AppLovin reported second-quarter results after the close on August 5. Revenue of $1.92 billion grew 52.8% year on year, adjusted earnings of $3.76 met consensus, and adjusted EBITDA of $1.61 billion rose 58% with margins roughly 300 basis points wider than a year earlier. The shares fell 19.7% the next session to $335.67, a fifty-two-week low, on more than twice average volume.

What actually went wrong

Revenue landed $18 million below the $1.94 billion consensus, under 1%, and below the company’s own guidance range. That last part is the whole story. AppLovin had never missed its guidance midpoint since listing in 2021, and the shares carried a fifty-times forward multiple because those forecasts were treated as a floor rather than an estimate. What came out of the price was that premium, not a repricing of demand.

Management’s explanation is narrow. Gaming still supplies most of revenue, and its growth depends on the cadence of machine-learning model improvements: when a model improves, advertisers hit their return targets at higher budgets. Chief executive Adam Foroughi said the pace of improvement was lighter than usual during the quarter and that the next step up landed just after it closed. Sequential growth was 4%, against double-digit prints in eleven of the previous twelve quarters.

Three claims that can be checked

MAX publisher earnings grew double digits sequentially, which is inconsistent with weakening advertiser demand. The share of publisher waterfalls was unchanged, which is inconsistent with competitive loss. And consumer advertiser spend ran 28% above the fourth-quarter 2025 seasonal peak during the seasonally weakest quarter of the year. All three point away from a structural problem.

Third-quarter guidance of $2.055-2.085 billion implies 46-48% growth and 7-8% sequentially, and the finance chief said it assumes only those model releases already live. The Securities and Exchange Commission separately closed its inquiry with no action.

Where the risk sits

Foroughi described the growth engine on the call as a series of A/B tests with no guarantee of a lift in any three-month window. That is honest, and it is also the bear case: the multiple this business deserves was always lower than the one it carried. The creative bottleneck blocking e-commerce self-service is also unchanged: the platform still cannot hand a small advertiser a usable long-form video.

At 18.9 times this year’s adjusted earnings and 15.8 times next year’s, the shares sit below Meta’s multiple while growing revenue roughly ten times faster. The rating is Buy, staged rather than committed at once, against a probability-weighted fair value of $521.

Valuation workings, the risk table and the entry and exit levels are laid out in the complete AppLovin Corporation (APP) report on our Reports page.