Good morning. My name is Krista, and I'll be your conference operator today. At this time, I would like to welcome everyone to the Roblox Q3 2023 Earnings Conference Call. All lines have been placed on mute to prevent any background noise. After the speaker's remarks, there will be a question and answer session. If you would like to ask a question during that time, simply press Star, followed by the number 1 on your telephone keypad. If you would like to withdraw your question, again, press Star one. Thank you. I will now turn the conference over to Stefanie Notaney, Senior Director, Financial Communications. Stefanie, you may begin. Thank you, Krista. Good morning, everyone. Thank you for joining our Q&A session to discuss Roblox's Q3 2023 results. With me today is Roblox's Co-founder and CEO, David Baszucki, and CFO, Mike Guthrie. As a reminder, our shareholder letter, press release, SEC filings, supplemental slides, and a replay of today's call can be found on our investor relations website at ir.roblox.com. On this call, we will make some brief opening remarks and reserve the rest of the time for your questions. Our commentary today may include forward-looking statements, including, but not limited to, expectations of our business, future financial results, and strategy. Forward-looking statements are subject to risks, uncertainties, and assumptions that could cause actual results to differ materially from those described in our forward-looking statements. A description of these are included in our SEC filings, including our most recent reports on Form 10-K and Form 10-Q. You should not rely on our forward-looking statements as predictions of future events. We disclaim any obligation to update these statements except as required by law. During this call, we will also discuss certain non-GAAP financial measures. Reconciliations between GAAP and non-GAAP metrics can be found in our press release and supplemental slides. With that, I'll turn the call over to Dave. Thank you, and welcome everyone to our Roblox Q3 2023 earnings call. A couple of quick highlights: We're going to continue to talk about our delivering growth strategy, highlighting our vision of reimagining how we come together, and continued moving forward on our mission to connect a billion people with optimism and civility. We'll highlight some of the product velocity and new monetization opportunities we've been working on. We're also going to give an update on our commitment to deliver operating leverage, which we shared in our letter. Finally, highlighting really the breadth of our business, which is connecting people around the world, connecting people of all ages and all backgrounds, and highlighting the complexity of a business that spans so many cohorts. Today, for a first time, we are going to give you a little hint at guidance looking forward, and Mike will do that after my discussion. Diving into Q3, a real strong quarter all around. GAAP revenue, Q3, $713 million, up 38% over Q3 2022. Diving more into how we run the business, which is on bookings and cash. Bookings of $839.5 million, up 20% year on year. Cash flow from operations, $112 million, and $59 million of free cash flow. Our DAUs hit 70 million, which is up 20% year on year. Total hours of engagement, 16 billion in Q3, which is up 20% year on year. And from a liquidity standpoint, over $2 billion of net liquidity and $3.1 billion of total cash equivalents and investments. And finally, on a covenant-Adjusted EBITDA basis, we hit $81 million. So let's talk a bit about our growth, growth initiatives. This goes all the way back to our S-1 filing, which is connecting everyone around the world, a platform for everyone of all ages, Roblox everywhere, and a vibrant economy. On connecting the world, just want to highlight, not just the growth in many really big future areas around the world, but also the monetization acceleration in these areas. Japan, DAUs up 66%, hours up 64%, bookings up 174%. Germany, DAUs, 27%, hours, 30%, bookings up 75%. Brazil, DAUs, 23%, hours, 23%, bookings, 62%. India, a huge opportunity for Roblox. DAUs up 53%, hours 49%, bookings up 76%. We continued to also show great progress in our vision that Roblox is a platform for everyone, irregardless of age. On the DAU side, in our 13 through 16 cohort, we saw a 22% growth, 17 through 24, 27% growth, and 25 and up, 25% growth on the DAU side. Hours, very similar growth, 13 through 16, 23%, 17 through 24, 29%, and 25 and up, hours of engagement up 28% year-on-year. On our vision that Roblox is everywhere, and this goes to the vision that immersive 3D multiplayer in the cloud simulation that connects and helps people communicate should run on every device. We brought forward two new devices recently. First, Meta Quest, which as of October 31, had over 2 million lifetime installs. Then Sony PlayStation, which we announced at our developer conference and launched on October 10, has had over 15 million downloads in October. And our goal is really for Roblox to be everywhere. On the size of a vibrant economy, we're innovating in many ways there, continuing to grow our monthly unique payers. We had 14.7 million average monthly unique payers in the quarter, which is a record. We have some really exciting innovative extensions to the platform coming. One is, really in the vision of Roblox, giving our developers and creators the tools to run their own business. We are going to be introducing subscriptions for developers, so their experiences can host subscriptions. We plan on launching that this month. Advertising, we're going to share a lot more at Investor Day, but we are making amazing progress here. We've hired Stephanie Latham