Swiss Re: cat lull is luck, not a license to cut

Two quiet years have a way of erasing a market's memory of tail risk, and Swiss Re is heading into Rendez-Vous de Septembre and the January renewal season with a warning that 2025 and 2026 catastrophe losses ran below trend because the industry was lucky, not because the underlying risk has faded. Urs Baertschi, Swiss Re's CEO of property and casualty reinsurance, said in remarks reported by Insurance Business America that recent performance has been “the luck of the draw”: “last year and also in 2026 we seem to be on the lucky side of things. But that does not reflect the underlying risk landscape.”

Swiss Re Institute estimates that global insured natural catastrophe losses grow at a real, long-term rate of 5% to 7% annually, pushed by rising exposures, higher asset values and shifting hazard patterns; on that trend alone, 2026 losses would land around $148 billion. The tail matters more for January renewals: a peak-loss scenario expected roughly once a decade could reach $320 billion, and a season like 2017's Harvey, Irma and Maria could push annual insured losses above $120 billion without a single record-breaking event.

A premium built on the two most recent loss years rests on a favorable sample; the $120 billion scenario is the more useful underwriting reference because it requires no outlier event, just a run of severe hurricanes in a single season. That kind of correlated year tests how much capital is genuinely available behind the towers being sold this January, and it is the scenario that a decade of below-trend years tends to push out of the discussion.

Baertschi's other message is that capacity alone is no longer the product: “Our clients need more than reinsurance capacity from us, they need risk expertise, data and solutions that help them navigate an increasingly complex environment.” The commodity part of the trade has been priced, and the remaining value sits in deciding which risks to write, at what attachment point, and under what terms—a decision that should push buyers away from rate alone and toward the structure of the coverage.

Europe's 2026 wildfire season is one sign that the composition of risk is changing, and wildfire remains the fastest-growing weather peril globally: insured European wildfire losses have climbed an estimated 8% to 11% annually over recent decades as people and assets move into fire-prone areas, a trend that has continued even as improved modelling and prevention measures have reduced some of the uncertainty around the peril. A hazard growing at that pace carries its own capital charge.

Data centers present the same challenge on a far larger scale: Swiss Re Institute projects cumulative global investment in data centers will exceed $6 trillion by 2030, translating into a related insurance premium opportunity of roughly $91 billion by the end of the decade. That figure sits inside a broader $200 billion premium opportunity across data centers and renewable energy infrastructure combined, while capital spending by the five largest cloud service providers alone is expected to top $600 billion this year.

Growth of that size is not automatic profit for reinsurers, and around 40% of U.S. data center capacity sits in zones with significant-to-very-high tornado risk; as facilities grow larger, their dependence on shared electricity grids, water supplies, technology and digital infrastructure concentrates risk both inside individual sites and across wider networks. That interdependence makes data centers less like a collection of independent properties and more like a correlated exposure, exactly the kind of risk a catastrophe bond or collateralized reinsurance contract needs to measure before pricing the limit.

Swiss Re's argument should push the renewal debate past whether rates rise or fall and into what the rates are paying for. If terms are set by the calm years, the market is effectively betting the next decade looks like the past two, even though the 5% to 7% annual growth trend and a $320 billion tail expected roughly once a decade point the other way. The structures written this January will still be in force when the next season of multiple severe hurricanes arrives.