Swiss Re warns AI data centers are outgrowing insurance industry's ability to cover them

AI data centers now cost up to $20 billion to build-double that with servers installed-and insurers are struggling to keep pace. Global data center insurance premiums are forecast to hit $24.2 billion by 2030, up from $10.6 billion in 2026.

Categorized in: AI News Insurance
Published on: May 29, 2026
Swiss Re warns AI data centers are outgrowing insurance industry's ability to cover them

AI data centers strain insurance models as risks mount

The insurance industry is struggling to cover the financial risks posed by massive AI data centers, according to a report from the Swiss Re Institute. Construction costs for a single facility can reach $20 billion-a figure that doubles once servers and graphics processing units are installed.

Global insurance premiums for data centers are expected to climb to $24.2 billion by 2030, up from $10.6 billion in 2026. Financing institutions increasingly require coverage matching the full cost of construction, driving the surge.

Natural disasters pose concentrated risk

More than a quarter of US data center capacity sits in areas experiencing three or more large-hail days per year. More than 40% of capacity is located in significant-to-very-high tornado zones where tornadoes of Enhanced Fujita 1 strength or greater occur at least three days annually.

Developers are clustering large data center campuses within roughly 20 miles of one another-such as in Abilene, Texas, and Virginia-raising the risk that a single natural disaster could strike multiple high-value locations simultaneously. A tornado's debris field can readily damage multiple buildings on the same campus.

Fire and cooling systems create new hazards

Fire accounts for just 10.9% of data center loss events but represents 42.3% of loss costs, according to an FM Global study cited in the report. Lithium-ion battery backup units integrated into server racks create an ignition source that "did not previously exist" in data processing equipment rooms.

Liquid-related losses account for nearly 24% of total data center loss costs. High-performance GPUs require direct-to-chip liquid cooling, and the scale of these networks creates risks from improper installation or maintenance. Around 30% of planned US data center capacity could include on-site power generation.

Traditional servers required 5 to 15 kilowatts per rack. AI servers can exceed 100 kilowatts per rack, making power supply the leading driver of business interruption, accounting for 45% of outages.

Underwriting obstacles emerge

Large data centers are sometimes presented to insurers through separate programs that cover buildings, equipment, and power plants independently. This structure makes it difficult to track overall exposure and accumulation risk.

Underwriting success depends on specialized technical assessment and disciplined accumulation management, the report said.

Recent failures underscore exposure

In September 2025, a lithium-ion battery explosion triggered a fire at South Korea's National Information Resources Service data center in Daejeon, shutting down hundreds of government digital services. The blaze disrupted mobile identity verification for banking and airports, postal services, transportation systems, and the government's internal intranet. South Korean President Lee Jae Myung called the incident "foreseeable" and ordered comprehensive infrastructure inspections across all government agencies.

In May 2026, a 12-hour fire at a NorthC data center in Almere, Netherlands, knocked services offline across multiple sectors. Utrecht University closed most of its buildings after its network went down. Bus and tram dispatch services were disrupted, hospitals reported partial system outages, and the national statistics bureau CBS experienced prolonged downtime.


Get Daily AI News

Your membership also unlocks:

700+ AI Courses
700+ Certifications
Personalized AI Learning Plan
6500+ AI Tools (no Ads)
Daily AI News by job industry (no Ads)