January 7, 2025. The Palisades and Eaton wildfires sweep through Los Angeles. $40 billion in insured losses — the largest wildfire event in recorded history. No major US hurricane made landfall all year. And yet global insured catastrophe losses still hit $107 billion, with secondary perils driving a record 92% of the total. The protection gap widened to $424 billion. Three quarters of global exposure is uninsured. Swiss Re's peak scenario for 2026: $320 billion. Wildfire growing at 12% annually. The insurance industry is being remade by fire. And the question it cannot avoid: at what point does risk become uninsurable?
Not investment advice. Data sourced from Swiss Re Institute sigma 1/2026 "Natural catastrophes in 2025: the persistent rise of wildfire and storm risk" (March 2026), Swiss Re Natural Catastrophe Insurance Resilience Index (June 2026), Munich Re NatCat 2025 Analysis, Gallagher Re 2025 Annual Review, Artemis cat bond data June 2026. All figures current as of June 2026.
The insurance industry has always priced catastrophe risk around two reference events: the 1906 San Francisco earthquake and Hurricane Andrew in 1992. These were the events that calibrated how much capital reinsurers needed to hold, how much primary insurers needed to charge, and what the upper bound of a single-event loss might look like. The Palisades and Eaton fires of January 2025 did not just exceed those reference points. They rendered the entire framework inadequate.
The fires started in Los Angeles County on January 7, 2025, driven by extreme Santa Ana winds and drought conditions that had reduced Southern California's vegetation to tinder. By the time containment was achieved, approximately 18,000 structures had been destroyed — nearly three times the recent California average annual loss — in neighbourhoods where the median home price exceeded $2 million. The combined insured loss: $40 billion. The largest wildfire event in sigma records. Achieved in a county where 12% of residential properties carry wildfire coverage — and where the major insurers had been withdrawing from the market for three years before the fires arrived.
The withdrawal pattern is the critical context. Allstate stopped writing new California homeowners policies in May 2023. State Farm followed in June 2023. Farmers Insurance capped its California exposure. AIG, Chubb, and others either exited or dramatically reduced their presence. By January 2025, the California FAIR Plan — the insurer of last resort, designed for properties the private market won't cover — had become the primary homeowners insurer in parts of Pacific Palisades. It was not capitalised to absorb a $40 billion event. The catastrophe that the insurers had predicted would happen eventually happened — in the market they had already left, to the policyholders who had no alternatives.
The architecture of catastrophe insurance and reinsurance was built around primary perils — the large, infrequent, well-modelled events. Hurricanes. Earthquakes. These are the events that catastrophe models have been calibrated to price over decades. The industry knew how to manage them: charge premiums that reflect the long-run expected loss, hold sufficient capital for a 1-in-100 or 1-in-250 year event, and reinsure the tail risk through the global reinsurance market.
Secondary perils were considered manageable — frequent enough to be diversifiable, small enough individually to be absorbed by primary insurers without significant reinsurance. Wildfires, severe convective storms, floods: these were the attritional losses, budgeted for, not feared. That framework is now broken. Secondary perils drove 92% of insured losses in 2025 — a record. They have driven above-trend losses for three consecutive years. Swiss Re has documented that insured losses from North American wildfires and European severe convective storms are growing roughly twice as fast as exposure alone would predict. The models underestimated the hazard. They underestimated the vulnerability. And they significantly underestimated the concentration of high-value assets in harm's way.
The fastest-growing peril in insurance history. Insured losses growing at 12% annually since 1990 — twice the rate of exposure growth. One in three Californians lives in the wildland-urban interface (WUI). Exposure in high-risk WUI zones has grown 1.9× faster than non-WUI areas in California since 1990. Reconstruction costs still 37% above pre-COVID levels. Europe's wildfire season is intensifying — Spain, Italy, Greece seeing record fire seasons — but has yet to generate a billion-dollar insured loss because of lower insurance penetration in at-risk areas.
Hailstorms, tornadoes, and damaging winds — the new attritional catastrophe. $51 billion in 2025, the third-costliest year for SCS after 2023 and 2024. US urban expansion in hail-prone regions combined with higher reconstruction costs producing elevated losses. Europe also experiencing intense hail seasons — a Brisbane hailstorm caused $1.8 billion in insured losses alone. SCS losses have exceeded $50 billion annually for three consecutive years — a level that was historically a major outlier year.
The peril with the largest protection gap by ratio. $3.4 billion insured against $15.4 billion economic loss — a gap that reflects the chronic underinsurance of flood risk globally. The Myanmar earthquake in March 2025 illustrated the extreme version: $11 billion in economic losses, $200 million insured. Southeast Asia flood season in 2025 caused $11 billion in economic damage across Thailand, Indonesia, and Malaysia — the vast majority uninsured. Germany's flood insurance penetration, at 57%, is held up as a global success story.
