Twenty-nine product liability verdicts of at least $10 million landed against manufacturers in 2025, totaling $12 billion. That is 47% of the $25.6 billion in nuclear verdicts Marathon Strategies tracked across every industry, in a year the total verdict count jumped 40.7% to its highest level since 2009 (Insurance Journal, August 25, 2026, citing Marathon Strategies).
A blended casualty severity trend cannot price that concentration. A manufacturers book riding an industry-average trend pick is carrying a damages exposure the manual class plan was never built to describe.
Key Takeaways
- Product liability absorbed 47% of 2025's $25.6 billion in nuclear verdicts, the second straight year the segment dominated the tail.
- The count of $10 million-plus verdicts rose 40.7%, the largest single-year jump Marathon has recorded. That is a frequency shift, and it raises attachment probability at every excess layer.
- Nuclear verdicts touched 68 industries, up from 48 in 2023. Classes with no large-loss history in the experience window now carry latent severity the manual differential has not caught up to.
- The Bayer Roundup litigation shows the driver: a $2.065 billion single-plaintiff verdict at a roughly 30-to-1 punitive ratio, now headed into a $7.25 billion capped class settlement.
- Primary general liability rate increases slowed to 2.6% while umbrella and excess accelerated to 5.3%, so only the top of the tower is still pricing for the trend.
Product Liability's Share of the Verdict Pool
Manufacturers do not represent anywhere near half of a diversified commercial general liability portfolio. Yet their product liability exposure absorbed just under half of everything the nuclear-verdict tracker counted in 2025.
The split did not hold at the same intensity in prior years. Marathon's 2024 edition put product liability at 32 verdicts totaling $13.9 billion, against a larger $31.3 billion overall pool, a lower concentration share than 2025's.
| Year | Product liability verdicts | Product liability dollars | All-industry total | Industries with a nuclear verdict |
|---|---|---|---|---|
| 2023 | – | – | – | 48 |
| 2024 | 32 | $13.9 billion | $31.3 billion | 55 |
| 2025 | 29 | $12 billion | $25.6 billion | 68 |
The manual class differential for products coverage is typically calibrated against a multi-year blend of verdict experience, spread across every class the manual covers. The blend assumes no single class dominates the tail in any given year. A tail that has now run 47% concentrated in one segment two years running is not a problem the next loss trend selection round can quietly absorb. It is a signal that the products differential needs a standalone severity trend, distinct from the general liability book it currently rides inside.
Frequency Moved as Much as Severity
The count of $10 million-plus verdicts rose 40.7% year over year, the largest single-year jump Marathon has recorded since it began the series. That is a frequency statistic, not a severity one. A severity-only trend selection built around the average or median size of a large verdict misses a shift in how often a large verdict happens at all.
For an excess or umbrella layer sitting above a manufacturer's primary limits, the relevant question is not how large the average nuclear verdict got in 2025. It is how much more often a claim in the book now has a real probability of attaching that layer in the first place. A 40.7% jump in frequency, concentrated disproportionately in one line, changes the attachment probability at every excess layer feeding off the primary product liability tower, independent of average verdict size.
68 Industries: Where the Class Plan Goes Blind
Nuclear verdicts touched 68 industries in 2025, up from 55 in 2024 and 48 in 2023 (Insurance Journal, August 25, 2026, citing Marathon Strategies). That breadth matters more to a products book than to almost any other line.
Product liability manual rates are built class by class. A class with no prior large-loss history in the experience period underlying its differential is treated as a lower-severity risk almost by construction.
Twenty additional industries picked up a nuclear verdict in two years. Classes that never fed the tail of a manual rate review now carry latent severity exposure the differential has not yet caught up to. A pricing actuary reviewing product liability differentials off a five- or seven-year experience window is still working from a period when several of 2025's newly affected classes had no large-loss signal at all.
The Bayer Case: Why the Damages Resist a Cap
The Barnes Verdict and Its 30-to-1 Punitive Ratio
The clearest illustration of what is driving the concentration sits in the agricultural chemical sector. A Cobb County, Georgia jury awarded John Barnes $2.065 billion against Bayer's Monsanto unit in March 2025 over Roundup-linked non-Hodgkin lymphoma (CNN, March 24, 2025). The award split $2 billion in punitive damages against $65 million in compensatory, the largest single-plaintiff verdict in Georgia history.
A roughly 30-to-1 punitive-to-compensatory ratio is not a number an actuarial severity model built on economic loss inputs can reproduce. Punitive damages track jury sentiment toward the defendant's conduct, not the plaintiff's measurable harm. No amount of medical-cost or lost-wage data in a claims file would have flagged that ratio in advance.
The $7.25 Billion Capped Settlement
Bayer's response is itself evidence of how large this exposure has become for a single manufacturer. The company proposed a $7.25 billion class settlement in February 2026 to resolve current and future Roundup claims from roughly 65,000 plaintiffs, structured as capped payments running over as long as 21 years. It also raised its litigation provisions from €7.8 billion to €11.8 billion, about $13.9 billion (Fortune, February 18, 2026).
