The global rush to build hyperscale data centers is concentrating exposure to natural catastrophes, cooling failures and fragile supply chains at a scale the insurance market is struggling to keep pace with.
The buildout represents one of the largest construction opportunities in modern history, but for underwriters and brokers, stacks unprecedented values at risk into single locations, often in secondary-peril territory, while aggressive construction timelines, specialized equipment and long-lead replacement parts stretch traditional builders risk and business interruption assumptions.
Achim Hillgraf (pictured), operations senior vice president of global growth strategies at FM, said the insurer expects another US$6.5 trillion to be committed to data center development by 2030. "There's going to be a huge boom, and there's going to be a lot of construction because those US$6.5 trillion worth of projects haven't broken ground yet," Hillgraf told Insurance Business.
The unprecedented demand for hyperscale facilities is already testing the limits of what the insurance market can absorb. S&P Global Ratings estimates total insurable values for a single hyperscale campus can reach US$20 billion to US$30 billion, including during the construction phase, compared with typical limits of US$5 billion to US$10 billion for major infrastructure projects such as bridges or tunnels.
Even the market's largest dedicated facilities fall short of the values now in the pipeline. Aon expanded its Data Center Lifecycle program to US$3.5 billion in April, while Willis has assembled roughly US$3 billion in hyperscale construction capacity, as Insurance Business has previously reported.
Zurich's global head of construction and surety, Kelly Kinzer, has similarly warned that at the largest end of the scale there is simply not enough capacity in the market to insure these projects to full value — a growing stress point as lenders increasingly demand full-value coverage
The gap is not theoretical for capital providers. Moody's insurance analysts noted that in March 2026, investors including Blackstone were reported to have passed on data center debt opportunities because of insufficient insurance. Where cover cannot be found, the residual risk flows back to sponsors and developers, with hyperscalers increasingly absorbing it on their own balance sheets, often through captives.
The global buildout raises different risk concerns depending on geography. In the United States, Hillgraf pointed to the increasing importance of so-called secondary perils, including convective storms, hail, tornadoes and wildfires. Although developers may avoid obvious catastrophe zones such as the Florida coastline, inland sites can still face severe weather exposures.
The trend is already visible in carrier loss experience. Severe weather has been the leading cause of loss in Zurich's US data center builders risk portfolio for the past three years, while a tornado at a data center site was among the leading loss causes in 2025. In 2026, 64% of data center capacity under construction sits outside traditional hubs such as Northern Virginia, pushing projects into interior markets where convective storm and hail exposure is elevated.
Europe has historically faced less severe catastrophe exposure than parts of North America or Asia, but Hillgraf said that assumption is becoming less reliable. He cited flooding, windstorms, occasional tornadoes and the emergence of heat domes (i.e., prolonged periods in which a high-pressure system traps hot air over a region, driving sustained extreme temperatures) as growing concerns.
That last risk is particularly significant, according to Hillgraf, because data centers depend on continuous cooling. Cooling failure is the second-most common cause of impactful data center outages after power problems, according to the Uptime Institute's Annual Outage Analysis 2025, which also warned that soaring AI demand is straining infrastructure designs, particularly around power and cooling.
In Asia, typhoons, earthquakes and volcanic activity add another layer of location risk. Hillgraf said developers should avoid exposed shorelines where possible and account for regional hazards during construction rather than after a facility has been completed.
The construction boom also comes with what Hillgraf described as broader megatrends of intensifying natural hazards and increasingly interconnected supply chains.
"Natural hazards are happening more often and they're happening more violently everywhere around the world," he said, adding that companies may be exposed directly through their facilities or indirectly through suppliers.
Supply chains, meanwhile, have become more efficient but also more fragile, with cost pressure encouraging businesses to operate with fewer buffers. Potential disruptions range from tariffs and trade restrictions to natural catastrophes, armed conflict and blocked shipping routes.
For data centers, that fragility could mean lead times for transformers, switchgear, precision cooling units and GPUs that stretch to months; a property loss that would sideline a conventional facility for weeks can leave a data center dark far longer.
In response to the demand for risk expertise, FM has established a specialist unit, FM Intellium, focused on insuring hyperscale data centers globally, with involvement beginning at the construction stage. The carrier's approach starts with site selection where possible, followed by physical protections (such as materials proven to withstand defined wind speeds or hail sizes) when location cannot be changed.
"We're very keen on engaging very early with those clients to make sure that we can help them build those facilities as resiliently as they need to be built before they come online," Hillgraf said.
