06/05/2026
Researchers from Caltech and UC Riverside estimate that by 2030, the air pollution linked to powering AI data centers could contribute to roughly 1,300 premature deaths each year in the United States, along with billions of dollars in public health costs.
The pollution isn't coming mainly from the data centers themselves. It comes from the power plants generating the electricity they need and from the diesel backup generators many facilities rely on.
The study also found that the health impacts are not distributed evenly. Lower-income communities are expected to bear a disproportionate share of the burden, raising concerns about who benefits from the AI boom and who pays the environmental and health costs.
These are known as 'externalities' of a technology or business, meaning, the damage they cause is passed on to humans or nature to deal with and pay for. If these externalities were paid for by the businesses themselves, they'd go broke. We see this sort of externality in many major industries that get away with large scale pollution and public health declines as they build and seel their product.
