Building AI data centers could unleash a new wave of e-waste. By 2030, AI-related technology could generate 8.6 to 13.1 million tons of waste each year. Much of this could be computers, smartphones, and other household and office equipment that users and companies are more likely to replace.
Building AI data centers could unleash a new wave of e-waste. By 2030, AI-related technology could generate 8.6 to 13.1 million tons of waste each year. Much of this could be computers, smartphones, and other household and office equipment that users and companies are more likely to replace.
This forecast was published by the environmental organization Basel Action Network (BAN).
The reason is the rapid pace of technology change. AI data centers require servers, accelerators, networking equipment, cooling and power systems, some of which will need to be replaced within a few years. The report also predicts that the increasing demands of AI will make conventional computers, smartphones and telecommunications equipment more obsolete. For example, new AI features are likely to prompt companies and consumers to upgrade their electronics.
This is where a significant part of the high-profile figure of 13.1 million tons comes from: in 2030, the data center infrastructure itself, according to the BAN model, will generate about 2.8 million tons of decommissioned equipment, and the accelerated replacement of other equipment will generate another 5.8–10.3 million tons.
This e-waste can contain lead, cadmium, mercury and other hazardous substances, and only about a fifth of this waste is currently properly recycled or disposed of worldwide, BAN notes.
Previously, dev.ua wrote that, according to another estimate, generative AI could create up to 2.5 million tons of e-waste annually by 2030. That study focused mainly on server hardware.