The most important number in 2025 insurance is a zero. No major US hurricane made landfall. Swiss Re explicitly characterised the below-trend total of $107 billion as "the result of favourable variability rather than any easing of underlying risk." The long-term trend implies $140 billion. A single major hurricane landfall on Miami or Tampa — cities that have grown enormously since the last major strikes — would generate insured losses well in excess of any historical event. Swiss Re's peak scenario for 2026: $320 billion. The question is not if but when.
Swiss Re Natural Catastrophe Insurance Resilience Index · June 2026 · Premium-equivalent terms
The mathematics of the protection gap are sobering. The global insurance resilience index — the share of natural catastrophe protection needs covered by insurance — is 27% in 2025, up from 25% in 2015. In ten years, the industry has closed the gap by 2 percentage points. At that rate, it would take another century to reach 50% global coverage. Meanwhile, as Swiss Re notes, "in absolute terms, the protection gap continues to grow, as there is simply more to protect." Economic growth, urbanisation, and asset value inflation are all expanding the denominator faster than insurance penetration is growing the numerator. The gap in premium-equivalent terms is $424 billion annually — the premium that would need to be collected to fully cover expected losses. That is more than the entire gross written premium of the US non-life insurance market.
Catastrophe bonds — securities that pay coupon returns to investors unless a defined catastrophe event triggers, at which point investors lose principal to cover insured losses — are the most elegant financial innovation in insurance of the past 30 years. They transfer catastrophe risk from reinsurers onto capital markets, dramatically expanding the capacity available to absorb large events. Cat bond issuance hit $16.1 billion year-to-date in 2026 — on pace for another record year. The market has grown from essentially zero in 1994 to over $50 billion in outstanding risk capacity.
The cat bond market proved itself definitively during the LA wildfires — structures designed to cover California wildfire risk triggered as intended, capital markets absorbed the losses, and reinsurers were able to manage their net exposure. Mexico doubled its parametric catastrophe insurance programme to $575 million at its 2026 renewal. Zenkyoren is sponsoring a $100 million Japan earthquake cat bond. ILS (Insurance-Linked Securities) as a broader category — including collateralised reinsurance, industry loss warranties, and sidecars — have created a permanent expansion of catastrophe risk capacity that is counter-cyclical to traditional reinsurance capital.
Traditional insurance requires damage assessment after a loss — an inspector visits, estimates damage, and pays out. In a catastrophe, this process takes months, by which time communities need the money immediately. Parametric insurance pays automatically when a defined parameter — wind speed, earthquake magnitude, rainfall level, temperature — exceeds a threshold, without requiring damage assessment. Payment is immediate, transparent, and requires no claims adjustment.
Parametric insurance is transforming coverage in emerging markets where traditional insurance infrastructure doesn't exist. The Kenya Livestock Insurance Programme pays farmers within two weeks of a drought trigger — before the animals die. Mexico's parametric catastrophe programme covers government emergency response costs. Caribbean nations use hurricane parametric covers to fund immediate disaster response. The Caribbean Catastrophe Risk Insurance Facility (CCRIF) paid out within 14 days of Hurricane Maria in 2017 — while traditional insurance claims were still being assessed. Parametric is not a replacement for traditional insurance — it is the only viable insurance structure for risks and populations that traditional indemnity insurance cannot reach.
The fundamental problem with catastrophe modelling is that historical data underestimates future risk in a changing climate. Models calibrated on 30 years of loss history are being used to price risk driven by conditions that have no historical precedent. AI and satellite data are enabling a new generation of risk models that use current physical conditions — real-time vegetation moisture, soil saturation, urban heat island effects, building materials from satellite imagery — rather than historical loss patterns.
Zürich Insurance has launched a data centre insurance product using satellite monitoring of cooling infrastructure risk. Jupiter Intelligence provides climate risk data to insurers, banks, and governments. Cape Analytics uses computer vision to assess property condition from aerial imagery — enabling more accurate underwriting without physical inspection. Floodbase provides parametric flood triggers from satellite altimetry. The companies building the data and modelling infrastructure for climate-adaptive insurance are the picks-and-shovels play in the insurance technology space. In 2026, AI is informing pricing, risk selection, fraud detection, claims payments, and reserving at the leading carriers — but only 30% of insurers have deployed generative AI in production, according to industry surveys.
Swiss Re's analysis of US adaptation projects found a median benefit-cost ratio of 1.86 — nearly $2 of avoided loss for every $1 invested. Cedar Rapids urban flood defences: 1.2× benefit-cost ratio. Middle Rio Grande levees: 9.63×. Germany's flood insurance penetration reaching 57% (from 20% two decades ago) has not just increased insurance take-up — it has funded flood defence investment that reduced losses. Adaptation measures — updated building codes, improved land-use planning, physical hardening of properties, controlled burns to reduce wildfire fuel loads — are the only mechanism that reduces the underlying risk rather than transferring it.
The investment gap is enormous. US federal adaptation spending is a fraction of annual catastrophe losses. California's investment in prescribed burns and defensible space around the WUI is expanding but still inadequate relative to the risk. Europe's Adaptation Strategy is the most comprehensive policy framework — but implementation varies dramatically by country. The reinsurance industry is increasingly engaging with governments on adaptation not out of altruism but out of commercial necessity: if the underlying risk continues to grow faster than pricing adjustments, some risks genuinely become uninsurable at any premium that policyholders can afford.