A Missouri court granted preliminary approval on March 4, 2026 (Claims Journal, March 5, 2026). The final approval hearing, originally set for August 19, has since been pushed to September 14, 2026 to give the class administrator time to process opt-out revocation requests (Seeking Alpha, August 2026).
A single manufacturer resolving mass-tort exposure through a capital-markets-style capped settlement, funded over two decades, is a different animal from a conventional excess-of-loss tower absorbing individual verdicts one at a time. It is also the kind of resolution other product liability defendants facing a similarly concentrated tail will now be measured against.
The Tail: 40 Verdicts Over $100 Million
Forty verdicts in 2025 reached $100 million and four topped $1 billion, both drawn from the same Marathon Strategies count (Insurance Journal, August 25, 2026). The report does not publish a median verdict figure, but the tail is lengthening faster than the median is likely rising. When the top four data points alone plausibly account for several billion dollars of the $25.6 billion total, the upper tail is doing most of the work.
High policy limits, and the excess layers built to absorb exactly this kind of event, are the part of a products tower most exposed to a tail growing in count and in size at the same time. The same severity-driven repricing is already showing up in umbrella and excess ILF selections industry-wide, where the connection between verdict severity and layer attachment is more direct than in primary casualty pricing.
Rate Environment: Primary Softens While Excess Hardens
The timing compounds the problem. General liability rate increases reported to CIAB decelerated from 4.2% in the first quarter of 2025 to 3.9% in the second, and to 2.6% in the first quarter of 2026 (CIAB, Q1 2026 P&C Market Survey). Umbrella and excess liability moved the opposite direction, accelerating from 4.8% in the first quarter of 2026 to 5.3% in the second, its 35th consecutive quarterly increase (Insurance Journal, August 20, 2026, citing CIAB).
| Line (CIAB rate change) | Q1 2025 | Q2 2025 | Q1 2026 | Q2 2026 |
|---|---|---|---|---|
| General liability | 4.2% | 3.9% | 2.6% | – |
| Umbrella and excess | – | – | 4.8% | 5.3% |
Ivans' renewal data shows the same GL deceleration carrying into midyear at finer granularity, down to a 4.99% renewal rate change in July 2026 from 6.85% at the start of the year. This site has tracked that trend against the CAS and Triple-I's 6.6% to 6.8% annual casualty severity benchmark.
Primary general liability rate discipline was softening at the exact moment a 40.7% jump in nuclear verdicts, concentrated in manufacturing, landed. The excess and umbrella layers sitting above that primary book are the only part of the tower still pricing for it.
The Desk
Four moves follow from the 2025 data:
- Give the manufacturers differential its own severity trend. Separate it from the general liability blend it currently sits inside, and build the selection around the 47% concentration figure rather than an industry-average casualty pick.
- Revisit policy-limit profiles before the next renewal. Test any manufacturers account with meaningful product liability exposure against the 68-industry breadth figure, particularly classes that entered the nuclear-verdict count for the first time in 2025 and have no prior large-loss signal in the experience period underlying their current rate.
- Split the tower in the broker conversation. On a products account, separate the primary GL layer, where rate is decelerating, from the umbrella and excess layers, where it is still accelerating. A blended increase across both masks which layer is actually absorbing the concentration risk.
- Track two dates. The Bayer settlement's rescheduled approval hearing on September 14, 2026 will show whether courts will let a capped, capital-markets-style resolution stand for a mass-tort products defendant. CIAB's next quarterly survey, expected in November, will show whether general liability's deceleration continues even as umbrella keeps hardening against the same verdict data.
Further Reading
- Umbrella Pricing Holds at 9.4% as Other Commercial Lines Soften: ILF Repricing and Nuclear Verdict Severity – how excess layers are already repricing against the same verdict severity data.
- Carriers Deploy AI Against Social Inflation as Nuclear Verdicts Double – how carriers are trying to model the jury-behavior drivers behind this trend.
- General Liability's 114% Combined Ratio Signals Rate Inadequacy – the primary GL book's underlying loss experience against the rate deceleration this piece documents.
- Ivans July Index: GL Renewal Rate Slips to 4.99% as WC Drag Eases – the month-by-month primary casualty rate deceleration referenced above.
- $604M Broker Verdict Turns Contingent Auto Into a Priced Exposure – a comparable case of a single nuclear verdict forcing a line with no established rating base to reprice.
Sources
- Insurance Journal, "Nuclear Verdicts Go Boom, Increase 40.7% in 2025" (August 25, 2026)
- Marathon Strategies, "Corporate Verdicts Go Thermonuclear: 2025 Edition"
- CNN, "Georgia jury orders Monsanto parent to pay nearly $2.1 billion in Roundup weedkiller lawsuit" (March 24, 2025)
- Fortune, "Bayer reaches $7.25 billion settlement over claims that Roundup caused cancer" (February 18, 2026)
- Claims Journal, "Bayer Wins Court Nod for $7.25 Billion Roundup Settlement" (March 5, 2026)
- Seeking Alpha, "Bayer granted court delay to September for Roundup settlement hearing" (August 2026)
- CIAB, Q1 2026 P&C Market Survey
- Insurance Journal, "CIAB Survey: Overall Soft Market Conditions Continue in Q2" (August 20, 2026)