%.

Circular economy strategies aim to minimize waste and increase the efficiency of IT hardware.

Tzachor said the strategy had three main objectives:

• Extend the use of existing equipment to delay the need for new equipment
• Reuse and Refurbish Components
• Extract valuable materials when recycling equipment

Gupta said he strongly agreed with the study’s findings.

“The reduction range of 16 to 86 percent reflects the immense potential of these strategies, especially if they are supported by policies and when widely implemented across sectors and regions,” Gupta said.

Gupta’s organization, Earth5R, has demonstrated how effective circular economy strategies can be, he said.

“Through our local programs and business partnerships, we are already encouraging local e-waste collection and recycling efforts that help businesses and consumers manage their electronic devices sustainably,” Gupta said.

He stressed that e-waste was a global crisis that required equitable, cross-border management strategies to mitigate the “environmental and health damage” caused when high-income countries export their e-waste to low-income regions.