Generative AI Could Generate Millions More Tons of E-Waste by 2030

Creative Intelligence is poised to worsen the E-Waste crisis

Who Saima S. Iqbal

Every time a creative artificial intelligence drafts an email or creates an image, the planet pays. Taking two pictures can consume as much energy as charging a smartphone; A single exchange with ChatGPT can heat up a server so much that it needs a bottle’s worth of water to cool it down. At scale, these costs add up. By 2027, the global AI sector could consume as much electricity as the Netherlands each year, a final calculation. And a new exam Nature Computational Science identifies another concern: AI’s massive contribution to the world’s growing e-waste pile. The study found that generative AI applications alone could add 1.2 million to five million metric tons to the planet by 2030, depending on the rate at which the industry grows.

This contribution would add to the tens of millions of tons of electronic products discarded by the world every year. Cell phones, microwave ovens, computers, and other ubiquitous digital products contain mercury, lead, or other toxins. When disposed of improperly, they can contaminate air, water, and land. The United Nations found that in 2022 about 78 percent of the world’s e-waste ended up in landfills or informal recycling sites where workers risk their health to remove the rare metals.

The worldwide AI boom is happening rapidly through physical data storage devices, as well as graphics processing units and other high-performance components required to process thousands of simultaneous calculations. This hardware lasts anywhere from two to five years, but is often replaced as soon as new versions become available. Asaf Tzachor, a sustainability researcher at Israel’s Reichman University who co-authored the new study, says his findings underscore the need to monitor and reduce the environmental impacts of this technology.

About supporting science journalism

If you like this article, please consider supporting our award-winning journalism subscribe. By purchasing a subscription, you’re helping to ensure a future of impactful stories about the discoveries and ideas that shape our world.

To estimate how much generative AI contributes to this problem, Tzachor and his colleagues looked at the type and volume of hardware used to run large language models, how long those components last, and the growth rate of the generative AI sector. The researchers caution that their prediction is a rough estimate that may change depending on a number of additional factors. More people may adopt creative AI than the authors anticipate, for example. Innovations in hardware design, on the other hand, can reduce e-waste in a given AI system, but other technological advances can make systems cheaper and more accessible to the public, increasing the number of uses.

The greatest value of this study comes from its focus on the broad environmental impacts of AI, says Shaolei Ren, a researcher at the University of California, Riverside, who studies responsible AI and was not involved in the new study. “Maybe we want these (creative AI) companies to slow down a little bit,” he says.

Few countries mandate proper disposal of e-waste, and those that do often fail to comply with existing laws. Twenty-five US states have e-waste management policies, but there is no federal law requiring electronics recycling. In February, Democratic Senator Ed Markey of Massachusetts introduced a bill that would require federal agencies to study and develop standards for AI’s environmental impacts, including e-waste. But that bill, the Artificial Intelligence Environmental Impacts Act of 2024 (which has not passed the Senate), would not compel AI developers to cooperate with its voluntary reporting system. Some companies, however, say they are taking independent action. Microsoft and Google have committed to reach zero net waste and zero net emissions by 2030; this would likely mean reducing or recycling AI-related e-waste.

Companies using AI have many opportunities to limit e-waste. It is possible to squeeze more life out of servers, for example through regular maintenance and updates or by switching to non-worn devices. Refurbishing and reusing obsolete hardware components can reduce waste by 42 percent, Tzachor and his co-authors report in the new study. And more efficient chip and algorithm designs can reduce the demand for hardware and electricity for creative AI. Combining all these strategies would reduce e-waste by 86 percent, according to the study’s authors.

There’s another wrinkle: AI products tend to be harder to recycle than standard electronics because the former contain a lot of sensitive customer data, says Kees Baldé, an e-waste researcher at the United Nations Institute for Training and Research, who was. Not with the new research. But big tech companies can delete that data and properly dispose of the electronics, he noted. “Yes, it costs something,” he says of the wider recycling of e-waste, “but the profits for society are much greater.”