Insurance: AI Investments Emerge (So Far) as Qualitative Comments vs Quantitative Drivers

The insurance industry’s investments in AI and the emergence of results are now regularly referenced on conference calls. Over the last few weeks, numerous P&C and life 2Q earnings calls include comments such as "contributing to margin improvement" and "a minor fraction of efficiency gains" and other qualitative management assertions. While comments have become regular, quantitative backing with financial figures has been absent. 

Seven insurers have addressed AI on Q2 2026 calls this cycle, but none have disclosed isolated dollar or bps figures that tie AI investments to reported results. This omission does not indicate AI isn't working inside these organizations. Instead, it indicates insurance companies haven’t yet disclosed evidence that would allow an outside analyst to size the investment or the return. The gap between what may be happening inside companies is not a verdict on ROI but is what emerged this week. 

Examples from the calls appear below and highlight a common thread across all seven companies: management comments are more positioning than proof. Companies reference AI as a productivity or efficiency lever, but even where a ratio figure entered the conversation the AI benefit was mixed up in other numbers (Travelers figures shown below), explicitly confounded by a non-AI driver, or just absent entirely. Chubb comes closest to tying performance with AI with its reference to tracking "hard dollars", but it still stops short of a disclosed number. 

Taken together, the pattern across all seven calls below is consistent. Management is confident enough in AI's contribution to raise it unprompted but is not yet willing or able to isolate its economic effect from the innovation, pricing, and integration activity it sits alongside. Only Travelers produced a ratio tied even partly to AI, and we show that figure requires calculation from the supplement rather than citing a disclosed number. Hamilton's cat-loss and reserve driven combined ratio deterioration this same quarter is a reminder that AI framing hasn't insulated any of these carriers from underwriting volatility.

Travelers: Q2 2026 call (7/17/26)

Management pointed to continued build-out of Travis, its digital quoting and distribution platform, and to AI's role in straight-through claims processing as part of a broader "innovation front". Pressed directly on whether these initiatives show up in margin, management confirmed AI and innovation investments contributed to the 0.5-point improvement in the business insurance underlying loss ratio, citing better risk selection and operating leverage. That figure is real, but it's a joint attribution; AI's discrete share within the 0.5 points isn't disclosed. 

Source: Travelers Supplemental Filings for period end 30-Jun-2026 

Chubb: Q2 2026 call (7/22/26)

Asked whether rising AI token costs were eroding efficiency gains, CEO Evan Greenberg said Chubb's token costs are "a fraction, a minor fraction" relative to the efficiencies, insights, and improvements gained. He added that Chubb tracks the comparison in hard dollars internally, but no dollar or bps figure for the efficiency side was shared. Interestingly, Greenberg separately framed Chubb's technology, data, and scale as a structural, secular advantage that itself, distinct from cost savings, was proposed by Chubb as an additional advantage.

MetLife: Q2 2026 call (8/6/26)

Direct expense ratio rose to 12.1% in Q2 2026 from 11.7% a year earlier. Management attributed that move to absorbing roughly 50 basis points of PineBridge integration costs, not to AI investment and expenses. CEO Michel Khalaf separately described AI as becoming a structural advantage given MetLife's policy and claims volume and added that AI investment was a "durable efficiency tailwind".

American Financial Group: Q2 2026 call (8/4/26) 

AFG said on its most recent earnings call that AI deployment has been decentralized across 36 business units and cited crop insurance (a data-intensive line) as furthest along in terms of business submission, document intelligence, and claims workflow tools. Management described "encouraging productivity improvements", particularly in claims, but no figures were provided. 

Selective: Q2 2026 call (~7/24–25/26)

Management said it is leveraging its "tools, granular insights, and differentiated operating model" to drive higher retention on its best-performing business while meaningfully lowering retention on underperforming business through rating actions. Retention in the worst-performing cohorts fell from 81% to 55%, with the renewal rate rising from 11.5% to 18%. Separately, Selective's Q2 2026 10-Q describes broader use of agentic AI in claims, underwriting, and risk management, citing early gains in accuracy, speed, and productivity. But that AI language is general and company-wide, not tied to the retention-targeting mechanism described on the call.

AIG: Q2 2026 call (8/7/26)

AIG cited the scaling of "Underwriting by AIG Assist" and "Claims by AIG Assist" to improve productivity and distribution but did not attribute any ratio move to these tools. Separately, without providing figures, AIG cited data center insurance for AI-related projects as a strategic growth initiative; a distinct business line insuring AI infrastructure, not internal AI use.

Hamilton Insurance Group: Q2 2026 call (date not confirmed)

CEO Pina Albo described AI investment as a "productivity and intelligence multiplier" helping underwriters and claims professionals process submissions and extract data, accelerating intake and improving data quality. It is a descriptive claim with no productivity, ratio, or volume figures attached. The same quarter, Hamilton's combined ratio deteriorated to 95.0% from 86.8% a year earlier, driven by roughly $50 million in catastrophe losses (primarily Middle East conflict related) and a separate $16 million reserve charge on certain Bermuda casualty lines. The two are unrelated developments, not one blended cause, but together they're a reminder that AI framing this quarter didn't correspond to or insulate against underwriting volatility.

