Photo: EPA
Extreme heatwaves, floods, and storms are increasingly threatening the operation of data centres that power artificial intelligence systems, according to CNBC.
Record-breaking heat in Europe has highlighted the vulnerability of this critical infrastructure. High temperatures increase the load on cooling systems and electricity grids, raising the risk of disruptions.
Patrick McBride, Zurich’s head of international construction insurance, said extreme weather events have become the leading cause of losses in data centre construction over the past three years in the United States, now accounting for about one-third of all damages. He stressed that severe weather can no longer be treated as a background factor.
The geography of new data centre construction is also shifting. According to McBride, 64% of new capacity is now being built outside traditional hubs, including regions such as West Texas, Tennessee, Wisconsin, and Ohio, where risks from tornadoes, hail, and strong winds are higher.
Research by First Street shows that around 79% of global data centre capacity is located in areas exposed to flooding, wildfires, extreme winds, or other climate risks. These conditions can lead to downtime, higher insurance costs, and expensive equipment repairs.
Experts warn that power grid strain is another growing issue. Cooling alone already accounts for about 40% of total data centre electricity use, and this rises further during heatwaves.
Technology companies are beginning to adapt. Microsoft is designing data centres for a wider range of climate conditions, using improved site selection, system redundancy, and real-time monitoring. Nvidia has upgraded cooling systems for its latest AI servers, allowing operation with coolant temperatures up to 45°C. Amazon, meanwhile, reported that its data centres used about 2.5 billion gallons of water in 2025, while slightly reducing overall consumption through efficiency measures and alternative cooling methods.
The UN has also launched a new initiative calling for greater transparency from tech companies on energy use, emissions, and water consumption, and urging a full transition to renewable energy for AI infrastructure by 2030.
