The boom in AI data centers is starting to affect not only the energy system but also the local climate. A new study has found that large data centers can significantly increase the temperature around them and create a heat island effect.
The boom in AI data centers is starting to affect not only the energy system but also the local climate. A new study has found that large data centers can significantly increase the temperature around them and create a heat island effect.
According to the study The data heat island effect: quantifying the impact of AI data centers in a warming world, published on arXiv, an international group of scientists analyzed satellite data on the temperature of the earth’s surface around AI data centers and concluded that after the launch of such facilities, the surface temperature in the surrounding areas increases by an average of 2°C. The authors attribute this to the heat generation of large data centers that operate to maintain AI services.
In the worst cases, the researchers recorded a temperature increase of up to 9,1°C. The effect was not limited to the area of the facility itself: the impact, according to the authors, can extend up to 10 km around the data center. Fortune, which reported the conclusions of the work, directly called this effect a new type of heat island associated with AI infrastructure.
The study also assesses the scale of the potential impact on people. According to the authors, such heating could affect the living conditions of more than 340 million people worldwide. This is not only about discomfort, but also about additional pressure on cooling systems, energy consumption and the overall quality of life in regions where large data centers are concentrated.
In fact, this is another signal that AI infrastructure has a very material price. While the market is discussing new models, chips, and billions of dollars in investments in computing power, another side of this race is becoming increasingly visible: the heat that this power emits into the environment.
The authors of the paper explicitly state that data centers are becoming part of the conversation about the environmentally sustainable future of AI. In their opinion, further scaling of infrastructure without more efficient cooling, more energy-efficient hardware, and changes in computing approaches could increase local thermal impacts in different regions of the world.
Previously, dev.ua wrote about how companies around the world sold $791.7 billion in semiconductors in 2025, a 25,6% increase from the previous year. Industry representatives expect that the amount could rise to $1 trillion in 2026, amid multibillion-dollar spending by tech giants on data centers for artificial intelligence.