as our VP of Global Partnerships to help supercharge this business. We've also, in addition to portal and image ads, started testing video ads in Q4, and looking for a launch of that in 2024. We'll share more of this at Investor Day. But on the overall product velocity and the background, we shared a lot of our 2024 vision at our developer conference in September, and those videos are all available online for anyone who's interested. We have a really strong pipeline of great innovation. On some of the big things we've been working around, which include voice and avatar simulation to more and more host real-time communication. Our voice DAUs for the quarter are now up 240% year-on-year. On avatar side, we are now live with what we call UGC avatars, which is user-generated content for avatars, allowing the community to participate in the creation of avatars. We're now live with facial animation, and over time, we expect all avatars on the platform to support real-time camera tracking and facial animation. At our developer conference, we introduced a new platform called Roblox Connect, which is a showcase of communication technology that will be open source and shows off a set of APIs that allow creators to embed communication capabilities into their own experience, including on mobile, ringing the phone to join with friends. A couple highlights, we'll touch more on Investor Day. On the AI side, we continue behind the scenes to more and more incorporate AI, AI up and down the stack. And in addition to, of course, added capabilities like Roblox Assistant, which we rolled out in October, that you'll hear more about on Investor Day, and Roblox, other ways of generating things. I think behind the scenes, I just want to highlight the efficiency that this is bringing us, both on safety, on moderation. We're moving many of our moderation pipelines more and more to AI, and this is simultaneously increasing quality as well as cost, and you can see that in our increased operating leverage. Finally, just want to highlight that at Investor Day in a week, we'll be covering a lot more of this. And now I'm going to pass it over to Mike for some more discussion on our operating leverage. Good morning, everybody, and thanks for joining. I think Dave did a great job of covering the highlights. I just want to point out that it was another quarter of very solid growth, and that growth was accompanied with a management of the growth and expense in our business. In particular, our fixed costs, which we've been talking about for the last few quarters, which are personnel costs and infrastructure costs. Both of those grew at a substantially lower rate this past quarter, and infra and trust and safety grew at a lower rate than our bookings growth, which is something we had talked about and indicated would happen this quarter. So we're quite happy with that. That provides quite a bit of operating leverage, and we see that, we're now in a moment where we can see our operating costs growing at a rate that trails our top line growth, which is exciting and should provide, healthy operating leverage in the future. We'll talk more about that at Investor Day, I'm sure, on this call. And at the same time, we've now completed our Ashburn, Virginia, data center, and so capital expenditures in the Q3 were down significantly from both last year and last quarter. And, we now have enough capacity to handle, even significant growth in traffic all of next year. So, substantially less capital expenditures in 2024 than you've seen over the last two years, which is a good setup for, not just operating leverage, but also cash generation. Finally, as we mentioned in our shareholder letter this morning, beginning next year, in 2024, we intend to begin providing financial guidance. Typically, when we report our annual results each February, we will provide guidance for the ensuing full year and for the Q1 of that year. For each ensuing quarter, we will provide forward guidance for the next quarter and update our annual guidance. This is pretty standard fare. Thus, on our Q4 earnings call next February, we will provide guidance for the full fiscal year of 2024 and for Q1 of 2024. In May of next year, we will provide guidance for Q2 of 2024 and update the guidance for the full year, and so on. We will guide on revenue, bookings, net income, and margins. We believe bookings provides a timelier indication of trends in our operating results that are not necessarily reflected in our revenue, because we recognize the majority of revenue over the estimated average lifetime of a paying user. The change in deferred revenue constitutes the vast majority of the reconciling differences from revenue to bookings. Since our Q1ly guidance will be for Q1 of 2024, we expect a question today about our views on consensus numbers for Q4 2023, the last quarter for which we will not provide guidance. Since we haven't provided any guidance information, we just want to reflect that based on facts that as of November 7, mean consensus for Q4 2023 is as follows: bookings of $1.065 billion, Adjusted EBITDA of $173.2 million, which correlates to our calculation of covenant-Adjusted EBITDA, and revenue of $809.3 million. While we are not going to provide formal guidance for Q4, we are comfortable with the consensus estimates for bookings and Adjusted EBITDA. Concerning consensus revenue, we want to remind everyone that revenue is much more complicated to forecast and model than our bookings. Paying users can buy both consumable and durable virtual goods. Bookings related to consumable virtual goods are recognized essentially immediately, while bookings related to durable virtual goods are recognized