The company's second-quarter revenue reached $1.5 billion, up 36% year-over-year.
Bookings totaled $1.6 billion, an 8% increase year-over-year, landing at the low end of the guidance range and below internal targets. Daily Active Users (DAUs) were 123 million, a 10% increase, while engagement hours were 29 billion, up 5% year-over-year. Operating cash flow was $318 million, up 60% year-over-year, and free cash flow reached $294 million, a 66% increase year-over-year.
Quarterly results fell short of expectations primarily due to monetization, measured by bookings per hour, being lower than forecast, particularly among the under-13 demographic. User registrations, retention, and engagement hours all met or exceeded expectations.
Q3 Guidance and Margin Outlook
The company expects Q3 bookings to be between $1.58 billion and $1.65 billion, which would represent a 14% to 18% decline year-over-year against a high comparison base. On margins, the decline in bookings scale will lead to deleveraging on fixed costs. Simultaneously, new AI-driven initiatives like Build, Roblox Reality, and Moments will push infrastructure costs higher. Breaking down the Q3 margin guidance, roughly half of the year-over-year margin compression is attributed to fixed cost deleveraging from lower bookings, with the other half stemming from the aforementioned AI-related investments, primarily incremental cloud GPU computing costs. The company's long-term view on margin potential remains unchanged, with the drivers for margin expansion still in place: bookings growth creating fixed cost leverage, and a shift in business mix toward lower-cost platforms improving COGS. Notably, the nature of this infrastructure investment differs from the past. Historically, infrastructure spending was tied to engagement hours and user numbers. This round is primarily driven by model training, making it more akin to a fixed cost, which should gradually show leverage effects as bookings grow. Furthermore, infrastructure costs related to Morpheus are initially planned to be offset by subscription revenue. The company maintains confidence in achieving long-term revenue growth exceeding 20% and its goal of capturing a 10% share of the global gaming market, with its US market share already significantly above that level.
Q&A Session Highlights
Q: What metrics should investors monitor to understand when headwinds from algorithm changes, Roblox Kids, and select accounts peak?
A: Several things are worth watching. For the under-18 demographic, we've clarified the direction, which includes the original vision of "you make the game." Not every user can figure out Studio, but in New Zealand, we see more daily users for Build than Studio. We believe AI-created games will eventually become a parallel activity to playing games. So, first, watch the frequency of engagement in the under-18 group. Second, we expect AI usage or tokens in the Build process to be generous, but for heavy Build users, we will offer enhanced capabilities within the Roblox Plus subscription, which might have its own token nomenclature. So, monitor subscription metrics for the under-18 group. Third, watch the ongoing rollout of the communications platform. For the over-18 demographic, we are highly focused on specific sub-segments and their growth rates, as well as the content types from the creator ecosystem that show stronger retention in these groups. Watch the growth rate for the over-18 group, as it represents the 80% market where we believe we can replicate our success with the under-18 demographic.
Q: Is the expected Q3 sequential user growth purely seasonal, or did you see trend improvement by the end of Q2?
A: To clarify, you mentioned "short-term engagement friction," but that's not what we observed. We are quite satisfied with engagement trends. Registration, retention, and overall engagement look healthy. The weakness that deviated from our original plan was concentrated in monetization, which will persist as we continue to adjust discovery and recommendation algorithms. Regarding DAU trends, seasonality is a significant factor. We also benefit from a full quarter's contribution from the relaunch in Russia. Those are the two main points influencing DAU trends.
Q: What is the timeline for seeing the long-term retention benefits from discovery engine updates translate into improved engagement and monetization?
A: A discovery algorithm connects users with experiences to drive long-term enterprise value and signals to creators what types of experiences are rewarded. The community feedback on our recent algorithm changes has been overwhelmingly supportive, with a feeling that we are tilting more toward "evergreen" games that retain players long-term, as opposed to short-term "cash grab" games. We made two decisions: extending the signal measurement window and using that feedback loop to update the algorithm, even while running more tests. The discovery system has become a self-improving system that estimates the "user-game" pairing yielding the best long-term retention. We are balancing the optimal mix of long-term retention and long-term monetization. In experiments, we see short-term monetization decline, followed by an inflection point and long-term improvement. We are pressing forward because this directly measures retention, not projected predictions. With the launch of Build, which dramatically increases content creation, we believe the algorithm is resilient enough to surface high-quality Build games for all users without flooding them with low-retention, AI-generated games.
Q: How will the cost structure change during the transition to the over-18 demographic and new content, and what will the financial model look like in 12-24 months?
A: Two points. First, our long-term view on margin potential is unchanged; the drivers for margin expansion persist. Bookings growth will generate fixed cost leverage, the business mix shift towards lower-cost platforms will improve COGS, and further improvements will come from fixed cost leverage in other parts of the business. Second, in the near-to-medium term, our investments in AI-driven product enhancements will keep infrastructure costs elevated. However, this spending is different from historical spending; it's more like a fixed cost driven by model training, unlike past infrastructure costs tied to engagement hours and user numbers. Over time, due to its fixed-cost nature and bookings growth, we expect to see leverage on these infrastructure investments. However, as the Q3 guidance shows, there will be short-to-medium-term pressure due to incremental cloud GPU computing needs.
Q: How do you view the evolving competitive landscape for time and attention, and how is the company positioning itself?
