Approximately 79% of global data-center capacity is exposed to elevated risk from acute climate hazards [1].
This vulnerability threatens the stability of the digital economy, as these facilities house the critical infrastructure required for everything from banking to artificial intelligence. Because the rapid expansion of AI-driven capacity often places infrastructure in vulnerable locations, the risk of operational disruption increases.
The findings come from a study released in June 2026 by First Street, a climate-risk financial-modeling firm [1]. The report identifies five primary categories of climate hazards that threaten these facilities: flooding, extreme heat, wildfires, wind, and drought [2].
These acute events can lead to increased downtime, higher repair costs, and rising insurance premiums for operators [1]. The study highlights a feedback loop where the infrastructure is at risk from the very environmental changes its own energy consumption helps accelerate.
"79% of global data-center capacity faces acute climate hazards, according to First Street," Diana Olick said in a report for CNBC [1].
The study authors said that AI datacenters are particularly vulnerable to the climate hazards that their own emissions help to intensify [3]. This risk is most prevalent across the largest and fastest-growing data-center markets globally [1].
As operators seek to scale capacity to meet AI demand, the pressure to build quickly may outweigh long-term climate resilience planning. The report suggests that the current pace of development is outpacing the implementation of necessary safeguards against extreme weather [3].
“79% of global data-center capacity faces acute climate hazards”
The concentration of digital infrastructure in climate-vulnerable zones creates a systemic risk for the global economy. As AI continues to drive a massive build-out of data centers, the industry faces a paradox where the physical growth of the cloud is threatened by the environmental instability caused by its own energy and carbon footprint.
