Skip to content

Do data centers harm local air quality?

Grid electricity creates a regional footprint. Diesel engines create intermittent local emissions and a real exposure pathway. Separate annual averages, permit limits, actual operation, dispersion, and cumulative neighborhood sources.

Key facts, with scope and limits.

Location-based footprint61 billion kg CO2e
Metric
Modeled greenhouse-gas emissions associated with data-center electricity
Scope
United States; balancing-authority generation mix
Period
2023
Status
modeled
Regional range243–1,249 lb / MWh
Metric
Annual total-output CO2e rate among selected eGRID subregions
Scope
Selected U.S. eGRID subregions; annual average, not marginal dispatch
Period
Generation year 2023
Status
administrative

Electricity emissions vary by region

Annual average electricity emissions vary widely by eGRID subregion. These values reflect the regional generation mix. They cannot predict the marginal plant or long-run resources serving a new facility.

EPA eGRID2023 Revision 2 total-output CO2e rates, pounds per megawatt-hour. Contractual renewable claims and physical location-based accounting are separate views.

Electricity emissions vary by region data
RegionValue (lb CO2e/MWh)
Upstate New York242.8 lb CO2e/MWh
California430 lb CO2e/MWh
Virginia / Carolinas596.3 lb CO2e/MWh
Northwest635.3 lb CO2e/MWh
Southwest706.2 lb CO2e/MWh
ERCOT736.6 lb CO2e/MWh
United States770.9 lb CO2e/MWh
SERC Midwest1,248.6 lb CO2e/MWh

derived orientation map

Where are the eGRID subregions named in the chart, and why are they not simply states?

Orientation map of the contiguous United States divided into EPA eGRID2023 subregions, with the regions used in the adjacent annual output-rate comparison identified. A key provides every selected region name without relying on color.

Contiguous U.S. · SRVC inspection active · select a region for detail

EPA eGRID2023 subregions used to interpret the adjacent annual output-rate comparison. Boundaries are representational; some places can be served by utilities in more than one subregion.
Scope
Contiguous United States eGRID2023 electricity subregions
Period
eGRID2023 geometry published January 15, 2025; adjacent generation-year 2023 rates use Revision 2
Metric
Geographic orientation only; the adjacent chart supplies lb CO2e/MWh

LimiteGRID subregions are approximate annual electricity-accounting regions, not state lines, exact utility territories, local pollution footprints, or a forecast of which generators a new load would cause to operate or be built. Some places have more than one applicable utility region.

Use the national baseline to read the local record.

National baseline

A data center's electricity footprint changes with the regional generation mix. EPA subregion rates vary substantially. Those annual averages still cannot tell you which plant a new load will dispatch or cause to be built.

Local case

Grid emissions are regional; generator emissions are local. A project page should show both without using a renewable contract to erase physical grid conditions or a permit maximum as if it were actual pollution.

What the evidence supports.

Data centers continuously pollute the neighborhood air.
Not established without the power source and operating record

A facility’s power source and operating hours determine whether combustion emissions are continuous or intermittent. A grid-supplied site with standby engines is different from one generating its routine power on site. Standby engines emit during tests and outages. Neither a national electricity footprint nor an annual permit cap establishes continuous neighborhood exposure.

Backup generators do not matter.
False: backup engines emit pollutants

Backup engines emit pollutants and must be included in the air assessment. A campus can aggregate many engines. One 2022 Virginia permit covered 123 units. Its annual caps do not show actual emissions; operating logs and fuel records provide that evidence.

Ask for these local records.

Without these inputs, a project-specific verdict is incomplete. Treat missing evidence as an open question.

  1. 01Electricity supplier, eGRID subregion, contractual claims, and location-based accounting
  2. 02Engine count, size, fuel, control tier, permit limits, and stack-test results
  3. 03Actual annual test, maintenance, commissioning, outage, and demand-response hours
  4. 04Any behind-the-meter prime generation kept separate from emergency backup
  5. 05Nearby sensitive receptors, cumulative permitted sources, monitoring, and complaint records

Sources used on this page.

  1. Official reportGrade B
    2024 United States Data Center Energy Usage Report

    United States; historical estimates through 2023 and scenarios through 2028

    Modeled national estimates with limited facility-level public data. Location-based emissions and indirect water do not include facility-specific contracts or behind-the-meter supply.
  2. Agency dataGrade A
    eGRID2023 Revision 2 summary data

    U.S. eGRID subregions and national total

    Annual average output rates attribute a regional generation mix. They do not identify short-run marginal dispatch or long-run resources caused by a new load.
  3. RegulationGrade A
    Stationary engine New Source Performance Standards

    Stationary compression-ignition engines in the United States

    Specific requirements vary by engine age, size, source classification, and use. The rule summary is not a facility operating record.
  4. PermitGrade A
    Air permit 74162

    One Vantage Data Centers campus in Sterling, Virginia

    This is one older Virginia permit. Its caps show maximum allowed emissions; actual emissions require operating data. Newer Virginia control guidance is more stringent.
  5. AuditGrade B
    Data Centers in Virginia

    Virginia, primarily FY2021–FY2023, with selected forecasts

    Virginia-specific. Several values are stakeholder estimates or model outputs, and future utility-cost scenarios are explicitly uncertain.
  6. Official reportGrade A
    Health Assessment Document for Diesel Engine Exhaust

    Human-health hazards from acute and chronic diesel-exhaust inhalation

    Establishes the hazard from diesel exhaust. Exposure and risk from a named engine fleet require site-specific evidence. Engine technology and controls have also changed since the assessment.
  7. Official reportGrade B
    Data centers: monitoring diesel pollution in Quincy

    Diesel-particle monitoring, health-risk assessment, and permitting in Quincy, Washington

    One community and one modeling process. Washington Ecology identified technical issues in the owner-funded 2018 assessment and later performed follow-up analysis.
Browse the complete source library