Lifestyle
Texas Data Center Boom Could Erase Decades of Air Quality Gains
The Verge reports that a new report from the Houston Advanced Research Center (HARC) finds that if all of Texas's planned data centers connect to the grid, soot and smog-forming pollution from the power sector could rise sharply. This piece covers the scenarios, key figures and what it means for ordinary people.
About 5 min read
What happened
The Verge reports that new research in Texas shows data centers are driving up pollution. It notes that Texas has become a hub of the physical infrastructure for generative AI. The wave of new data centers could push air pollution back to levels not seen since the 2000s.
As The Verge explains, the new report from the Houston Advanced Research Center (HARC) assumes a scenario in which all planned data centers connect to the Texas grid and are powered by the grid's expected generation mix, meaning the proportions of gas, renewables and other sources on the grid. In this scenario, soot and smog-forming pollutants climb sharply. According to The Verge, within just five years, the pollution would be enough to cancel out decades of air quality improvements.
Key figures in the report
According to The Verge, under this scenario, power-sector fine particulate matter (soot) would more than triple, reaching its highest level since 2006. Emissions of nitrogen oxides (NOx), which form smog, would rise 136%, close to 2004 levels. Carbon dioxide emissions would be 90% higher than in 2025.
| Indicator | Scenario result as relayed by The Verge |
|---|---|
| Fine particulate matter (soot) | More than triple, highest since 2006 |
| Nitrogen oxides (NOx) | Up 136%, close to 2004 levels |
| Carbon dioxide | 90% higher than in 2025 |
| With on-site renewables | NOx as low as one-tenth |
The power source determines the pollution
The Verge says whether developers and utilities meet demand with gas turbines or renewables will determine how severe the pollution is. If data centers combine grid power with renewable electricity generated on site, NOx emissions could be as low as one-tenth.
The Verge also points to the opposite possibility. Grid connections often involve long waits, and developers who build gas engines on site to avoid the wait could produce the highest pollution levels. The Verge adds that many developers use less efficient gas turbine technology to speed up construction, making this boom dirtier than necessary.
Data center size is usually measured by power consumption. The Verge says by that measure Texas is already the largest data center market, with 26 gigawatts (a unit of power) of existing and under-construction capacity. It also cites another report, published in March by Wood Mackenzie, finding that among data centers proposed in Texas, the capacity to be powered by on-site generation is four times that of any other state.
What it means for ordinary people
For ordinary readers, the computing behind AI services requires large amounts of electricity, and how that electricity is generated affects the air in nearby communities. HARC CEO John Hall said in a press release: "There is a good chance that our children and grandchildren will breathe dirtier air than we did." He also said that the transportation sector, the state and other stakeholders have worked for decades to bring emissions down, and that this progress is now at risk.
Note that, according to The Verge, these figures come from HARC's modeled scenarios, not actual measurements. The real impact depends on whether projects get built and which power sources they use.
Frequently asked questions
Who conducted this study?
According to The Verge, the report comes from the Houston Advanced Research Center (HARC).
Will air pollution definitely return to 2000s levels?
Not necessarily. The Verge notes this is a scenario in which all planned projects connect to the grid and use the expected grid generation mix. Many proposals are speculative and may not be built, and on-site renewables can reduce the impact.
Which approach produces less pollution?
According to The Verge, data centers that combine grid power with on-site renewable electricity could cut NOx to as little as one-tenth, while building gas engines on site could produce the highest pollution levels.
Why is Texas getting so much attention?
The Verge says Texas is the largest data center market by power consumption, with 26 gigawatts of existing and under-construction capacity. It also cites a Wood Mackenzie report finding Texas's proposed capacity powered by on-site generation is four times that of any other state.
Browse the latest news in this topic
Lifestyle
IBM's Granite Timeseries Ensemble Combines Four Forecasting Models So Teams Need Not Pick One
IBM Research has released Granite Timeseries Ensemble, which combines four Granite forecasting models into one default recipe so teams forecasting demand, usage or sensor data need not guess which model fits best. IBM's reported gains are small; the main benefit is robustness, at extra compute cost.
Lifestyle
Blacksmith, Which Helps Check AI-Written Code, Raises $66M Series C Led by Sequoia at a $1 Billion Valuation
TechCrunch reports that Blacksmith, which helps companies build and test software, including AI-written code, has announced a $66 million Series C led by Sequoia at a $1 billion valuation. Its CEO says the more code AI writes, the more companies need to know whether software can be trusted.
Lifestyle
Blackstone's Jas Khaira Says Fears Over AI Infrastructure Debt Are Overblown, Even as Borrowing Costs Rise
At TechCrunch Disrupt, Blackstone's Jas Khaira said the debt piling up to build AI data centers is not like the shaky banks of 2008. Here is how Blackstone lends to AI cloud providers, the strains now showing in credit markets, and why the cost of funding the AI boom matters beyond Wall Street.
Lifestyle
Meta Announces New Non-Voting Share Class to Fund AI Investment
According to The Verge, Meta said in its third-quarter earnings that it will issue a new class of non-voting shares for AI initiatives, potential acquisitions and employee compensation, with trading to begin in 2027. It shows how large the funding needs of the AI race are.