over the average life of a paying user, which is currently 28 months. The split between consumable virtual goods and durable virtual goods changes every month. In addition, the average life of a paying user can also change in any given quarter. Finally, because of the deferral period, in any given quarter, the revenue recognized has much more to do with prior bookings than with current bookings. Thus, for example, the growth in bookings over the past quarter will largely be reflected in our future revenues. Just as an example, I went back and looked at our bookings in the supplemental materials. If you think about the 28-month average life of a payer, that's about 9-10 quarters. So if you look at the growth rate of bookings over the last 9-10 quarters, you'll see that it is, those growth rates are below the consensus revenue estimate, which is about 40%. So we expect revenue in Q4 when we do the work on virtual versus consumable and the average life of a payer, assuming there's no change to that average life in Q4, we expect revenue in Q4 will be approximately $740 million-$750 million or a year-over-year growth rate of approximately 28%. So with that, we're done with our opening remarks, and we welcome questions. As a reminder, if you would like to ask a question, please press star one on your telephone keypad. We also would like if you could ask one question and one follow-up. Your first question comes from the line of Jason Tilchen from Canaccord Genuity. Please go ahead. Great. Good morning, and congrats on the strong results. I was just wondering, exchange fees as a% of bookings were down sequentially. I'm curious, you know, if there's anything one time that drove that and just higher level, how you view continuing to reinvest in developer community, as you see leverages in other areas. Thanks. Hi, Jason. We absolutely top priority is to continue to invest in the developer community. In fact, the strong results that you've seen over the last few quarters is a direct result of an explosion of creativity, new content, existing content being updated and upgraded, appealing more and more to our user base. So top priority is to continue to invest in our developer community. The sequential decline is primarily related to an issue with our prepaid cards. So there was an ability, and we can go through this in more detail if you want to get on a call with us one-on-one. But there was an ability, until recently, to actually buy Robux at a discount through prepaid cards in certain currencies. And as a result, the bookings number and the DevEx number... The DevEx number was artificially inflated versus the bookings number. The user was getting extra Robux per dollar. We've now closed that, and they're now buying Robux at the right price. That means they get slightly fewer Robux, which means the DevEx related to those bookings is slightly lower. But going forward, that's that loophole effectively is closed, and we will continue to be driving lots of opportunities for developers to grow the overall dollars in the system and the percentage of the economics that they get. And we'll talk more about that at Investor Day. I'm happy to discuss it on this call as well. Yeah, and we highlighted at our developer conference the growth rate of the top 10, the top 100, and the top 1,000 developers, and those all continue to grow quite rapidly as far as their revenue. Great. It's really helpful. Just one quick follow-up. In the shareholder letter, you called out bookings growth really strong in Western Europe and East Asia. Just curious if there was any underlying drivers of that performance that stood out in the quarter. Thank you. You know, across the Western Europe and East Asia, we have a few benefits that are really pushing the bookings growth rate. Number one, we have payer cohorts that are getting more and more seniority, and we know over time, as payer cohorts are on the platform for longer periods of time, they tend to monetize better. And so a lot of those payer cohorts are really hitting their stride in terms of maturity. Those are wealthy countries, so they have a lot of purchasing power, and so they, we've done very well there. We have seen lots of new content that has popped up in those markets. And so, lots of growth there. We're really growing across all ages in Western Europe and East Asia, from all the way from the youngest users on the platform to aged-up users. I wanna highlight also core technology, less on the monetization side and more on the user engagement side, is also underlying some of this. The quality of our platform in Japan has increased rapidly over the last few years. We highlighted semantic search and some of the other things we've been doing there, so this is also a result of product quality and the quality of our automatic translation system. Your next question comes from the line of Bernard McTernan from Needham and Company. Please go ahead. Great. Thanks for taking questions. Maybe just to follow up on the last one, you guys have been talking about Japan for a couple quarters now. Are there other kind of wealthy, under-penetrated countries out there that, is next on the list to target? Well, Germany, Spain, Italy, I mean, all throughout Western Europe, France, they're all growing really dramatically. And as you know, Bernie, it's less about an immediate target, and it's more about an organic growth that's driven, as Dave said, by great product technology, the fact that translation and search are better, the fact that there's a local developer community sprouting up in those marketplaces. So, it's organic and we believe very sustainable because it's organic. And so we're targeting really users