AI's footprint isn't just "in the cloud" — it's in our air.
Data centers contribute to air pollution through two primary pathways.
First, they demand enormous amounts of electricity, much of which still comes from burning coal and natural gas. This reliance on fossil fuels means every data center indirectly pumps fine particulate matter (PM2.5), nitrogen oxides (NOx), and greenhouse gases into the atmosphere.
Second, on-site backup diesel generators, permitted to operate for hundreds of hours per year, emit these same pollutants directly into surrounding communities.
The scale is massive. Training a single large AI model produces as much carbon as five cars over their entire lifetimes. By 2030, air pollution from U.S. data centers is projected to cause $20 billion in annual health costs. Some communities have already seen a 48% increase in greenhouse gas emissions since 2023 due to data center operations.
The pollutants these facilities release are small enough to penetrate deep into the lungs and enter the bloodstream, triggering asthma attacks, cardiovascular disease, and premature death. In Northern Virginia, the densest data center hub in the world, backup generators can run up to 50 hours at a time, contributing an estimated $200–$300 million annually in public health costs Virginia alone has approved 8,910 generators, nearly all of them highly polluting Tier 2 models.
The burden falls disproportionately on Black, brown, and low-income communities. In places like Bessemer, Alabama, data centers are sited near communities already facing severe environmental injustices. Low-income Black Americans experience the highest mortality rate from fine particulate matter air pollution, and the data center boom is deepening that gap.
Resources/ Sources
- Dr. Shaolei Ren (YouTube): Public health costs of AI data centers. Projected health burdens of $10-20B by 2028 due to air pollution, comparable to vehicle emissions.
- Historical red-lining is associated with fossil fuel power plant siting and present-day inequalities in air pollutant emissions
- Dr. Francesca Dominici: Research on air pollution and public health impacts.
- Karen Hao (MIT Technology Review): Training a single AI model can emit as much carbon as five cars in their lifetimes. This seminal article quantifies the environmental cost of natural language processing, revealing that the energy-intensive process of training large-scale deep learning models produces staggering amounts of greenhouse gas emissions.
- Molly Elise Bush: Construction and Consequences: The Human Impacts of Artificial Intelligence Data Centers. A human-rights-focused perspective on the physical and social consequences of data center development, highlighting the vulnerability of communities during long-term construction cycles.
- Harvard Business Review: Mitigating the Public Health Impacts of AI Data Centers, 2025. Analysis of public health costs from data center air pollution including PM2.5 and NOx emissions.
- Lincoln Institute of Land Policy: Land & Water Impacts of Data Centers. Examines land and environmental health impacts of data center siting.
- NAACP: Did You Know: AI and Environmental Justice. Fact sheet on AI infrastructure, pollution, and environmental justice impacts.
- EPA Title V Air Permits: Regulatory framework governing major sources of air pollution, including data centers with large backup generators.
- Virginia DEQ: Issued Air Permits for Data Centers. Public database of permitted emissions and operating conditions for data centers in Virginia.
- EPA Clean Air Act Resources: Federal air quality standards, permitting guidance, and compliance tools for data centers.
- Julie Bolthouse: The Local Impact of Data Centers. Discussion on community-level environmental and health impacts of data center siting.
- Nature Forward: Unpaid Toll: The Health Impacts of Data Centers.
- Alex de Vries-Gao: The carbon and water footprints of data centers and what this could mean for artificial intelligence. Research examining environmental and public health implications of AI-driven data center expansion.
- Regulatory Capture & Loopholes: Analysis of gaps in environmental regulation and enforcement affecting large infrastructure projects, including data centers.
- National Data Centers Tracker: Tool mapping existing and proposed data centers across the U.S.
- Move Past Plastic: Center Toxins Infographic. Visual overview of toxic exposures and environmental health risks linked to data center infrastructure.
- David Segal / NYT: A Data Center Wrapped in a Mystery Comes to the New Mexican Desert. Investigative reporting on opaque data center development and community concerns.
- Gateway 6 — Construction & Operational Permits: Permit monitoring and enforcement strategy for data center construction and operations.