The development of artificial intelligence infrastructure is not only a technological advancement but also turns into a serious environmental problem.
The rapid development of artificial intelligence increasingly requires the creation of more data centers to support powerful computing systems. However, alongside technological progress, the environmental burden also increases.
According to researchers, by the end of the decade, artificial intelligence infrastructure could become a significant source of greenhouse gas emissions in the USA.
Data centers are turning into a new source of emissions
Modern data centers consume large amounts of electricity. In many cases, they receive energy from power plants that use natural gas and other fossil fuels. This leads to an increase in carbon dioxide emissions and adds burden to the environment.
According to calculations by researchers from Cornell University, if the development of artificial intelligence continues at the current pace, data centers will emit between 24 and 44 million metric tons of carbon dioxide annually by 2030. This amount of pollution can be compared to adding 5 to 10 million additional cars on US roads.
Experts note that the rapid increase in demand for computing power is already forcing energy companies to expand electricity production, and not always from renewable sources.
Why Texas is in focus
According to research by the climate organization "Floodlight," cited by the American publication "Wired," Texas has become one of the leading regions for the construction of new data centers.
Journalists have discovered that some companies build their own natural gas power plants near these facilities. This not only ensures the smooth operation of servers but also significantly increases harmful emissions.
According to analysts from the Global Energy Monitor, the rate at which new gas-powered energy facilities are added in Texas is second only to China. This shows how rapidly the energy needs of the artificial intelligence industry are increasing.
How companies bypass environmental restrictions
The authors of the Floodlight study claim that at least 38 data centers in Texas have taken advantage of existing legislative features to obtain simplified permits for their energy supply.
In total, more than 2100 backup diesel generators are used in these facilities. According to experts' calculations, they emit approximately 2500 tons of nitrogen oxides annually — substances that worsen air quality and can negatively affect human health, especially those with respiratory diseases.
Additionally, journalists report that some operators initially applied for limited-emission relatively small data centers, and after obtaining permits, gradually expanded their infrastructure.
Large-scale projects receive simplified approvals
As written in "Wired," some large facilities initially undergo ecological procedures designed for small establishments.
For example, the data center in Abilene, Texas, which is part of the massive Stargate project estimated to cost 500 billion dollars, can be mentioned. The facility is equipped with 62 backup diesel generators, even though such approvals are usually prepared for facilities that use only one or two such energy sources.
The environmental cost of artificial intelligence development
Experts note that the problem goes beyond increasing carbon dioxide emissions. The operation of data centers requires large amounts of water to cool servers and put a significant load on regional power grids.
Therefore, many scientists believe that the further development of artificial intelligence should be accompanied by active use of renewable energy sources, implementation of more energy-saving technologies, and stricter environmental requirements for new facilities.
According to the authors of the study, artificial intelligence has the potential to accelerate scientific and economic progress, but without responsible approaches to infrastructure development, its rapid expansion may lead to serious environmental consequences. (lent.az)
20.07.2026, 06:58
201
views
Zara Editor