According to recent analysis from specialty Lloyd’s insurer MS Amlin, just over half (51%) of planned US data centre projects worth $670 billion are located in states at high risk of severe convective storms (SCS).
MS Amlin’s analysis examined more than 670 data centre projects under construction or in planning across the US, finding that 320 facilities are classified as high risk for tornadoes, large hail and damaging winds.
The study found that existing data centres in states highly exposed to SCS are valued at almost $20 billion, suggesting that future AI infrastructure in storm-exposed regions could be nearly 40 times the value of existing facilities.
Overall, 56% of the 670 planned US data centres—representing nearly $800 billion in investment—are located in states highly exposed to hurricanes, severe convective storms, earthquakes or winter storms.
MS Amlin revealed that 27% of data centres, representing $440 billion in investment, are planned in states at high risk of winter storms, which can disrupt power networks and create complex business interruption risks.
Meanwhile, 21% of planned data centres, amounting to $340 billion of investment, are located in states at high risk of hurricanes.
In addition, data centres in high-risk earthquake states account for 3% of planned facilities, representing around $12 billion in investment.
MS Amlin said the findings highlight the scale of investment flowing into states at risk of natural catastrophes as development of new hyperscale facilities shifts to southern regions where land and power are more favourable.
Martin Burke, MS Amlin’s Chief Underwriting Officer, said, “These numbers highlight both the opportunity and the risk. Hundreds of billions of dollars of new digital infrastructure are being directed towards regions at higher risk of potentially destructive severe convective storms. When assets of this scale cluster in hazard prone regions, the potential loss severity from a single storm event can rise very quickly. This is a growth opportunity for the specialty insurance market, but the risks must be properly managed and understood.”
He added, “As AI investment accelerates, insurers must adopt more advanced ways to manage aggregation risk. If the industry is slow to address this challenge, it could restrict the deployment of capital and roll out of AI infrastructure.
“Our proprietary database of hundreds of US data centre projects lets us capture the risk not just from tightly clustered facilities but also from supporting infrastructure like power generation. This provides a far more accurate picture of overall exposure.
“This visibility allows us to deploy capacity responsibly to support the sector’s growth while maintaining underwriting discipline. The ability to monitor aggregation risk is becoming increasingly important as this class continues to grow.”