A: We believe the world needs more "play," which is a broader concept than just Roblox. Play is between people, involving creation, being together, communicating, and watching others. Our redesign of the app, bringing Build and Moments to the homepage, is part of this vision to mirror the physical world with a digital world supporting "play." We are optimistic that Build will generate a volume of content creation far exceeding anything from the Roblox Studio era, while professional developers creating complex works will also leverage Build's capabilities. The definition of "game" is changing. We are entering an era where, similar to how video editing became common, AI will accelerate the creation of games, from 2D puzzles to complex 3D multiplayer games, on mobile devices, enhanced in Studio. This will reshape the entire game landscape. The vision is that everyone on the platform is a creator and builder.
Q: As free cash flow grows, how do you prioritize capital allocation between buybacks, organic investment, and M&A?
A: Our capital allocation strategy is clear: primary focus is on aggressively investing in organic growth, including R&D, personnel, tokens, and infrastructure. We are diligent in these investments but recognize the rapidly changing world, so we maintain ample "dry powder" for agility. Our balance sheet and free cash flow trends provide sufficient firepower for organic investment, buybacks, and M&A. Our historical M&A has focused on acquiring technical talent, which we can continue. Future larger deals are possible if they make strategic sense. Regarding stock price influence on buybacks, two points: Our buyback structure is designed to offset dilution from employee equity, so we buy more shares when the price is low and fewer when it's high, creating a natural moderating mechanism. We also have the capacity to accelerate or increase buybacks, which we continually evaluate.
Q: As the company no longer provides full-year guidance, can you provide boundary references for Q4 and the full year? Is 2026 still a growth year with full-year bookings higher than 2025?
A: We are not providing Q4 guidance due to the numerous moving parts outlined in the shareholder letter and our commentary. We feel good about long-term trends. We are navigating normalization after a high comparison base, making platform changes in discovery and safety, and launching significant new products like Build and Moments. Many of these initiatives are in very early stages, and we cannot predict their exact impact over the next three to six months. Our focus is on pursuing these initiatives as quickly as possible. You can assume our actions in Q3 and Q4 align with our long-term goals, but we cannot offer more specific guidance beyond what we have shared.
Q: Can you elaborate on the internal tests showing that longer retention compensates for lower hourly monetization?
A: We are running many experiments and iterating on algorithms. In A/B tests comparing new algorithms against the old discovery system, we see an immediate change in user retention, which has a compounding effect on our business flywheel by driving more users and engagement. However, dollars per hour, our monetization metric, is immediately and significantly impacted. Based on the curves from several weeks of data in each experiment, we believe a crossover point exists where the benefit of incremental retention will outweigh the short-term impact on bookings. We are not giving a specific timeline as we continue to iterate, and our focus is on finding ways to improve the algorithm to preserve retention gains while reducing the near-term bookings impact. The key is that by observing signals over a longer period and measuring directly, we capture long-term revenue associated with retention, not just early revenue. We are optimizing for a direct measure of long-term retention and long-term monetization, not over-extrapolating potentially noisy short-term monetization signals.
Q: Can you quantify the financial investment for Morpheus or the related cost headwinds this year?
A: Morpheus is a small part of our numerous AI initiatives. The simplest way to understand the cost is by looking at our Q3 margin guidance and the year-over-year margin compression. Roughly half is from fixed cost deleveraging due to lower bookings, and the other half is from these AI-related investments. These investments support new features like Build, Moments, safety improvements, and model training for Roblox Reality. The vision for Morpheus is to provide photorealistic multiplayer games, which doesn't exist today. We believe a hybrid path is the solution: 3D synchronization in the cloud plus client-side upscaling to achieve photorealism. The infrastructure costs for this will initially be offset by subscription revenue. The game will run in both a regular Roblox mode and a Morpheus video-realism mode, with access to the latter initially requiring a subscription fee.
Q: Will the newly supported 2D content be an acquisition funnel for non-platform players, or is it internal demand? How will it differentiate from typical 2D mobile games?
A: There's a strong alignment between what Build is doing and our goal of removing boundaries on content types. When users are presented with Build, they don't limit themselves to a 3D obstacle course or a specific puzzle game. There's a great intersection in supporting everything people want to build. Many users don't consciously distinguish between 2D single-player, 2D multiplayer, 2.5D, or 3D games; they see it all as "play." As we reach older demographics, there is ample demand for 2D. Any experience built on Roblox has significant advantages: multi-language support, global reach, social capabilities, our economic system, our infrastructure, and interoperability. This is about broadening the range of game types on the platform. Ultimately, it aligns with Build, where we don't restrict what people can create and show their friends.
Q: What is the difference in monetization patterns for the under-13 demographic that caused the impact this quarter?
A: The impact is more related to the comparison base. Looking at the high concentration of engagement hours from last year's viral, high-monetization hits, which were particularly popular with younger users, we see a mix shift. As those games have been replaced by games with more "normal" monetization levels, the impact is most pronounced in the younger user base. We don't see a structural change in the behavior of younger users; it's essentially a reflection of the current popular games not having the same extremely high monetization levels as last year's hits.
Q: What have you learned from the Incubator and over-18 initiatives, and how will these projects evolve?
A: We have a significant amount of content from incubator partners, studio partners, and the existing community. We view ourselves as a UGC platform but also recognize the value of building close relationships with top creators to guide them on what performs well. We recently introduced several technologies with incubator partners, including dramatically improved avatars and worlds that perform well on both low-end Android and high-end PC. We also introduced a technology called Slim, which allows complex avatars to run with high performance on mobile. These technical opportunities are significant for our partners. They also participate in the 50% DevEx premium for games targeting the over-18 demographic. The quality of content from our incubator partners, studio partners, and the existing developer community is very high.
Comments