The rapid expansion of AI data centers poses an emerging public health disaster, primarily driven by massive hydrocarbon and fossil-fuel energy demands. Emissions from supporting power plants and continuous diesel backup generators release harmful fine particulate matter (PM2.5) and, causing an hundreds of thousands of annual premature deaths and many billions of dollars in public health damages in the U.S.

The public health threats caused by the AI data center boom affect local and regional communities through several specific hazards: 

• Air Pollution and Particulates: Burning fossil fuels to power servers and running banks of emergency diesel generators create acute spikes in toxic emissions. Academic models project that data center-linked pollution could contribute to 600,000 asthma symptom cases by 2030.
• Targeting Vulnerable Communities: Large tech firms increasingly build these hyperscale facilities in economically disadvantaged areas or regions already burdened by high industrial pollution. These “sacrifice zones” exacerbate pre-existing health disparities and disproportionately affect communities of color.
• Severe Noise Pollution: Continuous operation of massive cooling fans and mechanical equipment emits low-frequency hums that often exceed the EPA’s recommended limits of 55 decibels near residential areas. This constant noise disrupts sleep and is linked to chronic stress, cognitive impairment, and cardiovascular risks.
• Water Scarcity and Contamination: Massive liquid-cooling systems consume millions of gallons of water daily, straining local drinking supplies. Furthermore, the chemical agents utilized in server fire suppression systems (often per- and polyfluoroalkyl substances, or PFAS) pose ongoing chemical safety risks to local water tables.

The rapid expansion of AI data centers is triggering a quiet but severe public health crisis, costing the U.S. economy $25 billion annually in environmental and health damages. These massive facilities require an unprecedented amount of power and water. Because their immense energy demands heavily outpace the integration of clean energy, data centers are driving a resurgence of fossil-fuel combustion. The resulting regional air pollution, toxic emergency emissions, and water depletion are creating severe, localized health consequences that have community advocates and researchers sounding the alarm.

The primary pathways through which AI infrastructure poses a public health risk include:

Air Pollution and Respiratory Illness

To stay online during power grid failures or peak load periods, data centers deploy large arrays of massive diesel backup generators

• Toxic Diesel Particulate Matter: These generators emit diesel soot, which is a known carcinogen.
• Extreme NOx Output: Data center diesel backups can release 200 to 600 times more nitrogen oxides (NOx) than a natural gas plant to generate the same amount of energy.
• Asthma and Mortality: Projections suggest that by 2030, U.S. data center air emissions could cause 600,000 cases of asthma symptoms and 1,300 premature deaths annually, accounting for over one-third of all American asthma deaths.
• On-Site Power Generation: High-profile AI projects are opting to build their own local fossil-fuel plants. For example, the Harvard T.H. Chan School of Public Health highlighted that the xAI Colossus 2 project in Mississippi plans to utilize 41 on-site gas turbines, threatening to blanket the region in tens of millions of dollars of annual health damages.