around the globe, including more and more users in the US, Canada, and markets that we're already very strong in. But the countries in Western Europe and East Asia just are showing particularly strong growth and very healthy monetization. And highlighting, when we talk about product quality in Japan, that's platform-wide product quality, not Japan-only product quality. And that goes all the way down to raw product performance on low-end devices, for example, which can help accelerate countries like India. Search can help accelerate all countries. So really, we target the platform as a whole generally. Understood. And then just on the cost saving initiatives, in talking about internal efficiencies, especially in infrastructure, obviously, you guys are doing a great job pushing on that. It feels like just from the shareholder letter, the commentary, that there might be maybe more to do, or you're discovering that there's more to do. Is that the right read, especially in infrastructure? Just we see internal efficiency initiatives. It sounds like it might be a broader push than just growing it below bookings, but just wanted to double-click on it. I want to highlight a couple efficiencies that we actually get from a quest for quality as opposed to a quest for raw efficiency. More and more, as our safety platform moves to AI, it's been driven by a quest for just increasing the quality of everything we do around safety and moderation. A by-product of that is a lot of efficiency. I think on the infra side, as we get into a high level of resilience, that gives us a lot more room going forward to grow without a lot of CapEx. So some of these are less discoveries than natural evolutions. May I just add, Bernie, one last thing to that? The growth rate that we had to build into through COVID was so significant that, you know, we made a decision to invest heavily and aggressively, and we're happy that we did. And obviously, the platform and the scale of the platform is much, much larger and much more capable than it was before. But now we're sort of through that macro adjustment of hyper growth that the economy took when the pandemic started. It should be much more rational and reasonable and predictable growth, and that allows us to invest and be more efficient, really, throughout the income statement and the balance sheet. I want to highlight one other nuance on efficiency, and this gets into what we're doing on voice safety and voice moderation. We're doing very sophisticated work to help with voice safety and civility, and we're doing this on our own infrastructure. And there's incredible efficiency by using the same infra that we use for real-time 3D simulation of all of our players, using that exact same infra for inference for voice safety. Understood. Thank you both. Your next question comes from the line of Omar Dessouky from Bank of America. Please go ahead. Hey, guys. Thanks for taking the question. I actually wanted to double-click a little bit more on this infrastructure question, because the numbers are quite remarkable. So if I just look at your results, it looks like your hours in the Q3, your total engagement hours were up 15%, but your infrastructure cost per hour was down 18%. You know, so that's pretty strong, and your infrastructure costs came in well below my estimates. So is there any one-time events that caused this, you know, very high cost savings on that line? And how should we think about your infrastructure cost per hour, kind of going forward? Would they be at a similar rate as they were in this quarter? Or, you know, will they kind of come back up to where they were in the Q2? And then I have a follow-up. I, Omar, I actually feel the one-time event has happened in the past, which was on core data centers, a fair amount of CapEx to build out resiliency. And we're through that now, and we have resiliency on core data centers. On edge data centers, where we run simulation, once again, we've got great efficiency there on that capacity. So I don't think there's a forward recurring event. I think we've done a lot of that in the past. Yeah. Then, Omar, as it relates to—so first thing, you, you had a comment about hours growing at 15%. I, I wanna make sure I understood that. I think the hours grew at 20%. But just- No, quarter. I'm just talking about quarter-over-quarter, Mike. Quarter-over-quarter- Oh, okay ... your hours were up 15%. Yeah. Yet, your cost per hour was down 18%, quarter-on-quarter. Okay, understood. So we're gonna show some more data at Investor Day, but one of the factors is what Dave just talked about, which was investment, building a lot of capacity, not unlike building a factory or anything else, and now absorbing that capacity. The second one is challenging the team to focus on a cost to serve target that declines over time, and to get more efficient as an engineering organization. And they've done an incredible job, and we believe have plenty in front of them, and they're gonna continue to drive down this metric of cost per thousand hours served. So, we actually see continued leverage in the infrastructure line, to answer your question, and that should be reflected in the future in lower, lower numbers as a percentage on the income statement, lower numbers as a percentage of either bookings or revenue. Okay, awesome. Okay, and then maybe just focusing on, you know, sort of your top line and your guidance. You know, what, what was it that increased your confidence and your ability to guide accurately, you know, as compared to, let's say, three to six months ago? Well, I wanna do a shout-out to the whole financial modeling team. We've always had a fair amount of confidence in our ability