Conclusion

For investors tracking insurance AI investment as an efficiency lever, the gap between commentary and data was evident on calls this week. Underwriting and expense-ratio forecasts increasingly assume AI-driven efficiency gains will show up in the numbers. However, until carriers disclose isolated, ratio-level figures rather than qualitative framing, investors have no reliable way to verify that a given quarter's results reflect genuine AI-driven efficiency rather than favorable loss experience, reserve releases, or the impact of reinsurance purchases. 

Statistical supplement 

FactSet Fundamentals 

The Travelers Companies, Inc. Financial Supplement - Second Quarter 2026

Macro Drivers and Impact on Insurance Earnings

Macro summary: Cooling producer prices will help P&C companies. This is an early, favorable signal for easing claims-cost inflation. The improving jobless rate supports life companies. specifically workers' comp premium bases and life/group-benefits persistency. Elevated equity levels continue to support AUM-linked fee income, though the repeated correction-and-recovery pattern remains a volatility signal for life carriers' hedging costs. Payroll and claims data point to a labor market that's decelerating but not distressed, with limited read through risk to either segment's exposure base.

Insurance earnings implications: Producer prices fell 0.3% in June, the largest goods-driven drop since mid-2022, while unemployment eased to 4.2% from November's 4.5% peak. Equities (S&P 500 at 7,709.96, Nasdaq at 26,348.35) sit near highs after two corrections over the past year. Payrolls grew 57,000 in June, and jobless claims held at 199,000, both near recent-average, decelerating-but-stable levels.

Our Macro Tracker table below lists key economic data and the potential impact on insurance company earnings. The right-hand column identifies the specific, potential impacts on company earnings. 

Final demand PPI was unchanged in July (seasonally adjusted), following a 0.3% decline in June, pulling the year-over-year rate down to 4.7% (unadjusted) from 5.5% in June. It's the clearest sign yet that producer-level cost pressure is easing alongside consumer prices. 

Earnings implications: PPI is the more direct proxy for auto and property repair/replacement cost inflation in P&C reserving. The drop to 4.7% year-over-year paired with a second flat monthly print supports a genuine easing in the loss-cost trend rather than a one-month blip, a modest tailwind for combined ratios. 

Earnings implications: Record-adjacent equity markets lift variable annuity and separate account fee income and AUM-based revenue for life insurers. For P&C, the benefit is mainly indirect through investment portfolio total return on equity and alternative asset allocations.

The unemployment rate edged down to 4.1% in July from 4.2% in June, even as payroll growth stalled, a divergence driven by a modest pullback in labor force participation rather than hiring strength. 

Earnings implications: A still-low unemployment rate supports persistency on group life and disability books for life insurers; the linkage to P&C is indirect and is not a meaningful earnings driver this period. 

Nonfarm payrolls fell by 23,000 in July, the weakest print since early 2026 and well below the prior six-month average, even as revisions added a combined 93,000 jobs to March and April. 

Earnings implications: Slowing payroll growth softens the base for group life and voluntary benefits premiums (tied to covered payroll), while a smaller labor pool has limited near-term read-through for P&C personal lines exposure growth. 

Initial jobless claims held at 199,000 for the week ending August 1, the third straight week below 200,000, while the 4-week average fell to a cycle-low 198,750, but continued claims jumped 24,000 to 1.80 million. 

Earnings implications: Low initial claims point to stable workers' compensation claim frequency for P&C carriers, but the rise in continued claims signals longer unemployment durations that bear watching for group disability claim duration at life insurers. 

AUM Through the Cycle

The table directly below presents AUM roll forward data for PRU, EQH, and HIG from 2Q21 through 1Q26, individual company net change (as a percentage of beginning AUM), and S&P 500 quarterly price return. The pattern that emerges is both consistent and actionable: AUM balances follow S&P 500 returns.

Updated company AUM data for Equitable and Voya show increases in AUM. This result is expected given the equity market data we tracked throughout 2Q. 

Company AUM Roll Forward Detail

Equitable up: Company/Security - Assets Under Management - EQH-US - FactSet 

Hartford drop: Company/Security - Assets Under Management - HIG-US 

Prudential drop: Company/Security - Assets Under Management - FactSet 

Principal: Company/Security - Operations by LOB - FactSet 

Voya up: Company/Security - Assets Under Management - FactSet 

Manulife drop: Company/Security - Assets Under Management - FactSet 

Accessing Insurance Insight Reports

FactSet Insight blog: Read more insurance sector analysis in our previous blog articles.

Document search: Access insurance insight reports from the FactSet Workstation using the Document Search function. Search for "Insurance Tracker: Event of the Week".

Today’s Top News: Insurance insight reports are also delivered on the Workstation through StreetAccount’s Today’s Top News, or “TTN”. Aside from providing access to insight reports, TTN provides an insurance-focused daily sector synopsis, updated events calendar, and “insurance reads” that highlight current news stories that impact the insurance sector.

To access TTN, select the Today’s Top News tab on the FactSet Workstation and select Insurance from the drop-down menu (both highlighted below).

Insurance Solutions

Deep sector data and functionality shown in this report are available through the FactSet Workstation. Learn more about FactSet insurance solutions that combine investment research, portfolio construction, and risk management in a cloud-native platform. Our comprehensive tools enable investment and actuarial teams to enhance asset modeling and capitalize on market opportunities.

This blog post is for informational purposes only. The information contained in this blog post is not legal, tax, or investment advice. FactSet does not endorse or recommend any investments and assumes no liability for any consequence relating directly or indirectly to any action or inaction taken based on the information contained in this article.