AppLovin added $54.3 billion in five sessions; lead-plaintiff papers in its suit are due November 16
Lead-plaintiff papers in the AppLovin securities class action are due November 16, nine days after a five-session rebound added $54.3 billion to the stock. The suit does not contest whether the company's AI ad models work; it contests the improvement schedule the company allegedly promised.
Richard Tangthree-minute read
Investors who bought AppLovin+9.76% — AppLovin, up 9.76 percent today between February 12 and August 5, 2026, have until November 16 to ask the court to make them lead plaintiff in the securities class action against the company, and the shareholder-rights firms competing to represent them are down to nine days to say so 12. Their claim is precise: the AI advertising company sold a story of constantly improving models, and the story broke twice in one summer 1.
The third quarter proved the machine works. What no quarter can prove is that the machine runs on a schedule, and the schedule is what was allegedly promised.
The cycle is real, and the quarter proved it
AppLovin's Axon engine predicts which ad wins an auction and what it is worth; advertisers set a target return on ad spend, and when prediction improves, more of their budget deploys profitably at the same target 3. Third-quarter revenue rose 55% to $2.18 billion, roughly $100 million above the top of the company's own guidance, after a partial rollout of the next model generation 3.
The 10-Q, its quarterly filing, shows what that means in the unit that matters: nine-month net revenue per installation up 68% on 7% fewer installs 3. This is a price-and-prediction business, not a reach business, and prediction just had its best quarter of the year.
AppLovin's growth dipped inside the class period and re-accelerated after the model rollout
Growth dipped inside the class period, then re-accelerated on the new models
- Year-over-year revenue growth
- Estimate
Data
| Year-over-year revenue growth | |
|---|---|
| Q3 '25 | 68.2% |
| Q4 '25 | 65.9% |
| Q1 '26 | 59% |
| Q2 '26 | 52.8% |
| Q3 '26 | 55.5% |
| Q4 '26 guide (estimate) | 43.5% (43–44%) |
The alleged promise was a pace
The complaint, filed September 16 and disclosed in the 10-Q, quotes the "virtuous cycle" pitch and a May call in which management said it had "100% seen faster improvements to the models" and saw no reason for that to slow 35. Then a Bank of America−0.48% — Bank of America, down 0.48 percent today note on soft June e-commerce growth took 12.65% off the stock on July 13, to $442.85 1.
On August 6, after second-quarter revenue of $1.92 billion missed consensus and management called model improvement "lighter than normal" and the generative video-creative tool "still [a] work in progress," shares fell 19.66% to $335.67 1. Four analysts cut their price targets that day 5.
On November 4, chief executive Adam Foroughi said the quiet part: model gains "won't land on a perfect quarterly schedule," and "we won't promise a cadence" 3.
The same investors are plaintiff and rebound
Those two drops are the loss the case will size, and the investors who ate them are now riding the rebound: five straight winning sessions added $54.3 billion in market value while the S&P 500 fell, and the stock is still down 41.9% for the year after entering the print as the index's worst performer 78. The sell side that cut targets in August raised them after the beat 6.
AppLovin says the allegations lack merit 3, and a court, not a quarter, will decide intent. But the record already carries the lesson for anyone pricing AI companies: a model-improvement cadence is a research process, not a release schedule, and this market had priced it as a schedule.
The cycle is real. The promise of a timetable is what is on trial, with papers due November 16.
Get Applications by email, free
How AI is changing the software people and companies use. In our daily email, every weekday.
