AppLovin (APP) Q3 2026 Earnings Call: Q&A Summary

Q: What are the biggest improvements in the new model generation compared to Axon 2, and what is the rollout timing?

A: Adam Foroughi stated they are starting to roll out the new models, which offer more room to optimize than the previous generation, though he could not commit to a specific full rollout timeline.

Q: Is the new architecture generalizable across media types and client types?

A: Adam Foroughi confirmed it is possible to take the architecture in that direction.

Q: Should we think of this as Axon 3.0, and how impactful will it be?

A: Adam Foroughi described it as the next generation of the model, noting it is technically complex and the team is excited, though it remains in a partial rollout phase.

Q: How do you frame the sequential deceleration in the Q4 revenue guide compared to Q3?

A: Matt Stumpf explained that their guidance philosophy remains consistent and includes normal seasonality, the model updates already in place, and increased data center costs.

Q: What is the update on the progress of the referral program?

A: Adam Foroughi noted it is a multi-year, multi-quarter trajectory, but the rate of pixel accounts is accelerating and will compound over time as more customers gain experience.

Q: Does the new model change expectations for gaming ad revenue growth, given the law of large numbers?

A: Adam Foroughi reiterated a commitment to 30% long-term growth, noting that the consumer vertical is growing triple digits and provides significant headroom, making it difficult to judge verticals in isolation.

Q: How much of the 30% long-term growth will come from model improvements versus new customer additions?

A: Adam Foroughi stated the components are hard to predict, but the consumer vertical is early in its evolution and offers a long runway for growth.

Q: What is embedded in the Q4 guide for gaming versus consumer growth?

A: Matt Stumpf declined to break out the specific splits but noted that both segments are strong for different reasons and the guide reflects seasonality and model improvements.

Q: How prominent is CTV in the product roadmap compared to six months ago?

A: Adam Foroughi said pilot campaigns are promising, but the product is not yet accretive to the numbers, though the team is gaining confidence.

Q: Do you have line of sight to a tipping point for the consumer flywheel, and what incremental investments are required?

A: Adam Foroughi explained that data and model architecture go hand in hand, and they are looking for ways to acquire data to train larger, more complex models.

Q: Are you generating enough data organically, or do you need to acquire it?

A: Adam Foroughi stated they are performing well with current data, but models need more data to improve, so they are looking for ways to get data into the system faster.

Q: Do you see the same 3% to 5% self-learning tailwind from intra-quarter model retraining?

A: Adam Foroughi said it is difficult to isolate specific percentages, but self-learning is inherent in deep learning models and compounds into the business results.

Q: How are consumer advertising budgets shaping up for the holidays, and should we expect a seasonal reset in Q1?

A: Adam Foroughi said they follow normal patterns where Q4 is strong and Q1 is traditionally weaker, though gaming's Q5 phenomenon helps insulate the business.

Q: How has the learning budget or time for e-commerce customers improved?

A: Adam Foroughi noted that learning is two-sided and customers must validate results with incrementality studies, but the product is improving every quarter.

Q: Are you concerned about ROI compression from new bidders in the auction?

A: Adam Foroughi argued that industry checks are often flawed, and while better targeting increases CPM, it also improves ROAS, which is why they are growing quickly.

Q: What is the update on the market and competition?

A: Adam Foroughi stated that while other companies are performing well, it is a zero-sum game, and the company put up a massive quarter despite the competition.

Q: Is a more diversified ad format the goal for e-commerce, or are you waiting for Gen AI video creation?

A: Adam Foroughi said both are important, as diversity of content creates uplift, and the ad format will continue to evolve over the next few years.

Q: Is ad creative still a primary barrier to success for customers?

A: Adam Foroughi said it is not a barrier but an opportunity cost, as customers who invest in creative and ad count have more success on the platform.

Q: What is the outlook for supply growth through 2027?

A: Adam Foroughi noted that the MAX market is growing at a strong double-digit rate every year.

Q: How do you frame the cost of supply acquisition?

A: Adam Foroughi explained that there are two drivers: more supply and rising costs, and the company has consistently improved its technology faster than the market to grow.

Q: Does the added buyback authorization give you flexibility to be more aggressive?

A: Matt Stumpf stated the meaningful step-up in Q3 was opportunistic due to stock price dislocation, and they have the ability to continue being opportunistic.

Q: Why is the APAC region finding stronger product-market fit for e-commerce?

A: Adam Foroughi explained that APAC customers often spin up new sites specifically for the channel, which avoids the attribution complexity faced by larger, long-standing Western brands.

Q: What is the ideal M&A candidate or data you are looking for?

A: Adam Foroughi said they look for valuable data and customer relationships that can accelerate growth, similar to the MoPub acquisition, to compress a decade of work into a shorter timeframe.

Q: What is the take rate directionally?

A: Adam Foroughi stated they do not talk about take rate, as it is not a function they optimize to.

Q: Who owns the data in your transactions?

A: Adam Foroughi clarified that there is no magical data access, as MAX is a fair marketplace where auction data belongs to the bidder.