As AI revolutionizes industries and transforms everyday life, the infrastructure supporting its growth demands significant energy resources. However, this rapid technological expansion carries a hidden cost: the environmental and public health impacts of emissions from energy-intensive data centers. By 2030, the health consequences of this trend could include 600,000 asthma cases and 1,300 premature deaths annually in the United States.
AI technologies rely heavily on data centers, which are power-hungry facilities requiring vast amounts of energy to store and process information. Many of these centers in the US still depend on fossil fuels, a major source of harmful air pollutants like fine particulate matter.
Researchers Shaolei Ren (University of California, Riverside) and Adam Wierman (California Institute of Technology) revealed alarming projections about this issue. “Public health impacts are direct and tangible impacts on people, and these impacts are substantial and not limited to a small radius of where data centers operate,” Ren explains. Pollution from these facilities is not contained locally; it travels across states, affecting people nationwide.
To illustrate the severity, training a single large AI model generates emissions equivalent to over 10,000 car roundtrips between Los Angeles and New York City. By 2030, data center emissions could lead to over $20 billion in public health costs, double the cost attributed to the US steel industry and comparable to emissions from millions of vehicles in California.
Communities near data centers are already feeling the impact. Virginia’s Data Center Alley serves as a stark example, where gas-powered backup generators contribute to 14,000 asthma symptom cases annually. The public health cost of these emissions ranges between $220 million and $300 million per year at current operating levels, potentially surging to $2–$3 billion if emissions approach state-authorized limits.
Even states far from data centers, such as Florida, are affected by the dispersal of airborne pollutants.
Despite the concerning outlook, some tech companies are taking proactive measures. Investments in renewable energy projects and nuclear power are helping offset emissions. However, many data centers continue to rely on fossil fuels, which could increase the US demand for natural gas to levels comparable to the energy consumption of states like New York or California by 2030.
This study, published in arXiv, is among the first to quantify the public health impacts of data center emissions in financial terms. “The question around the health impacts of artificial intelligence and data center computing is an important one,” Benjamin Lee, a researcher at the University of Pennsylvania noted.
