Most P&C Insurers Stuck in AI Pilot Mode as Top 10% Pull Ahead on Revenue and Share Price

Only about 10% of P&C insurers have moved AI beyond pilots to scalable systems, and they achieve higher revenue growth and share price gains than peers, according to Capgemini.
By: | May 14, 2026
AI collaboration concept

The property and casualty insurance industry’s AI ambitions are running well ahead of its results. While 40% of P&C insurance leaders say AI is meeting their expectations, those expectations started low, gains have been largely marginal, and 42% of insurers are not tracking AI metrics at all, according to the World Property and Casualty Insurance Report 2026, published by the Capgemini Research Institute.

The report, based on surveys of more than 2,200 insurance employees, executives and policyholders across 20 markets, found that 60% of carriers remain stuck in exploration or proof-of-concept phases.

The Architecture Mismatch

The report identifies what it calls an “architecture mismatch” — a structural ceiling created by fragmentation across business units and a lack of organizational expertise to drive innovation at scale.

The mismatch manifests across three dimensions simultaneously, the report said.

  • Strategically, only 35% of the top 20 global P&C insurers explicitly link their AI strategy to business outcomes beyond efficiency. Nearly half of AI decision-makers across financial services require a positive return on AI investment within a year, creating tension between short-term pilots and longer-term capability building.
  • Technically, 81% of insurers cited legacy systems and IT architecture as barriers, 74% pointed to data quality and cross-functional accessibility issues, and 61% flagged regulatory and compliance constraints.
  • Organizationally, 67% identified a shortage of AI skills, 55% pointed to unclear ownership of AI initiatives, and another 55% highlighted the absence of clear return on investment.

A spending imbalance compounds these problems. On average, 72% of AI investments go toward technology and infrastructure, with only 28% directed at change management, the report found.

A Small Group Breaks Away

Roughly 10% of P&C insurers — which the report labels “intelligence trailblazers” — have broken through the pilot phase by treating AI as a core operating capability rather than a technology initiative. These organizations demonstrated 21% higher revenue growth and a 51% greater increase in share price over a three-year period compared with mainstream insurers, the report said.

What distinguishes trailblazers is not spending levels but strategic approach. They address strategy and talent, technology, and organizational adoption simultaneously rather than sequentially.

Trailblazers are 3.9 times more likely than mainstream insurers to direct AI investments toward change management, 3.1 times more likely to have developed shared explainable AI infrastructure, and twice as likely to organize cross-functional teams around shared outcome KPIs.

Yet even trailblazers face unresolved challenges. Most AI still operates at the individual task level, despite 49% of employee time being spent on cross-team collaboration. Only 12% of insurers report high maturity in data readiness. And AI has largely been layered on top of existing human-designed workflows rather than prompting process redesign, the report said.

The Employee and Customer Gap

The consequences of stalled transformation are visible on both sides of the insurance transaction. Among employees, 47% of those with access to AI tools report that their workday remains unchanged even after 18 months, the report found. Meanwhile, 43% of employees cite job security as a top concern about AI, and 25% worry the transition will increase rather than reduce their workload.

On the customer side, 86% of policyholders still describe their relationship with insurers as reactive and transactional, even as 51% say they are very comfortable with insurers using AI to provide proactive services. Policyholders want capabilities such as usage-based pricing, cited by 46% of respondents, personalized advice at 38%, and risk prevention alerts at 21%.

The report describes what it calls an “expert-centric” organizational model built around four interconnected elements: leadership that defines human-AI boundaries, human experts who set standards for when synthetic execution can be trusted, a synthetic execution layer handling high-volume routine work, and orchestration managers who translate business strategy into AI governance principles.

“The gap between these intelligence trailblazer insurers and most other organizations isn’t just a technology problem,” the report said. “It’s about organizational redesign decisions yet to be made.”

Obtain the full report here.

The R&I Editorial Team can be reached at [email protected].

More from Risk & Insurance