Corporate nuclear verdicts of $10 million or more reached 135 in 2024, up 52% from 89, totaling $31.3 billion, and the industry added $16 billion to prior-year P&C liability loss estimates in that year's reserve reviews. The tool carriers are reaching for is AI: venue scoring, litigation prediction, automated settlement ranges. It is trained on the loss history that social inflation has already made unrepresentative.
Key Takeaways
- 135 nuclear verdicts totaling $31.3 billion in 2024, a 52% increase in count and 116% in value from $14.5 billion, spread across 55 industries, 34 states and 77 courts.
- 49 thermonuclear verdicts above $100 million, up 81.5% from 27, with five exceeding $1 billion against two the prior year.
- The commercial auto liability 12-to-60-month development factor more than doubled between 2007 and 2024, which is a structural break that multi-year link ratio averages understate by construction.
- Attorney-involved claims average $77,807 in indemnity against $15,936 unrepresented, a 390% difference on durations 295% longer, which is the signal the triage models are built to catch.
- $16.1 billion of litigation finance assets across 42 active providers, a funding structure that removes the settlement pressure the historical data was generated under.
What 2024 Actually Recorded
Marathon Strategies counted 135 corporate nuclear verdicts of $10 million or more, the highest since tracking began in 2009 and a 52% rise on the 89 of 2023. Total value reached $31.3 billion, up 116% from $14.5 billion, across 55 industries, 34 states and 77 courts, all records.
The tail moved faster than the count. Thermonuclear verdicts above $100 million rose 81.5% to 49, from 27, and five exceeded $1 billion against two in the prior year. The median nuclear verdict reached $51 million, up from $44 million. Concentration by sector was uneven: beverage at $8.5 billion, entertainment at $4.7 billion largely on the NFL Sunday Ticket antitrust verdict, and agricultural chemicals at $2.3 billion.
The composition matters more than the totals. The Institute for Legal Reform's review of more than 1,288 nuclear verdicts from 2013 to 2022 found noneconomic compensatory damages exceeding combined economic and punitive damages in six of ten study years, so the growth sits in the least predictable award category. Swiss Re's Social Inflation Index put the non-economic component at 5.4% annually between 2017 and 2022, reaching roughly 7 percentage points by 2023, and cumulatively 57% added to US liability claims over the decade. CAS and Triple-I sized the excess liability losses at $231.6 billion to $281.2 billion.
The Models Learn From the Wrong Decade
Carriers now run four classes of tool across the claim-to-verdict pipeline, and the case for them is genuinely strong at the triage end.
Venue scoring assigns risk from historical verdict data by court, judge and jurisdiction; Premonition, on a database above 325 million cases across 3,124 civil courts, finds 30.7% of case outcomes determined by the judge-lawyer relationship. Early flagging targets the sharpest split in the data: claims with attorney involvement average $77,807 in indemnity against $15,936 unrepresented, a 390% difference, on durations 295% longer. Outcome prediction and settlement range estimation feed authority decisions directly.
The problem is the training window. Social inflation's current episode accelerated in the mid-to-late 2010s on litigation funding maturing as an asset class, reptile theory becoming standard plaintiff practice, and shifting juror attitudes. A model fitted on claims from 2005 to 2018 learns development patterns from an environment where those forces were absent or nascent.
The break is measurable. CAS and Triple-I found the calendar-year 12-to-60-month development factor for commercial auto liability more than doubled between 2007 and 2024. Other liability occurrence severity compounded at 6.8% a year from 2015 to 2024, more than double the 3.2% CPI rate. A multi-year average of link ratios across that period does not describe either regime.
At the tail the data is not merely stale, it is absent. Nuclear verdicts are extreme by definition, and pre-2018 datasets contain few of them. A gradient-boosted model fitted on 200,000 closed claims from 2010 to 2017 might hold fewer than 50 claims above $10 million in indemnity. A model cannot estimate a distribution from fifty observations of it.
That is why the $16 billion of prior-year liability reserve additions in 2024 is the operative number for reserving rather than the verdict count. Recency weighting and synthetic augmentation both narrow the gap without closing it, because the drivers are exogenous to the loss records. An explicit social inflation trend load on top of the model output is doing work no amount of feature engineering on internal claims data can do, and ASOP No. 43's requirement to consider significant risks and uncertainties is where that judgment gets documented.
