McKinsey Puts Gen-AI's Insurance Prize at $50 to $70 Billion: The Soft-Market Catch

McKinsey estimates generative AI could unlock $50 billion to $70 billion of insurance industry revenue, concentrated in marketing and sales, customer operations and software engineering. It sets that figure against two decades in which premiums compounded at 4.9% annually while pretax profit grew only 4.3% (McKinsey, August 2026).

Whether that gain reaches the loss ratio or gets competed back to policyholders is a pricing question, not a technology one.

Key Takeaways

1.5 to 3.0 points of technical-result improvement is what more than 50 leaders from Europe's largest insurer groups told McKinsey gen AI could deliver, alongside 10% to 20% productivity gains.
A 60-basis-point gap between 4.9% premium growth and 4.3% profit growth, compounded since 2005, is the stalemate the estimate is meant to break: $8.3 trillion of premium producing roughly $580 billion of pretax profit.
27% to 32% has been the P&C expense ratio band for two decades, with a 10-point spread between leaders below 22% and laggards at 32%. That spread, not the average, is where the revenue estimate sits.
Commercial P&C premiums fell 1.2% in the first quarter of 2026, the first quarterly decline since 2017. A softening market is the condition under which a shared efficiency gain becomes price rather than margin.

Where the $50 Billion to $70 Billion Sits

The report frames the opportunity against four structural problems it dates to 2005: fading relevance, high distribution cost, flat productivity and a slow pace of change. Relevance shows up as protection gaps that widened even through digitization, a $133 billion natural catastrophe gap in 2025 and a cyber gap McKinsey puts near $900 billion, with less than 1% of global cyber costs currently insured.

Distribution is structural rather than cyclical. Roughly 85% of U.S. property and casualty premium and 95% of life premium still moves through agents, brokers and managing general agents, channels that consume 10 to 25 cents of every P&C premium dollar and up to 80 cents of first-year life premium.

Productivity has the longest paper trail, and it is where the estimate actually lives. McKinsey puts the average P&C expense ratio at 27% to 32% since 2005, essentially flat, with a 10-point gap separating leaders below 22% from laggards stuck at 32%. Marketing and sales, customer operations and software engineering, the three functions the estimate names, are exactly the cost centers that decide which side of that divide a carrier sits on.

The top-line arithmetic supplies the urgency. Global gross written premium reached an estimated $8.3 trillion in 2025, growing about 4.9% annually since 2005, while pretax profit grew roughly 4.3% to about $580 billion. Personal lines premium has fallen as a share of global GDP, from 1.2% in 2019 to 1% in 2023 (Reinsurance News).

Why a Shared Gain Lands in Rate, Not Margin

The technical-result estimate is a partial-equilibrium number. It models what one carrier gains while its competitors hold still, which is the right simplification for a single firm's business case and the wrong one for an industry. Insurance pricing clears like a market where dozens of rated, regulated competitors quote the same risk and the buyer can shop the difference.

That is where the actuarial mechanism bites. An indicated rate change is set off credibility-weighted loss experience and current trend, not off a projected productivity gain. Once enough of the market's loss experience reflects AI-assisted underwriting, the indication converges lower for everyone, whether or not any individual carrier intended to hand the margin back. The 1.5 to 3.0 points the survey respondents expect to keep is the exact quantity a softening market gives away first.

The 2026 pricing record is the natural experiment. Commercial premiums fell 1.2% in the first quarter, with commercial property down 5.5%, workers' compensation down 3.7% and cyber liability down 3.5%, and brokers citing expanded underwriting capacity and more aggressive competition for new and renewal business (Council of Insurance Agents & Brokers). Personal auto shows it from the other side: a 94.4 net combined ratio in 2025, the strongest in years, with premium growth slowing to 3.6%, the weakest since 2020 (Insurance Information Institute).

Cyber is furthest along the same cycle. Global cyber rates fell for a 12th consecutive quarter through the second quarter of 2026, down 4% year over year, while average ransom demands climbed 47% to more than $1 million (cyber rate adequacy and loss-cost trend). Cyber underwriting has had the deepest AI-assisted analytics penetration of any commercial line, the head start McKinsey's framework predicts should produce the largest technical-result gain. Filed rate decoupled from loss-cost trend toward price instead.

The Build-Versus-Buy Version of the Same Problem

Aviva is the counter-case, and it is worth stating plainly. More than 80 AI models across its UK book cut liability assessment time by 23 days, improved claims routing accuracy by 30% and reduced customer complaints by 65%, for roughly £60 million, about $82 million, of savings in 2024 alone (Risk & Insurance). That is a disclosed carrier result rather than a survey response, and it supports McKinsey's finding that AI leaders generated 6.1 times the total shareholder return of laggards over five years.

Read against the pricing data, that 6.1-times spread measures how long a first-mover window stays open rather than a permanent structural advantage. BCG's 2026 AI Radar found P&C carriers have tripled AI spending as a share of revenue while only 38% generate measurable value at scale (coverage). Morgan Stanley projects a 200-basis-point expense ratio reduction and $9.3 billion of industry operating income by 2030, but only after a net negative $2.4 billion in 2026 as implementation cost outruns realized savings (stress test).

The consequence for a pricing actuary is a build-versus-buy question rather than a technology one. A carrier licensing the same off-the-shelf underwriting or claims-triage tool its competitors license reaches the 10-to-20% productivity gain fastest and commoditizes fastest, because every licensee of the same vendor model books the same improvement on the same timeline. That is the precise condition under which the gain converts to price. A rate plan built on a durable AI-driven technical-result advantage is a bet on vendor commoditization that the 2026 cyber and personal auto data does not support.

McKinsey Puts Gen-AI's Insurance Prize at $50 to $70 Billion: The Soft-Market Catch

Key Takeaways

Where the $50 Billion to $70 Billion Sits

Why a Shared Gain Lands in Rate, Not Margin

The Build-Versus-Buy Version of the Same Problem

Further Reading

Sources