Key facts
- AI datacenters will create a "tsunami" of discarded electronics, tripling global e-waste in 25 years, according to BAN.
- The e-waste generated could fill shipping containers enough to encircle the globe six times.
- E-waste typically contains toxic materials like lead, mercury, cadmium, and forever chemicals.
- The report estimates 4,871 new datacenters are in the US construction pipeline.
- AI datacenter equipment is projected to be replaced every two to five years.
- AI developments may force rapid replacement of consumer electronics.
The rapid expansion of datacenters to support artificial intelligence is projected to create a significant surge in electronic waste, according to a new report by the Basel Action Network (BAN). The report warns that this could triple the world's annual generation of toxic e-waste within the next 25 years. The volume of discarded electronics could fill enough shipping containers to circle the globe six times.
BAN highlights that current e-waste management is inadequate, with only one-fifth of electronic trash being properly handled. This waste often contains hazardous substances like lead, mercury, cadmium, and persistent chemicals, alongside valuable metals that are difficult to reclaim. Experts agree that the US faces substantial challenges in managing its e-waste, with much of it ending up in landfills or being improperly disposed of overseas.
The report notes that 4,871 new datacenters are planned in the US, many of which will be dedicated to AI. It estimates that AI-driven equipment, including servers weighing 1,360 kgs each and associated networking and cooling infrastructure, will need to be replaced every two to five years. Furthermore, advancements in AI are expected to accelerate the replacement cycle for consumer electronics.
McKinsey & Company estimates that the tech industry will require nearly $7 trillion in capital outlays globally over the next five years to build datacenters for AI. BAN contrasts this spending with the cost of ending world hunger for 75 years, noting the significant financial commitment to AI infrastructure. The report also points out that individuals are often unwitting participants in this AI buildout, using AI services without full awareness of the environmental consequences.