AI Data Centers Strain Insurers' Capacity and Underwriting

The acceleration of AI data-center construction, much of it financed outside hyperscalers by private equity and private credit, is creating outsized, concentrated insured exposures that are stressing capacity in global insurance markets. Private infrastructure data-center deals have repeatedly topped $10 billion, and brokers say single campuses now represent multibillion-dollar insurable values.

Data centers built for generative AI require denser power and cooling footprints, specialized construction standards and sustained capital through long build-and-ramp cycles. Those technical characteristics increase both the magnitude of potential insured losses (construction, physical damage, business interruption) and the complexity of modelling correlated risks such as grid failure, water shortages and cyber incidents tied to concentrated compute assets.

Insurance brokers told CNBC they are forming dedicated teams and bespoke policies to underwrite large campuses after capacity constraints made it nearly impossible to reasonably insure a $20 billion campus in 2023, according to Tom Harper, data-center lead at a major broker. Preqin and market data cited in reporting show private infrastructure transactions consistently above $10 billion last year. Sector analyses from Allianz and Munich Re highlight construction and utility constraints—Allianz reported 3.2 GW under construction in APAC and 13.3 GW in planning (early 2025 data)—while law firms flag emerging litigation risks when financial stress hits (declining GPU values, construction delays, demand shocks).

Risk modelers, underwriters and ML ops teams must rethink exposure aggregation and scenario design: physical loss modelling must incorporate cascading grid/water risks and longer-tail business-interruption scenarios tied to compute demand cycles. Finance teams and lenders should anticipate covenant and valuation stress if GPU values or demand soften. Claims teams will face novel litigation and allocation questions across insurers, lenders and operators.

insurer capacity allocation decisions, the terms of bespoke policies (sub-limits, aggregation language), utility infrastructure investments in key hubs (e.g., Virginia’s data-center corridors), and litigation outcomes tied to financing disputes. Expect continued product innovation (parametric elements, layered reinsurance and increased coordination between insurers and financiers).