to model the future. So this is less about confidence. Mike maybe can comment to the real, real world. Well, well, first of all, Omar, you said the existing guidance. So we haven't given any guidance yet. We're gonna guide for the Q4. I merely commented on Q4. A number of reasons, there's certainly demand. We talked to a lot of investors who would really like a deeper insight into our forecasting, and the best way to do that is through guidance. And I think having gone through such a big cycle of COVID and coming out of COVID, and the big investments that we've made, and now with margins and operating leverage coming out the other side, it actually does become easier to have that conversation and show people the models going forward. So I think it'll be a healthy dynamic for everybody. Yeah, and I wanna highlight the complexity, not really complexity, but the breadth and the scope of our vision internally. We're building a platform that is connecting people around the world in a bunch of different regions and different countries, a bunch of different age groups, and more and more are moving to several economic models that support that virtual currency, advertising, and more to come. And then on top of that, we have complexities on the GAAP revenue side, as Mike mentioned, as far as deferred revenue. It's a fairly broad, ambitious vision, and we think guidance can help that complexity. Awesome. Thank you, guys. Thanks, Omar. Your next question comes from the line of Clark Lampen from BTIG. Please go ahead. Thanks, morning. I have a question on average spending rates. You guys highlighted a number of products throughout the prepared remarks that seem, you know, whether in the late stages of, I guess, sort of pre-release or early release right now, that seem likely to be accretive to spending. Are we at a point now where, you know, maybe those products collectively can offset some of the mix shift headwinds to monetization we see from international expansion, and maybe average spending from here sort of lifts going forward? That's question one. Hey, Clark, good to talk to you. First thing, I would like to highlight that in the Q3 of 2023, so the average payer monetized at $19.02 over the quarter, which was up from $18.11 the prior Q3. So, so monetization on a payer basis has been creeping up over the last few quarters on a year-over-year basis, and is basically catching up to those peak moments in the pandemic, when monetization was incredibly high. Overall, if you look at it on a DAU basis, some of the discussion around the prepaid cards that I mentioned earlier has been a bit of a drag on it, so I think we'll work through that as well. So, yes, I do believe I'm strongly optimistic about monetization. Overall, in addition, we've—as Dave just highlighted, the platform is built now, 70 million daily active users with a virtual economy that is very large and growing at a very healthy clip... and it is set up, as we've always talked about, and goes back to when we went public, that we will grow the virtual economies or vibrant economies on top of our platform, including advertising and other things that we'll talk about at Investor Day. Those things, as they funnel in, obviously, will also move monetization up materially. So we're pretty excited about that. So overall, as we look at monetization, whether it's per user or per payer, we like the trends that we're seeing right now. Makes sense, and I'll take- There's a little bit of a talk track sometimes about the geographic mix shift, which I think is what you're getting at. I just wanna remind everybody that in the U.S. and Canada, as an example, as users grow in terms of older users, they monetize better. There's a mix shift going on even in our core. And so even with a younger user base, where monetization and bookings are growing more slowly, we've definitely seen a positive mix shift to older users who do spend more. So it is not necessarily the case that just because other markets are growing faster, it has to have a detrimental effect on the rate of bookings per user or per payer. And overall, bookings growth is really what we focus on. Understood. You talked about economy changes. You know, also, it looked like there were some adjustments that were made to sort of packaging and then the Robux per unit of currency values, earlier this quarter. Could you talk about, you know, maybe how those packages are gonna sort of affect the user and developer community, if at all, and maybe secondarily, how, how could those changes, I guess, impact the P&L moving forward? Thanks. Yeah, I just wanna highlight, you may be seeing small experiments that we're constantly running on the economy. We don't have any big plans for a massive economy shift. Consider these nudges right now. Thank you. Your next question comes from the line of Eric Handler from Roth MKM. Please go ahead. Good morning, and thanks for the question. A couple of things on advertising. First, I imagine it's very small, but were there any advertising contributions in the Q3? Hey, Eric. There were, and you're correct, they were very small. Okay. Um- I do wanna. A little bit of a preface to Investor Day. While the actual ad revenue is pretty small, there is a vibrant economy around brands and the content that brands are publishing on the platform. And, there is a virtual economy related to brands, just like there's a virtual economy. Or the virtual economy around brands is integrated into our virtual economy overall. So while the ad dollars themselves were small, they're already contributing to the platform. Great. And then maybe I'm getting ahead of myself here, and you'll, you'll talk more next week, but now that