What Roundup tells us about the next mass tort.

In the spring of 2018, a California jury found that Roundup had caused a school groundskeeper’s non-Hodgkin lymphoma and awarded him $289 million. Bayer’s share price dropped 11% in a single trading session. It would drop another 30% over the next ninety days. By the time the global settlement framework was announced two years later, the company had set aside more than $16 billion against the litigation. The 10-K for fiscal year 2017, filed three months before the verdict, described Roundup-related claims in two paragraphs.

The litigators saw it coming. The plaintiffs’ bar had been mobilizing since 2015. The IARC classification had landed in 2015. The first MDL panel order consolidating cases came down in October 2016. By the time of the verdict, there were 8,700 plaintiffs already in the docket. None of this was a secret. It was, in fact, the loudest possible signal that something was about to happen — to anyone who was reading dockets instead of reading 10-Ks.

This is the central, repeating pattern of every mass tort in modern memory. The legal market sees the wave forming twenty-four to thirty-six months before the financial market does. The lag, in our experience, is the most reliable source of alpha in litigation-aware investing. It is also the most reliable source of catastrophic surprise for a CFO who finds out about a reserve adjustment at the same time as their auditor.

How a mass tort actually forms

There is a recognizable sequence. We have seen it run cleanly on Roundup, Zantac, talc, opioids, hair-relaxers, paraquat, 3M earplugs, and the social-media MDL. The shapes change. The sequence does not.

1. A science enters the public record — a regulatory classification, a meta-analysis, an internal document leak.
2. Two or three outlier verdicts come out of plaintiff-friendly venues, each one large enough to be on the front page of the regional bar journal but not yet on the financial press.
3. An MDL panel consolidates the early cases. Filing volume jumps by an order of magnitude in the next four quarters.
4. The defendant’s 10-Q reserve line moves. The 10-K legal-proceedings section gets one more paragraph, slightly hedged.
5. The settlement framework — or the bellwether trial that forces it — arrives. The share price catches up.

Steps one through three are visible to any litigator paying attention. Steps four and five are visible to everyone. The gap between three and four is where the financial market is wrong. It is also where, in five separate dockets right now in 2026, our methodology is pointing.

Five mass torts in motion

We are not going to score these dockets in detail in a public post. But we will describe the shapes, because the shapes are recognizable, and we think the readers of this blog — GCs, funders, analysts, partners — should be running their own model against them.

1. PFAS in consumer products

The “forever chemicals” story has been told in the press for half a decade, but the litigation has only recently broken out of the municipal-water-supply lane and into consumer products — cookware, cosmetics, contact lenses, food packaging. Filing volumes in 2025 were up roughly 4x over 2023. Several large CPG defendants are now named in cases their 2024 10-Ks did not anticipate. The first bellwether verdicts in the consumer-products track are expected within twelve months.

2. Generative AI training-data copyright

The first generation of training-data cases — NYT v. OpenAI, the various publisher suits against Anthropic and Meta — have produced a settlement template and a procedural posture that other plaintiffs are now copying. The plaintiff side is no longer fragmented. The defense side is more exposed than its disclosures suggest, because the operative claim is increasingly being framed as a per-work statutory damages question rather than a fair-use question. The exposure math, if a single bellwether goes badly, is non-trivial.

3. Social-media addiction (teen mental health MDL)

Now a consolidated MDL in N.D. Cal., with school districts, state AGs, and individual plaintiffs all in the docket. The science is contested, the discovery has produced unusually quotable internal documents, and the bellwether schedule is tightening. This is the docket most likely to produce a Roundup-shaped verdict in the next twenty-four months.

4. Hair-relaxer carcinogenicity (uterine cancer MDL)

The N.D. Ill. MDL has crossed 10,000 plaintiffs. The scientific paper that started the litigation is methodologically contested, but the plaintiffs’ bar is now coordinated, the bellwether selection process is underway, and the defendant universe is consolidated enough that the per-defendant exposure math is unusually concentrated.

5. GLP-1 side-effects (gastroparesis MDL)

The newest of the five, but moving fast. The MDL was consolidated in 2024, the docket is growing, and the underlying drug class is a generational blockbuster — meaning that even a low per-plaintiff settlement value compounds into a number that can move a balance sheet. The pharmacovigilance question is genuinely open, which is both why the litigation exists and why the variance on the outcome is wide.

The lag is the alpha

We are not in the business of stock-picking. We are in the business of making litigation risk modelable — which is, by definition, the upstream of every interesting trade, reserve decision, and underwriting call in this space. But the mechanism is worth being honest about.

Roundup was a thirty-month lag between the verdict that should have repriced Bayer and the date the financial market did. Opioids was longer. Zantac was shorter. Across the eight torts we have back-tested most carefully, the median lag is roughly twenty-two months between the first measurable signal — first MDL consolidation, plus a non-trivial bellwether verdict — and the first material reserve adjustment.

Twenty-two months is an enormous window. It is the window in which the relevant analysts could have repriced the exposure, the GC could have re-set the reserve, the insurer could have repriced the layer, and the funder could have built the position. In most cases, none of those things happen. The 10-K describes the matter in a paragraph. The market reads the paragraph. The lag persists.

Why we publish this

Because we think the next decade of litigation is going to be defined by the fact that this lag is closing. The plaintiffs’ bar is more sophisticated and better capitalized than it was in 2015. The financial market is paying more attention to legal risk than it was in 2015. The tools available to model these dockets — including, frankly, ours — are better than they were in 2015. The lag is going to compress. The defendants and analysts who close it first will be the ones who used the period when the data was already public but no one was reading it.

The next Roundup is already filed. The 10-K that describes it is already public. The question is whether your model is faster than your auditor.

Methodology note

We score active mass torts on five dimensions — claim signal, venue concentration, plaintiffs’ bar mobilization, defendant exposure, and settlement velocity — and aggregate them into a single forward-looking score. The methodology is one of the six research tracks that runs on every matter we take on. We are happy to walk serious counterparties through the per-docket scoring under NDA. Email us at info@valarhq.com.