A study led by researchers from Cambridge University, Nanyang Technological University, and partner institutions has found that Artificial Intelligence data centres are generating a localised warming phenomenon around their sites, raising land surface temperatures by an average of 2 degrees Celsius in the months after they open, with some locations recording increases as high as 9.1 degrees Celsius and the effects detectable up to 10 kilometres away.
The researchers, who used NASA satellite data to measure land surface temperatures globally from 2004 to 2024 and cross-referenced findings against more than 11,000 Artificial Intelligence data centre locations worldwide, have described the phenomenon as the “data heat island effect,” drawing a direct comparison with the urban heat island effect that causes cities to run warmer than surrounding areas due to concentrated human activity. The study found that more than 340 million people living within 10 kilometres of a data centre could be exposed to these temperature increases, with researchers describing the effect as having a significant influence on communities and regional welfare that should be factored into global conversations about environmentally sustainable Artificial Intelligence development.
The energy demands behind these temperature effects are considerable. According to the International Energy Agency, data centres consumed approximately 415 terawatt-hours of electricity in 2024, representing about 1.5 percent of global electricity supply, growing at roughly 15 percent per year over the previous five years, with projections pointing to nearly double that consumption by 2030. Hyperscale data centres, the largest facilities built by major technology companies to support cloud computing and Artificial Intelligence at global scale, typically require between 100 and 300 megawatts of electricity to operate continuously and house a minimum of 5,000 servers. That energy output generates enormous heat that must be managed through liquid cooling systems consuming vast quantities of water, with a single 100-megawatt hyperscale facility estimated to use approximately 2.5 billion litres of water annually, equivalent to the yearly needs of around 80,000 people. As of June 2026, more than 11,600 data centres are active worldwide, with the United States hosting more than 4,300, the highest concentration of any country, followed by the United Kingdom, Germany, France, China, and India.
The financial scale of investment flowing into this infrastructure is equally striking. Goldman Sachs estimates combined capital expenditure of $5.3 trillion between 2025 and 2030 across the four largest hyperscalers alone, covering Microsoft, Amazon, Alphabet, and Meta. Major projects underway or announced include Meta’s $27 billion Hyperion campus in Louisiana, Microsoft’s $20 billion campus expansion in Wisconsin, Amazon’s $25 billion investment in Mississippi, Google’s $15 billion Project Spade in Missouri, and Oracle’s Project Stargate in Texas, a dedicated Artificial Intelligence supercluster for OpenAI with capacity projected between 1.2 and 2 gigawatts.
Follow the SPIN IDG WhatsApp Channel for updates across the Smart Pakistan Insights Network covering all of Pakistan’s technology ecosystem.
