The AI Boom, Coal Plants, and the Air You Breathe

Photo by Johannes Heel on Unsplash

The artificial intelligence boom is reshaping the energy landscape in ways that are reverberating through air quality policy and public health. A Reuters investigation published in April 2026 details how surging electricity demand from data centers has contributed to the rollback of federal soot standards, keeping some of the nation's dirtiest coal plants online longer and hitting already-burdened communities hardest.

The story is centered on St. Louis but extends far beyond it. What's happening there is a case study in the tension between energy demand and the air that millions of people breathe every day.

50 Gigawatts of New Demand

The U.S. Department of Energy estimates that AI and data center growth will create 50 gigawatts of new electricity demand by 2030, a nearly 4% increase over the roughly 1,300 gigawatts produced by all U.S. power plants in 2025. Data center energy consumption has more than doubled in recent years, rising from 2% of national electricity use in 2018 to 4.4% by 2023, with projections reaching 6.7% to 12% by 2028.

That demand has to come from somewhere. While new solar, wind, and nuclear projects are in the pipeline, they take years to build. In the short term, the gap is being filled by fossil fuel plants that were expected to retire.

What Happened to the Soot Standards

In 2024, the EPA adopted tougher standards for fine particulate matter (PM2.5), the microscopic soot particles linked to heart disease, respiratory illness, stroke, and premature death. Those standards were set to take effect in 2027 and would have required the most polluting plants to either slash emissions or shut down.

In February 2026, those standards were scrapped before they took effect, part of a broader effort to ensure the grid can meet surging electricity demand. Reuters interviewed 20 air quality activists and health advocates who all identified the AI boom and the policies supporting it as the biggest potential threat to U.S. air quality.

St. Louis: Ground Zero

St. Louis ranks 475th in air quality among 501 U.S. metro areas. Residents breathe "good" air during only about one-third of the year.

A major driver is the Labadie Energy Center, a coal-fired plant about 40 miles west of the city. According to EPA data, Labadie produces the highest combined total of sulfur dioxide and nitrogen oxides among U.S. coal plants and emits soot at a rate two to three times that of nearly every other coal plant in the country. Reuters' analysis of the EPA's Co-Benefits Risk Assessment (COBRA) tool estimates the plant's pollution drives an economic burden of up to $5.5 billion per year, with about $820 million of that falling on St. Louis area residents.

The predominantly Black neighborhoods of North St. Louis already have some of the city's worst air quality. PM2.5 levels there regularly exceed federal safety limits, compounded by pollution from nearby highways, rail operations, and industrial sources.

The disparities are national. According to the NAACP, 78% of African Americans live within 30 miles of a coal-fired power plant, compared to 56% of non-Hispanic whites. A 2019 study published in Environmental Science & Technology found that soot pollution from power plants kills African Americans at a rate 25% higher than the national average.

The Broader Health Cost

The health toll is not confined to communities near coal plants. A University of California, Riverside study estimates that AI's electricity demands will cause up to 1,300 premature deaths annually in the U.S. by 2030 from particulate pollution generated when fossil fuel plants run longer and harder to meet data center load. Researchers project up to 600,000 asthma symptom cases per year, with a public health cost approaching $20 billion.

The problem extends beyond the grid. Data centers also rely on banks of on-site diesel backup generators. A Harvard-affiliated analysis of a single data center in Loudoun County, Virginia estimated $53 million to $99 million per year in health damages from its permitted emissions, driven by PM2.5 from eight natural gas turbines and 51 diesel generators. More than 2.5 million people live in areas where the study estimates measurable increases in fine particulate matter from that one facility alone.

Meanwhile, environmental nonprofits report that the rate of coal plant retirements has slowed sharply. At the rate seen in 2022, all U.S. coal plants would have been shut down by 2040. At the current pace, they could remain operational until 2065.

Why Local Air Quality Data Matters More Than Ever

When pollution sources are sustained or expanded, the need for granular, real-time air quality data increases. Regional monitoring networks were designed for a world where pollution sources were gradually declining. In a world where new demand is keeping old plants running and adding new on-site generation, the gaps in coverage become more consequential.

Fine particulate matter doesn't announce itself. PM2.5 is invisible at low concentrations, and levels can vary significantly over short distances depending on wind, terrain, and proximity to sources. A neighborhood a few miles from a power plant or data center may have very different air quality than one on the other side of town.

Community sensor networks fill those gaps. The PurpleAir map provides real-time PM2.5 readings from thousands of sensors worldwide, giving residents the hyperlocal data they need to make informed decisions: when to keep kids indoors, when to run filtration, and when to advocate for change with data in hand.

What You Can Do

• Know your air. Check the PurpleAir map regularly, especially if you live near industrial sources, highways, or data center developments.
• Monitor your neighborhood. An air quality sensor at your home or community site contributes data to the public network and creates a record of local conditions over time.
• Understand PM2.5. Fine particulate matter is the pollutant driving most of the health impacts described in this article. Our Particulate Matter 101 post explains what it is, where it comes from, and why it matters.
• Engage locally. When data centers, power plants, or industrial facilities are proposed in your area, air quality data from community sensors gives residents and advocacy groups a concrete, evidence-based voice in the process.

Further Reading

• How the AI boom derailed clean-air efforts in one of America's most polluted cities — Reuters, April 2026. The investigation that prompted this post.
• Analyzing air pollution health, economic risks from AI data centers — Harvard T.H. Chan School of Public Health. Research on the PM2.5 health costs of on-site data center power generation.
• The Hidden Public Health Cost of AI — Peer-reviewed study projecting 1,300 premature deaths and $20 billion in health costs from U.S. data center emissions by 2028.
• Datacenter boom keeps dirty coal plants alive in the US — The Register, April 2026. Research on slowing coal retirement rates driven by data center electricity demand.
• Global energy demands within the AI regulatory landscape — Brookings Institution. A broader look at AI's energy footprint across the globe.
• American Lung Association: State of the Air — Annual U.S. air quality report card.

Related Posts

• Particulate Matter 101 — What PM2.5 is, where it comes from, and why it's one of the EPA's six criteria pollutants.
• 12 Health Effects of Particulate Matter 2.5 — What PM2.5 does to the body, from short-term irritation to long-term disease.
• How to Reduce Exposure to Air Pollution and High Levels of PM2.5 — Practical steps for when air quality is poor.
• Air Quality Index Guide: What Should You Do in Each Level? — A walkthrough of all six AQI levels and what actions to take.