Study: AI Data Centers Projected to Cause 1,300 Premature Deaths and $21.5B in US Health Costs Annually by 2030
Summary
- • AI data centers could cause up to 1,300 premature US deaths per year by 2030 through fossil-fuel-driven air pollution.
- • Associated healthcare costs are projected at up to $21.5B annually — roughly double the burden of the entire US steel industry.
- • Virginia's massive data center hub already exports fine-particulate pollution across seven neighboring states.
- • Researchers urge regulators to require air-quality disclosures, noting tech sustainability reports completely omit these pollutants.
Details
Projected Annual Deaths by 2030
Using EPA statistical models, researchers project AI-driven air pollution could cause up to 1,300 premature US deaths annually by 2030, driven by surging data center electricity demand.
Healthcare Cost Projection
Associated healthcare costs covering cancer, asthma, chronic illness, and lost productivity are projected at up to $21.5 billion per year — roughly double the public health burden of the entire US steelmaking industry.
Virginia Regional Health Costs
Northern Virginia, the world's largest data center hub, already causes an estimated $190M–$260M in annual regional health costs; running backup generators at maximum permitted capacity could spike that figure tenfold to $2.6B/year.
Transboundary Pollution Spread
Pollution from Northern Virginia data center backup generators regularly drifts into Maryland, West Virginia, Pennsylvania, New York, New Jersey, Delaware, and Washington D.C.
Single Model Training Footprint
Electricity consumed to train Meta's Llama-3.1 (launched July 2024) generated air pollution equivalent to more than 10,000 round-trip car journeys between Los Angeles and New York.
Energy Consumption Forecast
McKinsey projects data centers will consume 11.7% of total US electricity by 2030, up from 3.7% in 2023, with AI workloads as the primary driver — making data centers the fastest-growing US energy consumer.
Pollution Source Mechanism
Data centers do not emit pollution directly but draw large electricity volumes from grids still powered by coal and natural gas; backup diesel generators during grid spikes add fine particulates and nitrogen oxides.
Corporate Disclosure Gap
Tech companies routinely publish sustainability reports on carbon and water metrics but almost completely omit harmful air pollutants. Researchers call for standardized regulatory disclosure requirements for air-quality impacts.
Stat = quantified data point; Research = study methodology or finding; Policy = regulatory/disclosure issue
What This Means
A new study quantifies a largely invisible public health cost of the AI infrastructure boom: the fossil fuels powering data centers and their backup generators are projected to kill up to 1,300 Americans per year by 2030 and impose $21.5 billion in annual healthcare costs — a burden that currently appears nowhere in corporate sustainability disclosures. Because tech companies report only on carbon and water, communities near data centers lack reliable information about local air quality, creating a significant accountability gap. The research strengthens the case for regulators to mandate air-quality reporting as a standard sustainability metric, and could fuel pushback in communities near planned data center expansions. It also adds urgency to the broader debate about AI's energy footprint at a moment when data center electricity demand is tripling as a share of the US grid.