Portuguese 
English 
Spanish 
6 min of reading 
0 comments 
In Phoenix, a study from Arizona State University indicates that AI data centers linked to artificial intelligence may intensify thermal pollution and urban heat, with an average of 0.9 ºC downwind and peaks of 2.2 ºC, leading hot cities to rethink location, cooling, and urban green areas.
AI data centers may be creating a new urban heat pressure in Phoenix, where a study linked structures used by artificial intelligence to thermal pollution. The research indicates an average increase of 0.9 ºC downwind and peaks of up to 2.2 ºC near these facilities.
The research analyzed the metropolitan area of Phoenix, considered the hottest in the United States, and brought an additional concern to the debate about artificial intelligence. According to the portal Xataka, besides energy and water consumption, data centers can also increase the so-called thermal pollution, turning residual heat into an urban, environmental, and even public health problem.
Study measured heat near AI data centers in Phoenix
The researchers focused on Phoenix precisely because it is a region already dealing with high temperatures. In this context, any increase in heat can have a significant impact on residents’ lives, energy consumption, and the functioning of urban infrastructure.
-
A Soviet cosmonaut left Earth for a routine mission, spent 311 days stranded on the Mir station, and returned as the “last citizen” of a country that had already disappeared.
-
Engineer transforms chopsticks discarded by restaurants into premium furniture, presses used bamboo until it becomes harder than maple and more resistant than oak, and shows how food waste can turn into high-value design.
-
Doctoral student from UEPB creates solar desalination device that transforms brackish water into potable water, costs much less than traditional systems, and is already improving the lives of families who previously walked hours for water in the semi-arid region.
-
“X-ray” technology used in aerial surveys will map the beaches in different seasons of the year to identify where the sand is disappearing. A city located in SC is undergoing the largest beach widening in Brazil and now wants to understand the dynamics of the sea.
The study evaluated how the air expelled by AI data centers behaves after passing through cooling systems. These centers need to keep servers at a controlled temperature, and to do so, they release hot air into the external environment.
According to the analysis, this air can be released between 14 and 25 degrees Fahrenheit above the ambient temperature. In dense or already very hot regions, this difference can contribute to warming streets, blocks, and nearby residential areas.
The most striking conclusion is that the heating is not confined to the building. The areas downwind showed an average temperature 0.9 ºC higher, with peaks of 2.2 ºC compared to reference locations.
Sensors in vehicles helped map the temperature difference
One of the relevant points of the research is the method used to measure the phenomenon. The study was described as the first known to use high-precision sensors mounted on vehicles to compare air temperature before and after passing through a data center.
This strategy allowed for the observation of heat behavior on an urban scale. Instead of analyzing only internal measurements or theoretical estimates, researchers monitored the air temperature in areas near data centers.
The data indicated that residual heat travels through the surroundings. The increase was detected up to about 500 meters from the source, a distance equivalent to approximately five blocks in Phoenix.
This means that AI data centers can affect not only their immediate grounds but also nearby neighborhoods, depending on wind direction, urban design, natural ventilation, and the presence or absence of green areas.
Thermal pollution enters the debate on artificial intelligence
Until now, much of the discussion about artificial intelligence and digital infrastructure focused on energy and water consumption. The new study adds another issue: the heat these facilities return to the environment.
AI data centers operate with high computational capacity, requiring intense cooling to keep servers running. The greater the demand for AI, the higher the pressure tends to be for new processing centers.
Thermal pollution is less visible than smoke or waste, but it can be directly felt by those living nearby. In hot cities, an increase of 1 ºC or 2 ºC can intensify discomfort, increase air conditioning use, and expand risks during heatwaves.
The study does not claim that data centers should be banned. The central discussion is another: how to plan these structures so that they do not worsen the thermal conditions of already vulnerable neighborhoods.
Heat cycle may increase air conditioning use
The problem can become a cycle. The data center releases hot air to cool servers. This air heats the neighborhood. Nearby residents and businesses use more air conditioning. The devices, in turn, also emit heat outside.
This process can reinforce the so-called urban heat island, a phenomenon where built-up areas become warmer than less dense or more wooded regions. In places like Phoenix, where the heat is already extreme, the effect can be even more noticeable.
The research indicates that a single data center can generate as much residual heat as a small city of 40,000 homes. This data helps to gauge the potential impact of these facilities.
When the heat from a digital infrastructure adds to common urban heat, the problem is no longer just technological. It starts to involve public health, urban planning, energy, housing, and environmental justice.
Location of data centers becomes a strategic decision
David Sailor, leader of the study, argues that the findings should serve to rethink the integration of data centers into cities. The goal would not be to prevent these structures, but to treat them as sources of industrial heat.
Among the suggested solutions are reorienting air outlets, distancing facilities from sensitive residential areas, and creating parks or green zones to cushion the temperature increase. These measures can reduce the direct impact on nearby neighborhoods.
The location becomes as important a decision as the technical design. Installing AI data centers in already hot, dense areas with little vegetation can exacerbate existing problems.
On the other hand, planning these structures with ventilation corridors, buffer areas, and environmental criteria can allow digital infrastructure to grow with less thermal impact on the population.
AI expansion could double demand for data centers
The concern increases because the computational capacity projected for the data centers to be built in the United States is expected to double by 2030, according to the cited study. This growth accompanies the accelerated expansion of artificial intelligence.
More models, more services, and more processing mean more servers operating continuously. This also means more cooling requirements and greater generation of residual heat.
The race for AI is creating a much larger physical infrastructure than many users realize. Although artificial intelligence seems to work “in the cloud,” it relies on buildings, energy, water, cooling, and networks installed in real territories.
If this expansion occurs without adequate urban planning, hot cities may feel the impact unevenly. Neighborhoods near large facilities may start to experience more heat precisely where the population already faces high temperatures.
Spain also appears in the alert about new projects
The study was conducted in Phoenix, but the discussion can apply to other regions trying to attract data centers. The source cites the case of Spain, especially the Autonomous Community of Aragon, which has gained prominence in agreements with large technology companies.
Amazon and Microsoft have data center projects planned for the metropolitan area of Zaragoza. The cities of Villamayor de Gállego and Villanueva de Gállego, both less than 20 kilometers from Zaragoza, also appear in the debate for receiving plans related to this type of infrastructure.
These initiatives promise to boost the local economy, generate investments, and strengthen the region’s position in the digital sector. At the same time, they raise questions about energy, water, environmental impact, and now also residual heat.
The lesson from the study is that the arrival of AI data centers needs to be evaluated before construction, not after the problem appears. Prevention can be more efficient than trying to correct already formed heat islands.
Artificial intelligence also occupies space in cities
The research shows that artificial intelligence is not just a digital phenomenon. It relies on a physical network of infrastructure that occupies land, consumes resources, and interferes with the surrounding environment.
The AI data centers may be essential for modern services, but they also need to be part of discussions about urban planning, climate, and public health. The challenge is to balance technological innovation with quality of life in cities.
In hot regions, this balance becomes more urgent. If a data center increases the temperature in the surrounding area, the impact may fall on residents who do not directly participate in the economic gains generated by the installation.
And you, do you think cities should limit the installation of AI data centers near residential neighborhoods, or does the advancement of artificial intelligence justify new constructions even with the risk of more urban heat? Share your opinion.
Be the first to react!