The world's leading catastrophe reinsurer and the primary source of global nat cat loss data. Swiss Re's sigma reports are the industry bible. P&C returns on equity of 18–19% projected for 2025 — the highest in two decades. Leads the industry on climate risk modelling and adaptation investment advocacy. The hardening reinsurance market after 2017–2022 losses produced exceptional profitability at the cost of tighter coverage terms.
Co-equal with Swiss Re in scale and scientific approach. Thomas Blunck's January 2025 statement — "traditional assumptions no longer hold" — is the most consequential public statement any reinsurer made about climate risk. Munich Re's proprietary NatCatSERVICE database tracks every catastrophe event globally. The company that prices climate risk most accurately has the greatest long-term competitive advantage in a world where climate risk is growing.
Buffett famously exited Florida hurricane reinsurance in 2021, citing inadequate pricing. He returned in 2023 when pricing had improved sufficiently. The lesson: Berkshire's disciplined capital deployment — only writing risk when the price is right, regardless of market pressure — is the template for risk capital management. Its balance sheet strength makes it the counterparty of choice in peak-event scenarios when other reinsurers face capital constraints.
The world's oldest insurance market and the primary venue for specialty risk. Lloyd's syndicates are the most agile underwriters of emerging and unusual risks — cyber, satellite, pandemic, climate litigation. Lloyd's Future at Lloyd's strategy is digitising the market. 2025 performance was strong — the market outperformed after years of remediation under CEO John Neal (now under leadership scrutiny after probe disclosed). Cat bond issuance funnels significant ILS capacity through Lloyd's structures.
The primary insurance layer that faces policyholders directly. Chubb, AIG, and Zurich are the three largest global commercial insurers. All three have been managing California wildfire exposure — reducing limits, adding exclusions, repricing. Chubb has been the most disciplined — exiting unprofitably priced risks before events rather than after. The primary market repricing of 2022–2025 was led by these three, who raised commercial property rates 20%+ in consecutive years.
The Bermuda reinsurance market — capitalised after Hurricane Andrew in 1992 and rebuilt after 9/11 and Katrina — is where catastrophe risk expertise lives. RenaissanceRe's proprietary risk models and capital efficiency have made it one of the best-performing reinsurers over two decades. Everest's $530 million Kilimanjaro III Re retro cat bond in 2026 demonstrates the ILS market deepening. Gallagher Re projects P&C ROE of 18–19% for these specialists in 2025.
Some risks are becoming genuinely uninsurable at premiums that make economic sense — and the industry is not admitting this clearly enough. When Allstate, State Farm, and Farmers exit California, they are not making a temporary commercial decision pending better pricing. They are making a structural assessment that the risk is growing faster than any actuarially sound premium could absorb while remaining affordable to the households who need the coverage. The California FAIR Plan — the insurer of last resort — is not a solution. It is a fiscal time bomb. When the next major wildfire hits, the FAIR Plan will be insolvent, requiring either a state bailout or policyholder assessments that collapse the coverage market entirely. The honest acknowledgement that some areas cannot be insured at affordable premiums is the prerequisite for the honest conversation about whether those areas should be built in at all.
The $424 billion protection gap is not primarily a market failure — it is a price and affordability failure. Insurance is available in most markets. It is simply unaffordable for the majority of the exposed population. In Bangladesh, flood insurance exists. It costs more than a year's income for the farmers most at risk. In the Philippines, typhoon coverage is available. Most smallholders cannot afford it. The protection gap is not closed by better products or more sophisticated models — it is closed by reducing the premium through risk reduction (adaptation), government subsidy (catastrophe pools), or parametric structures that reduce administrative costs. The industry's commercial incentives push it toward insuring the insurable and repricing away from the uninsurable. But the people most harmed by the protection gap are those with the least political and economic power to advocate for solutions.
The 2025 reinsurance profitability record — 18–19% ROE — is partly a function of the underwriting discipline that produced adequate pricing, and partly a function of not having a major hurricane year. Gallagher Re projects P&C returns on equity of 18–19% for 2025. This looks like a strong industry. It is — for one more year without a major event. A $320 billion year — Swiss Re's peak scenario for 2026 — would test the capitalisation of the entire reinsurance market simultaneously. The Florida homeowners market is already structurally impaired. The California wildfire market is de facto a state insurance system. A major hurricane season on top of ongoing wildfire and SCS losses would reveal whether the post-2017 restructuring of reinsurance pricing and terms was sufficient to maintain solvency through a genuine peak-loss year. The industry's current profitability is the profit of a market that got lucky in 2025. Swiss Re said exactly this. The question for 2026 is whether that luck continues.
Long-horizon thinking on capital, technology, and the forces shaping the next decade of wealth creation. Written from first principles. Not consensus. Not noise.