The rapidly rising demand for insurance programmes for ever larger data cent res is creating new risks for  insurers and those who finance the projects.

New research by Standards & Poor’s has found insurance markets are adjusting to a step‑change in asset scale, as hyperscale data centre campuses can extend into the $20 billion–$50 billion total insured value range during the construction phase alone, well beyond traditional single‑risk benchmarks.

As a result, who bears the residual risk will become increasingly important. As capital investment pushes projects beyond traditional fully insurable frameworks, insurance may no longer be assumed to fully mitigate tail risk. An integrated view across insurance, project finance, and corporate balance sheets will become central to assessing credit risk.

The report added: “Capital spending to build hyperscale data centre campuses has reached tens-of-billions of dollars per site, pushing beyond the levels at which traditional property and construction insurance markets have historically provided full replacement-style coverage at a single location. These constraints primarily apply to physical construction and real asset development. As a result, insurance for data centres is increasingly structured through probable maximum loss (PML) or maximum foreseeable loss (MFL)-based and layered programs that may only cover part of the total project value, leaving a larger share of exposure outside the insurance coverage than has been typical for large-scale infrastructure. By contrast, insurance coverage for high-value computing equipment- particularly GPUs- may present a distinct and potentially more acute challenge, especially as emerging financing structures begin to treat GPU fleets as separately financed assets at scale.

The research found:

• As capital spending to build hyperscale data centre campuses reaches tens of billions of dollars per site, full insurance coverage may no longer be possible when compared with traditional data centres.
• This insurance gap could increasingly function as a capital-structure constraint, where available coverage supports the portion of asset value lenders may view as recoverable in downside scenarios.
• With insurance programs structured through probable maximum loss or maximum foreseeable loss-based and layered programs that may only cover part of the total project value, a larger share of tail risk may be shifting into financial structures.
• As a result, assessing the underlying risk of these mega infrastructure projects is increasingly complex for market participants and may currently be underappreciated, given evolving and non-standard insurance frameworks and the growing migration of tail risk into financial structures for data centre lenders.

S&P added. “Rather than scaling linearly with asset size, coverage is increasingly structured through PML or MFL-based limits, layered placements, and global capacity sourcing, indicating challenges in full limit capacity. PML and MFL differ in the level of potential damage and the assumptions made during the assessment.

“PML assumes partial impairment of safety measures during a loss event, while MFL projects a more severe, worst-case scenario where all safeguards completely fail. Layered placement, consisting of multiple insurers across different layers or tiers, allows each insurer to attach at a specified loss level and exhaust at a defined limit.

“This structure enables access to higher overall coverage limits than any single insurer would typically provide alone.

“Aggregation risk has become a key constraint for insurers, with large campuses concentrating significant insured values at single sites and exposures accumulating across construction and operation phases. This is further complicated by multiphase development, as large data centre campuses are built in phases rather than all at once.

“Coverage must adapt dynamically as new buildings come online, requiring “stacked” limits across construction phases and transitioning into operational property coverage. This evolving exposure profile challenges conventional underwriting frameworks.

“ While overall insurance capital remains substantial, capital deployment insuring hyperscale campuses is becoming more selective, with greater emphasis on risk engineering, transparency of exposure, and portfolio-level accumulation controls, particularly as reinsurers face limited visibility across ceded portfolios.”