In Australia and New Zealand, and across the world, the number of data centres and their capacity is growing rapidly. Cloudscene lists more than 300 data centres in Australia and 80 plus in New Zealand, a snapshot of the speed at which digital infrastructure is scaling across the Tasman. CBRE forecasts Australia’s live data centre capacity to rise from around 1.3–1.4GW in 2025 to about 1.8GW within three years, while still warning demand is likely to outstrip supply - a dynamic that keeps projects moving fast and sometimes ahead of comfortable risk transfer.
From the building phase through to operations, these power-hungry purpose-built facilities with their dense racks of servers, cooling systems, security layers and third-party supply-chains, present brokers with challenging insurance risks. Data centres are high-value, high-dependency assets where BI severity, nat-cat exposure, equipment breakdown and cyber risks can make underwriting tricky and policy wordings far more negotiated than for an ordinary commercial property.
In recent months, that complexity has collided with the challenges of softening market. Against that backdrop, parametric insurance — long talked about, often misunderstood — is increasingly being positioned as an add-on to traditional placement structures, particularly for nat-cat and “non-damage” triggers that can shut down a data centre without leaving a crater in the building.
For Lynn Roehrig (pictured), Descartes Underwriting’s Australia and New Zealand head of business development, the starting point is geography - and the fact that data centres are not one homogenous risk class.
“Natural catastrophe risks vary significantly depending on where data centres are located,” said Roehrig.
On Australia’s eastern seaboard, flood and cyclone exposure is a recurring underwriting theme, while extreme heat can become a loss amplifier by driving higher cooling demand and stressing equipment performance. New Zealand adds its own peril stack - including earthquake - alongside flood and storm risk in key commercial hubs.
The bigger issue for capacity providers, though, is how modern data centre rollouts concentrate exposure.
“Data centres tend to be clustered in the same zones due to fibre connectivity, power availability and access to water for cooling,” said Roehrig.
Hyperscale and edge sites do not appear randomly on a map; they form clusters that follow power availability, fibre routes and cooling constraints - which is exactly how insurers end up with accumulation risk. This clustering can bite in two directions. A single large event can create correlated losses across multiple sites. And even if a facility escapes physical damage, it can be knocked offline by disruption to a critical dependency - grid supply, substation assets, water access, third-party network, or even access roads.
That’s where the most common friction appears between insured expectations and policy response. Traditional BI typically needs physical damage at the insured premises to trigger. But a data centre can face severe losses from a prolonged external power outage, a utility failure, or offsite damage that still forces downtime and SLA credits.
Underwriters say the “best” risks are increasingly differentiated by resilience, not just location: dual utility feeds (or credible alternatives), demonstrable redundancy in cooling (N+1 or better), tested failover and maintenance regimes, modern detection/suppression systems, and clear evidence that operators can isolate faults rather than trip whole halls. For brokers, that engineering story is becoming inseparable from the insurance story - because it’s what can turn a tough nat-cat or outage conversation into a placement that attracts capacity, narrows exclusions, and improves BI terms.
Parametric’s pitch to data centres is not that it replaces property and BI - it’s that it fills the holes: sub-limits, deductibles, waiting periods, excluded perils and the grey area of non-damage triggers.
Roehrig said parametric can also be structured for construction and commissioning risk (including DSU), where delays can cascade through a project’s financing model. It can be designed around coordinates, supplier locations, project timelines and weather-intensity thresholds - effectively putting a pre-agreed value on delay, then paying quickly when an objective trigger is met.
Descartes, he said, is underwriting to meet that demand in the region. “Descartes Underwriting is ready to deploy up to AUD 110m+ in capacity per risk for construction, commissioning and operational phases in Australia, Asia and the Pacific region,” he said.
For operational sites, the argument often lands on SLA exposure - especially for hyperscale contracts where penalties may apply at 99.9% or 99.99% availability thresholds. Downtime can force immediate service credits and create liquidity stress, even before a conventional claim is adjusted and paid.
“While traditional claims take months to settle, parametric insurance pays in days, providing the cash to cover penalties without straining the balance sheet,” said Roehrig.
This is as much a finance problem as an engineering one: data centre ownership structures, including SPVs, can be highly sensitive to timing, cash flow and covenant pressure. Speed of payment becomes a coverage feature.
The market is unlikely to treat every site equally. Nat-cat exposed locations, weaker fire resilience, poor maintenance and single points of failure - like a single grid feed or single cooling path - will continue to face tougher terms and pricing. But as build pipelines expand and insured values climb, parametric is increasingly being used as a pragmatic mechanism to keep transactions bankable and operations resilient - especially when the peril is real, the dependency is external and the need for cash is immediate.
