SIRC 2025: From capacity to AI - 5 themes insurers should act on now
This year's SIRC felt bigger for a reason. More people, more countries, and sharper conversations with one goal: secure quality capacity ahead of 1/1 while building a smarter, leaner book.
If you didn't make it, here's the signal without the noise - and the moves to make before the window closes.
1) Capacity is back, but it's selective
There is more capacity on the table than last year, especially for well-performing books. But reinsurers are sticking to discipline. Attachment points remain elevated in many property-cat layers, aggregate cover is still scrutinized, and wordings are tighter.
Expect appetite to skew toward clean programs, clearer data, and stable underwriting behavior. Casualty conversations are calmer but still focused on long-tailed loss trends and social inflation.
- What to do: Package a clean story. Show rate adequacy, TIV updates, exposure change controls, and event-level loss performance. Pre-empt with options so partners can say "yes" to something.
- Be ready to trade terms: occurrence vs aggregate, paid reinstatements, or co-participations to land price and structure.
2) 1/1 pricing: adequacy first, retentions hold
Demand is up on higher TIVs, construction inflation, and more realistic PMLs. Pricing is stable to firm on property-cat where results justified it; softening remains selective and data-driven. Retentions are sticking in many programs, and reinstatement terms matter again.
Non-modeled and secondary perils are getting a brighter spotlight: flood, convective storm, and wildfire aggregates will face questions, especially where data is thin.
- What to do: Bring peril-level analytics, not just modeled outputs. Show how you control frequency and severity (deductible strategies, underwriting rules, sub-limits, and aggregation guardrails).
- Have alternatives ready: top-and-drop layers, multi-year talk tracks, and structured quota share/XS combos.
3) AI moves from pilots to production
Underwriting triage, exposure data extraction, and claims fraud checks are moving into daily workflows. The winners aren't chasing shiny tools; they are fixing data pipelines, governance, and user adoption. Regulators are leaning in on fairness, accountability, and auditability for AI in financial services.
If your AI plan can't show measurable lift in quote speed, loss ratio, or expense ratio - it will stall. Keep it small, specific, and traceable.
- What to do: Stand up a simple AI control framework (use cases, approvals, logs, model monitoring, and human oversight). Align with guidance like MAS's FEAT principles for AI in finance for a pragmatic baseline. View FEAT
- Upskill your underwriting, claims, and actuarial teams on prompt quality, review workflows, and data hygiene so tools don't drift into shadow IT. For practical training by role, see AI courses by job.
4) Climate risk goes granular in Asia
Conversations have moved from "climate is a risk" to "show me the asset-level signal." Typhoon tracks, flood depth, and surface water exposure are being tied directly to pricing and reinsurance structures. Clients want clearer terms on secondary perils and business interruption.
Parametric interest is real where cover gaps persist, but basis risk and data sources need a hard look.
- What to do: Refresh hazard maps and vulnerability curves with local data, not just global defaults. Prove how new intelligence actually changes underwriting rules or attachment strategy.
- Use transparent sources to support your view of risk. For physical risk context, see the latest assessments from the IPCC. IPCC reports
5) Retro and alternative capital: useful, but picky
ILS and cat bond interest remains solid, but investors still want clarity on trapped capital, event definition, and reporting cadence. Retro is available, yet terms favor reinsurers with consistent performance and crisp data.
Expect basis risk conversations to get technical fast. Sponsors that invest in clean, frequent reporting and independent validation will win better pricing and faster closes.
- What to do: Standardize event reporting packs now (loss snapshots, exposure change logs, and third-party attestations). Decide which risks truly fit index or parametric structures.
- Pre-wire investor Q&A on model choices, sensitivity tests, and climate adjustments to avoid surprises.
What smart cedants, brokers, and reinsurers are doing before 1/1
- Pre-empts with 2-3 structure options and clear walk-away points.
- Short, factual loss narratives with side-by-side actions taken post-event.
- Underwriter-ready exposure files: consistent schemas, audited TIV updates, and peril tags.
- AI in production for one narrow task (e.g., bordereaux cleanup), tracked by a single KPI.
- Climate addendum: secondary peril view, not just cat model PDFs.
- Retro/ILS playbook with reporting SLAs and independent validation.
Bottom line
Capacity is available for disciplined books. Pricing rewards clarity. AI is useful when it fixes a specific bottleneck - and you can audit it. Come to the table with proof, options, and speed. You'll get the terms you want, and you'll keep them.
Your membership also unlocks:
