S04.6 · Insurance
A ~$715B global capital pool assuming ceded risk from primary insurers, now softening as capacity outpaces cedent demand.
Reinsurers assume risk ceded by primary insurers to spread portfolio exposure, a market with roughly $715B in dedicated global capital as of 2025E (Aon/Guy Carpenter estimates), of which third-party and insurance-linked-securities (ILS) capital alone is projected at $114B in 2025, up 7% (AM Best/Guy Carpenter). Unlike primary insurance, this is a genuinely concentrated underwriting layer — the top ten reinsurers write the large majority of global ceded premium. The cycle position matters more than the structure right now: pricing is softening at the January and midyear 2025-26 renewals because capacity has outpaced demand following the post-2023 hard-market capital rebuild, and soft pricing on extreme capital intensity is where underwriting discipline gets tested.
Capital grew steadily from 2023 through 2025 as the sector rebuilt following a hard market, and softening pricing is the arithmetic consequence of that inflow outrunning cedent demand (Guy Carpenter, Insurance Journal). Bermuda, Europe — Munich Re, Swiss Re and Hannover Re among the largest — and London/Lloyd's dominate underwriting capacity. Cedents are global, but catastrophe-exposed US and Asia-Pacific business is what actually drives the pricing cycle.
Upstream, retrocession and ILS/catastrophe-bond capital backs reinsurers' own risk; downstream, primary carriers cede business through reinsurance brokers such as Guy Carpenter, Aon Re and Howden Re. Pricing is highly cyclical, and combined ratios swing with catastrophe loss years rather than tracking a trend. Capital intensity is extreme. Retail regulatory protection is minimal, since the counterparties are sophisticated insurers rather than consumers, but rating-agency scrutiny of capital adequacy is heavy — a reinsurer's rating is itself a commercial asset, because cedents will not place meaningful volume with a counterparty below a given financial-strength threshold regardless of price. The rating is the license to trade.
The capital base is being rebuilt from new sources. ILS and catastrophe bonds are an expansionary adjacency, pulling in capital-markets money that would not otherwise underwrite insurance risk directly — genuinely new capital, not recycled premium. Life reinsurance and longevity risk transfer extend the model into mortality and longevity exposure. MGA fronting — providing rated paper for a delegated underwriter without retaining the risk — is the defensive move, a response to capital-markets investors seeking more direct access to underwriting economics rather than paying a reinsurer's margin for it.
Agentic AI automates catastrophe-model aggregation and treaty exposure analysis, compressing the actuarial and analytics headcount cost line that has historically supported treaty pricing. It has not touched the segment's real moat, which is capital allocation and cedent relationships built over decades, not analytical throughput. Analytics automation is already underway; AI-priced parametric retrocession is 5-10 years from meaningful scale, limited less by modeling capability than by the scarcity of tail-risk loss data needed to train and validate a pricing model with confidence.