Funding Removed the Pressure the Data Was Generated Under
The confounding variable that AI claims models are structurally unable to observe is who is paying for the plaintiff's case.
Third-party litigation funding turns a lawsuit into an investable asset. Westfleet Advisors counted $16.1 billion in assets across 42 active capital providers from mid-2023 to mid-2024, with projections of $18.9 billion in 2025 and possibly beyond $67 billion annually by 2037 at roughly a 10.7% compound rate. Burford Capital alone reported $7.3 billion in capital commitments.
The mechanism is specific and it targets exactly the part of the distribution settlement models are trained on. A funded plaintiff faces no personal financial pressure to settle, because the funder absorbs downside and covers legal cost in exchange for a share of recovery. That removes the pressure that governed the low and middle of the severity distribution in every year of the training data. A model calibrated on pre-funding settlement behavior will underestimate both the probability that a claim goes to trial and the size of the verdict when it does.
The industry has not aligned its own incentives against this either. The 2026 CLM study found 80.6% of claim executives reporting rising indemnity over three years, 75.8% higher defense costs and 85.2% larger policy limit demands, and for the first time in eleven years 36.1% now believe sophisticated defense spending reduces indemnity. Yet 50.8% of carriers rarely or never approve defense firms' use of AI tools, and only 6.2% have moved past hourly billing. The efficiency the models are supposed to deliver is being bought under a fee structure that pays for hours.
Further Reading
- Social Inflation and Litigation Trends 2026: The $529 Billion Challenge Reshaping Casualty Insurance
- Schedule P Data Maps the Casualty Reserve Problem Across 2021-2024 Accident Years
- Soft Market Returns to P&C: A Reserve Adequacy Playbook for the 2026 Pricing Downturn
- Machine Learning for Loss Reserves: The ASOP Compliance Gap
- AI Fraud Detection in P&C: Testing Deloitte's $160B Savings Claim
- Eight States Target the TPLF Blind Spot With Mandatory Disclosure – How new state and federal disclosure rules could make litigation funding status observable for AI claims models, addressing the confounding variable gap identified in this article.
Sources
- Marathon Strategies, "Corporate Verdicts Go Thermonuclear: 2025 Edition," May 2025 - marathonstrategies.com
- TransRe, "Social Inflation Overview," November 2025 - transre.com
- U.S. Chamber Institute for Legal Reform, "Nuclear Verdicts: An Update on Trends, Causes, and Solutions," May 2024 - instituteforlegalreform.com
- Swiss Re Institute, "sigma 4/2024: Social Inflation: Litigation Costs Drive Claims Inflation," September 2024 - swissre.com
- Casualty Actuarial Society / Insurance Information Institute, "Increasing Inflation on Liability Insurance - Impact as of Year-End 2024," November 2025 - casact.org
- Munich Re, Q1 2026 Quarterly Statement, May 2026 - munichre.com
- Travelers Companies, Q1 2026 Earnings Release, April 2026 - investor.travelers.com
- Chubb Limited, Q1 2026 Earnings Release, April 2026 - chubb.com
- Milliman, "U.S. Casualty Insurance 2024 Financial Results: What Kind of Market Are We In?," 2025 - milliman.com
- Westfleet Advisors, "2024 Litigation Finance Report," 2025 - westfleetadvisors.com
- Marsh, "Social Inflation and Nuclear Verdicts," 2025 - marsh.com
- Gallagher, "Three Key Drivers of Social Inflation," 2025 - ajg.com
- Sedgwick, "2025 Liability Litigation Commentary: Inside the Verdict," August 2025 - sedgwick.com
- NAIC, "Insurance Topics: Social Inflation," December 2025 - naic.org
- American Tort Reform Foundation, "Judicial Hellholes 2024-2025 Report," December 2024 - judicialhellholes.org
- Swiss Re, "U.S. Property & Casualty Outlook: The Past Weighs on the Present," July 2025 - swissre.com
- Gen Re, "Litigation Analytics: Turning Data into a Competitive Advantage for Casualty Insurers," August 2025 - genre.com
- Insurance Journal, "AI for the Defense: CLM Litigation Management Study," April 2026 - insurancejournal.com
- CLARA Analytics, "Litigation Risk Prediction Platform," 2026 - claraanalytics.com