Study reveals data centers’ long-distance damage to air quality
Data centers can affect health far beyond their sites because they draw electricity from an interconnected grid. Meeting that demand can increase emissions at power plants elsewhere, so the pollution footprint does not stop at a facility’s fence. The study links those extra emissions to premature deaths through air pollution. The researchers modeled which generators responded to data center demand hour by hour, and how electricity moved between regions. They then modeled how emissions of nitrogen oxides and sulfur dioxide formed pollution and traveled through the atmosphere. In parts of the Midwest, emissions combine with conditions that favor fine particulate matter, or PM2.5. The study estimates that data center electricity demand caused about 650 premature deaths in the United States in 2023: 513 linked to PM2.5 and 137 to smog. The authors say impacts are widespread, not limited to nearby neighbors. Their findings highlight why the location and operation of data centers matter for public health across a broader region.
How can data centers cause air pollution and health impacts in places far from the facilities themselves?
Data centers can affect health far beyond their sites because they draw electricity from an interconnected grid. Meeting that demand can increase emissions at power plants elsewhere, so the pollution footprint does not stop at a facility’s fence. The study links those extra emissions to premature deaths through air pollution.
The researchers modeled which generators responded to data center demand hour by hour, and how electricity moved between regions. They then modeled how emissions of nitrogen oxides and sulfur dioxide formed pollution and traveled through the atmosphere. In parts of the Midwest, emissions combine with conditions that favor fine particulate matter, or PM2.5.
The study estimates that data center electricity demand caused about 650 premature deaths in the United States in 2023: 513 linked to PM2.5 and 137 to smog. The authors say impacts are widespread, not limited to nearby neighbors. Their findings highlight why the location and operation of data centers matter for public health across a broader region.
How many premature deaths did the study estimate were linked to data center electricity use in 2023?
The researchers estimate that air pollution from electricity used by data centers caused about 650 premature deaths across the United States in 2023. The figure matters because it connects the electricity needs of data centers to health effects that can occur well beyond the places where the facilities stand.
The study divides the estimated deaths by pollution type: 513 were linked to fine particulate matter, or PM2.5, and 137 to smog. The researchers traced the emissions by modeling which power generators responded to data center demand, then used a chemical transport model to estimate how pollution formed and moved. They translated pollution into mortality rates using epidemiological evidence.
The 650 deaths represent about 5.3 percent of the nation’s total electricity-related mortality, according to the study. The estimate is for 2023, when data centers represented about 4 percent of U.S. energy demand. The researchers note that this share has likely increased as data centers continue to be built rapidly, making regional health impacts an important consideration.
Why did some Midwestern states have higher estimated death rates even though much of the data center demand was in Virginia and Texas?
The study found that estimated premature-death rates were highest in Iowa, Nevada, South Dakota, and Kansas. That may seem surprising because much of the data center electricity demand was concentrated in Virginia and Texas. But the location of a data center does not necessarily match the location of the power plants that respond to its electricity use or the communities exposed to resulting pollution.
The researchers modeled how electricity moves through the interconnected grid and which generators increase output in response to demand. They also modeled how atmospheric conditions affect pollution formation. In parts of the Midwest, power plant emissions meet conditions that favor the formation of PM2.5, which can harm health.
The study’s approach therefore traces impacts across regions instead of assigning them only to places near data centers. It estimates about 650 premature deaths nationwide in 2023 from data center-related pollution. The researchers argue that planning should account for these wider regional health impacts, including how facility location and operations affect public health beyond the immediate neighborhood.
What is PM2.5, and why can breathing it harm people?
PM2.5 is fine particulate matter, a type of air pollution that can harm health. The article identifies it as a pollutant that can cause heart attacks, stroke, and lung disease. In standard usage, PM2.5 refers to particles no larger than 2.5 micrometers; that size definition is general scientific context, not a detail provided in the article.
The study connects data center electricity use to PM2.5 through power plant emissions. Nitrogen oxides, or NOx, and sulfur dioxide, or SO2, are released by generators. The researchers’ chemical transport model simulates how these emissions contribute to PM2.5 formation and how pollution travels across regions. The article notes that atmospheric conditions in parts of the Midwest favor this formation.
The researchers estimate that 513 of the roughly 650 premature deaths linked to data center electricity demand in 2023 were from PM2.5. The estimate draws on epidemiological literature connecting additional air pollution to mortality. Because pollution can form away from facilities, the authors emphasize that health effects may reach people who do not live near a data center.
How can extra electricity demand in one region cause power plants in another region to burn more fuel?
Electricity demand in one place can affect power plants elsewhere because the U.S. grid is interconnected. Electricity can be transmitted between regions, so the generators that supply an additional unit of demand do not have to be located beside the customer. This matters because increased generation can bring additional emissions and health impacts to other areas.
The MIT researchers built a model that simulates how much electricity each generator produces each hour and how power moves between regions. It identifies which generators respond to data center demand. The article notes that generators responding to extra demand are often natural gas plants, although the response varies by place and hour.
Using this approach, the team traced emissions of nitrogen oxides and sulfur dioxide to power plants across the country, then modeled how pollution forms and travels. Their estimate of about 650 premature deaths in 2023 shows why local electricity use can have broader consequences. The authors say assessments should account for grid-wide and regional effects, not only facilities’ immediate neighbors.
How do gases such as nitrogen oxides and sulfur dioxide turn into pollution that can travel downwind?
Power plants emit nitrogen oxides, called NOx, and sulfur dioxide, called SO2. These gases can contribute to pollution that forms in the atmosphere rather than only at the smokestack. The article states that NOx contributes to ground-level ozone, or smog, and that both NOx and SO2 contribute to fine particulate matter, PM2.5.
The researchers used a chemical transport model to simulate how pollutants form and travel through the atmosphere across regions. Atmospheric conditions matter: the study notes that conditions in parts of the Midwest favor PM2.5 formation when combined with emissions. The article does not detail the specific chemical reactions, but standard atmospheric science explains that emitted gases can undergo reactions that produce ozone and particles.
This process helps explain why electricity demand in one place can affect air quality elsewhere. The study estimates 137 premature deaths from smog and 513 from PM2.5 in 2023, linked to data center electricity demand. The researchers say regional pollution and health effects should be considered when assessing data center impacts.
Why do data centers, especially those running AI computing, need so much electricity around the clock?
Data centers need electricity to run computing tasks, and the researchers assume that facilities draw power 24 hours a day, seven days a week. That constant demand matters because power plants may need to respond at different hours, producing emissions that vary by location and time. The article focuses on the effects of this electricity use, rather than detailing how much power particular AI tasks require.
As general context beyond the article, AI systems can require substantial computing to train models and to answer user requests. Those computing operations run on data center equipment, which must be powered and supported by the facility. The study itself does not separate AI electricity use from other data center computing tasks or quantify the energy use of particular AI systems.
The article reports that data centers accounted for about 4 percent of U.S. energy demand in 2023, and says that share has likely increased as facilities continue to be built rapidly. Its key finding is that electricity demand can cause distant emissions and health impacts. The authors call for attention to the broader public health effects of where data centers are located and how they operate.
This brief was written by AI from the original reporting and checked by other models. Names, figures and quotes come from the source; read it for full context.
Read more in the JupiteX app
Pulse is free. New stories every 4 hours, each one broken into the questions that explain it.
Or read more news on the web