AI Investment Surges in Insurance, But ROI Questions Persist
Insurtech funding snapped back in 2025, reaching $5.08 billion. Two-thirds of that total flowed to AI-focused companies. Momentum peaked in Q4 with $1.68 billion deployed across nearly 230 AI-centric deals, the strongest quarter since 2022 and involving 100+ companies.
It's the biggest year yet for AI in insurance. But the message from the market is clear: separate momentum from money-making. Efficiency that doesn't show up in the combined ratio, growth, or cash flow is a distraction.
Key 2025 Numbers at a Glance
- Total funding: $5.08B (+19.5% YoY)
- AI share: ~66% of all insurtech dollars
- Property & casualty: $3.49B (+34.9%)
- Mega-rounds: 11 deals (up from 6); five $100M+ rounds including CyberCube ($180M) and ICEYE ($174.81M, Series E)
- Geography: U.S. captured 55.74% of global deals (+5.16 p.p. vs. 2024); Silicon Valley's share jumped from 8.72% to 16.12%
Where the Money Landed
P&C insurtechs drew the lion's share, helped by larger rounds and a risk environment that rewards precision: better triage, faster claims, tighter pricing, stronger cyber and property analytics. The center of gravity intensified in the U.S., with Silicon Valley taking a much bigger slice of deal flow.
The headline: capital is available again, especially for AI that targets core insurance levers-loss costs, expense ratio, and profitable growth.
The ROI Paradox Every Carrier Feels
Gallagher Re flags a trap many teams are seeing firsthand: tools free up time, but the org doesn't redeploy that time to high-value work. You get efficiency on paper without a lift in productivity, profitability, or customer impact. Meanwhile, big tech poured more than $1 trillion into AI and data centers in 2025, and valuations outpaced revenue. Bubble risk questions aren't going away.
Set expectations on an 18-month window. If an AI initiative can't prove material value by then, reset or retire it.
A Simple Way to Judge AI Bets
- Product-level: Does this improve underwriting accuracy, claim cycle time, fraud catch rate, quote-to-bind, or CX? What's the before/after?
- Company-level: How does it move loss ratio, LAE, expense per policy, premium per FTE, renewal rate, or cash conversion?
- Industry-level: If everyone adopts it, where does durable advantage remain? Data moats? Distribution? Capital access?
Operator Playbook: Turn Efficiency into Profit
- Define the money metric first. Tie each pilot to one primary P&L outcome (e.g., -1.0 pt combined ratio in a line of business).
- Stage-gate your roadmap. Clear 30/60/90-day checkpoints with hard kill-switches if KPIs miss.
- Redeploy freed capacity on purpose. Move saved hours into higher-touch claims, producer enablement, and renewal retention tasks.
- Prioritize near-term wins. Claims routing/auto-adjudication, FNOL intake, subrogation, SIU triage, pricing segmentation, and producer quoting APIs.
- Vendor diligence beyond demos. Data rights, model lineage, security posture, indemnities, and exit terms; insist on loss ratio impact proofs, not vanity metrics.
- Data governance matters. Map consent, usage rights, retention, and audit trails-especially for health data and wearables. See HIPAA guidance.
- Control compute costs. Set usage caps, batch non-urgent inference, cache results, and track unit economics per prediction or automated claim.
- Backtesting is non-negotiable. Validate models across cycles and cat events; align with reinsurance and capital teams before scaling.
Life, Accident & Health: Data Streams to Watch
Wearables: 44% of Americans now use devices tracking sleep, heart rate, and activity. Firms like dacadoo, HealthIQ, and Lapetus aggregate signals for underwriters to refine risk classes and pricing. The practical lift: more precise segmentation and dynamic engagement programs-if consent, data quality, and behavioral follow-through are handled well.
Electronic health records (EHR): Hospital investments created rich population data. Insurtechs such as Qrvey and Human API help carriers interpret it for underwriting and product design. Expect faster evidence gathering and fewer manual reviews-provided governance and clinical validation are tight.
Genomics: Companies like FOXO Technologies analyze molecular markers to estimate mortality risk and disease susceptibility. Potential upside: price known conditions more accurately and shift the focus to preventative behaviors. Proceed with strict compliance, ethics reviews, and transparent customer consent.
What This Means for Insurance Leaders
Capital is back, but scrutiny is higher. Winners will connect AI directly to underwriting precision, loss-cost control, and distributor productivity-and prove it with numbers. Set a short runway, measure aggressively, and reallocate time you save into work that compounds.
Want to upskill your team for practical, ROI-focused AI work? Explore AI courses by job function to accelerate adoption and governance without the hype.
Your membership also unlocks:
