Revenue soared by 53% but still faced a sell-off! AppLovin (APP.US) saw its shares plunge by over 25% in after-hours trading, simply because its AI model upgrade lagged behind.

After the US stock market closed on Wednesday, mobile advertising giant AppLovin (APP.US) reported a significant profit surge but faced harsh judgment from investors. The company's stock price plummeted over 25% in after-hours trading due to its second-quarter revenue falling slightly short of market expectations and providing a less-than-inspiring guidance for the next quarter despite a 55% year-on-year rise in net profit and an adjusted EBITDA margin remaining at a rare 84%. The earnings report showed that for the second fiscal quarter ending June 30, AppLovin achieved revenue of $1.92 billion, a 53% year-on-year increase, but still fell short of the consensus estimate of $1.94 billion from analysts. The adjusted earnings per share were $3.76, slightly exceeding the market consensus of $3.75. Net profit reached $1.27 billion, up sharply by 55% compared to $820 million in the same period last year; adjusted EBITDA was $1.61 billion, showing a 58% year-on-year growth, continuing to demonstrate strong profitability. However, what raised concerns in the market was that this performance not only failed to exceed Wall Street expectations but was also below AppLovin's own internal guidance. In the subsequent earnings call, management identified the key reason for the disappointing results in one word: timing. The slowdown in the pace of model upgrades was explained by management as a timing issue. Co-founder and CEO Adam Foroughi candidly stated during the call that the companys game-focused advertising business is highly dependent on the performance improvements of its AI models. Every significant iteration of the model allows advertisers to invest more in their budgets while maintaining target returns on ad spending. However, in the recently concluded second quarter, this leap in model performance did not materialize as expected. The issue this quarter boils down to timing, Foroughi explained. The rhythm of substantial model improvements was slower than usual, and the next major enhancement in model performance happened to land right after the quarter ended. He emphasized that there was no observed weakening in advertiser demand or adverse changes in the competitive environment during the quarter, with publisher revenues from its aggregation platform MAX achieving double-digit quarter-on-quarter growth, and AppLovin maintaining a stable share in publisher bid waterfalls. This statement aimed to signal to the market that the growth engine itself had not stalled, but rather that the pace of technological upgrades had coincidentally misaligned with the end of the earnings report cutoff. For the current quarter, AppLovin provided guidance that reflects the contribution of new models. The company expects third-quarter revenue to be between $2.055 billion and $2.085 billion, with year-on-year growth of about 46% to 48%, and a midpoint of $2.07 billion slightly below the consensus expectation of $2.08 billion from analysts. Adjusted EBITDA is expected to fall between $1.71 billion and $1.74 billion, with an adjusted EBITDA margin around 83%. CFO Matt Stumpf pointed out that the third-quarter guidance has already accounted for increased training and computational infrastructure costs due to the rollout of the new models, but it does not include any potential future model releases that may still not be realized. He reiterated that the company focuses on absolute EBITDA and free cash flow as its core management metrics and that it will continue to invest as long as computing power investments can generate incremental revenue. Stumpf noted that, in the long term, the adjusted EBITDA margin is expected to remain just above 80%, but it may experience short-term fluctuations due to infrastructure investments. Beyond gaming, AppLovin is actively moving into broader consumer advertising fields such as e-commerce. Foroughi revealed that second-quarter consumer advertising spending set a new record, surpassing the typical peak season in the fourth quarter of 2025 by 28%. However, this segment's scale is still insufficient to fully offset the fluctuations in the gaming business, although management expects its contributions to gradually strengthen. During the period, the company opened its self-service advertising platform called AppLovin Ads Manager to the public. Foroughi stated that the initial target is medium-sized advertisers with budgets willing to absorb the learning costs of the new platform, rather than immediately competing for major brands or a large volume of small tail merchants. The system is currently capable of generating interactive landing pages efficiently, but it still needs to overcome technical hurdles in automatically producing high-quality long-form video ads. Once resolved or alternative solutions are launched, it will significantly lower the creative barriers for small and medium advertisers. In the long run, Foroughi believes that the combination of continuous optimization of the game advertising model and the expansion of consumer business is expected to support the companys revenue to achieve an annual compound growth rate of about 30%. In terms of cash flow, the free cash flow for the second quarter was $863 million. Stumpf explained that the cash conversion rate being below normal levels was mainly due to timing differences in international cash taxes and interest payments, not changes in profitability. He expects improvements in the third quarter, with the full-year free cash flow conversion rate likely to return to about 75% of adjusted EBITDA. The company's balance sheet remains robust, holding $3.05 billion in cash at the end of the quarter, total debt of $3.7 billion, and a net leverage ratio of only about 0.1 times, well below its long-term target of maintaining leverage around 1 time. Regarding share repurchases, AppLovin spent about $551 million to repurchase and retire approximately 1.14 million shares in the second quarter, a significant slowdown compared to the nearly $1 billion in repurchases in the first quarter. Stumpf clarified that this merely reflects a temporary drop in free cash flow for the season and that the companys attitude towards buybacks has not changed. As of the end of the quarter, about $1.8 billion in repurchase authorization remains available. In addition, Stumpf revealed that the voluntary inquiry by the U.S. Securities and Exchange Commission (SEC) has been concluded without any action recommendations, and the company believes this matter is not significant. Why the market is unimpressed: high expectations meet AI anxiety Prior to the earnings report, AppLovin's stock had already fallen about 40% from its high of over $740 this year, with forward P/E ratios retreating from extreme euphoria to around 25 times, much closer to the normative valuations for the advertising technology sector. Because of this, a clean and decisive beat could have triggered a rebound. However, what ultimately materialized were slight shortfalls in both revenue and guidance, providing new ammunition for short sellers. Deeper concerns still relate to AI disruption. Despite AppLovin consistently emphasizing that its Axon system uses AI for precise mobile ad matching and has successfully expanded beyond gaming, some investors remain wary of traditional software and advertising platforms that may be impacted by the AI wave. The analyst community is similarly divided: some optimists maintain target prices far above $700, believing the recent sell-off is excessive; however, more cautious investors suggest that after the platform matures, its easiest acceleration phase may already be over. Ultimately, AppLovins experience again reinforces the harsh logic of the current market: in a trading environment that ruthlessly pursues perfection, a company with annual revenue growth still exceeding 50% is treated as a disappointment simply because its progress is slightly slower than the most optimistic expectations. As revealed in its earnings report paradox by almost any conventional standard, this quarter was strong, but in this climate, the label strong is still far from enough.