Summary
A Houston Advanced Research Center assessment projects that Texas data-centre electricity demand could push some power-sector emissions back towards levels last seen in the early 2000s. The model’s results vary substantially with the power supply chosen.
A Houston Advanced Research Center (HARC) assessment projects that rising data-centre electricity demand could reverse some of Texas’s long-running declines in air-pollutant emissions. Its model estimates emissions through 2031 under different data-centre growth forecasts and power-supply approaches; the projected effects vary sharply depending on how the additional electricity is generated.
Texas’s power-generation sector cut nitrogen oxide (NOx) emissions by 67% and sulphur dioxide (SO₂) emissions by 80% between 2002 and 2024, according to the report’s sector comparison. HARC’s assessment examines whether the growing electricity load from data centres could change that trajectory. NOx and volatile organic compounds (VOCs) contribute to ground-level ozone formation, while PM2.5 refers to fine particulate pollution.
Emissions depend on the power supply
In a scenario where new data-centre demand is met by the projected grid mix, HARC estimates that 2031 NOx emissions would be 136% above the 2025 level, approaching the volume recorded in 2004. VOC emissions would exceed their 2002 volume by 32%, and PM2.5 emissions would reach the 2006 volume. SO₂ emissions would rise by 18% but remain below their 2002 level.
Those are projections of emissions, not measurements of the concentrations people would breathe or estimates of health effects. The figures also describe a particular supply scenario, rather than a prediction that every new data centre will use the same power mix.
HARC’s comparisons show why that distinction matters. Depending on the data-centre demand forecast, NOx emissions from power generation could be two to ten times higher if the new load were met with on-site combustion rather than grid electricity paired with solar and wind. The assessment also finds that even its cleanest power-supply configuration adds emissions relative to the model’s no-new-data-centre baseline: the added electricity demand still has to be met.
The report therefore points to power-supply choices as a major influence on the modeled air-pollution effects, alongside the amount of data-centre capacity built. HARC says the projections are intended to help policymakers and other stakeholders assess those trade-offs.
How HARC built the projections
The assessment uses three sources of projected data-centre capacity: ERCOT’s Adjusted Large Load Breakdown, transmission service providers’ large-load breakdowns, and BAXTEL data. HARC first modeled electricity and fuel use for a representative 1-gigawatt data centre, then scaled those estimates to the projected Texas capacity. The model assumes facilities operate 24 hours a day, 365 days a year, at 80% capacity, and that about 40% of their electricity demand goes to cooling.
For its baseline, HARC projects recent emissions trends forward without adding new data centres. It uses Texas Commission on Environmental Quality point-source emissions data from 2018 to 2024 and grid-generation information spanning 2017 to 2025, among other inputs. Data-centre emissions under each modeled supply approach are then added to that baseline.
These results depend on the assumptions behind the projections. HARC notes that actual data-centre developments may use a mixture of power-supply approaches, while its scenarios assign all new demand to a selected mix. It also scales results from one representative facility and assumes planned projects in the BAXTEL dataset come online on schedule. Changes in construction timing, facility design or Texas’s future grid and industrial emissions trends could shift the projected totals.