you're, you know, you're testing video advertising, you know, when I look at sort of the, the suite that you have for advertising, I imagine, you know, the digital billboards are probably the low end in terms of pricing. You've got this unique concept with portals, and now you have video. Can you maybe talk about the value proposition of those three buckets? Yeah. I'm gonna take a step back on the visionary value proposition, and then we can, you know, paint a picture a year or two out and back into it. And once again, this is no promise of product delivery or any future revenue. But the vision really is ultimately what we do in real life. For example, hanging out with friends at a shopping mall, going into various locations, browsing, trying on, and literally buying, is gonna be fully replicated on our platform. And ultimately, that buying behavior will go to both virtual goods as well as someday physical goods as well, and this will all happen on the Roblox platform. So advertising is a great way for brands, and we've mentioned some of our great partners, Gucci, Vans, Nike, and whatever, to bring essentially friends traffic to their 3D virtual destination, where people can interact with the brands, watch videos, try stuff on, and then ultimately buy virtual and physical goods. So we, we actually think that, that advertising is just the first step to a really a fully closed loop, long-term shopping solution. And Eric, just a quick comment, because we're gonna cover this more on Investor Day, and Christina, who you know runs this business at Roblox, will talk more about it. You're really talking about portal ads and video ads, and there are gonna be more than one type of video ads. We're gonna talk about multiple products. I wouldn't underestimate the value of video ads. Portals are very interesting and unique. We also have a massive user base and a very interesting age demographic that's highly engaged. So we really see value for brands and CMOs to look at various products on the platform. Ultimately, we're incredibly excited about portals, and we've seen good activity. When brands engage at that level, where they have their own, persistent experiences, it really is a base for a lot of other interesting engagement and monetization in the future. There's a real, there's a real breadth of products. Great. Thank you very much. Thanks. Your next question comes from the line of Brandon Ross from LightShed. Please go ahead. Hi. You recently launched on a couple of new platforms, including Sony PlayStation. We're just wondering if there's any way to size the impact- ... from those type of launches on DAU or in engagement, and if there's an update on any more platforms to come, and then I have a follow. I would say we wouldn't share any platform by platform engagement estimates. I think it's safe to say it's doing better than we expected, and I think the reason it's doing better than expected is we have so much amazing content that is already running on phones, tablets, computers, other consoles, that that's really just an accelerant of that existing content with a new destination. I actually, Brandon, I would add to that. It's also, you know, a little bit more of Metcalfe's law, so that those people on the network can connect to other people on phones and desktops and other things. And so we are seeing an impact that maybe we can't just explain with a discrete view of what's happening on PlayStation. It's really the benefit that we're seeing all over. But we're going to talk about download numbers next week and everything. They're very strong and very healthy, so we're super excited about it, but I think it's sort of multiplicative. On future platforms, the vision would be immersive 3D multiplayer communication and connection technology ultimately should run on any form factor and any device. Great. And then on the DevEx, just to clarify from earlier, was it the loophole that made DevEx only 20% of bookings in the quarter? And then, like, generally on DevEx, how are you thinking about your profit and cash flow generation, off the leverage and kind of cost control versus giving back to the dev community? Kind of what goes into that decision, and how will advertising affect that? So there's a lot in that question. So, first and foremost, our goal is to continue to push high growth in DevEx to the community. The investment that we make there is why we have great content and growing content, and so we're always looking for ways to grow that, overall. Our desire, Brandon, on leverage and cash flow is really to leverage the efficiency in our fixed costs, which is around headcount and infrastructure. Those are the primary fixed costs in the business. Generally, over the last few years, as we have seen efficiency in COGS, we've tried to share those pretty much dollar for dollar with the community, and that's been really effective. Historically, if you look back three or four years, the DevEx rates were quite a bit lower than they are today. So we're in a much more sustainable place and an ability to continue to push more of those economics to the developers. As we open up the economy, things like advertising, we fully intend for the community to participate in those economics. And so we're really excited about the scale and the growth and the ability to continue to invest in it, because that's really what's driving the flywheel, is content and users, and that investment that we make pays back. So we're always trying to grow it. Awesome. Thank you. Thanks, Brandon. Your next question comes from the line of Matthew Cost from Morgan Stanley. Please go ahead. Hi, everybody. Thanks for taking the questions. Maybe I'll start with one for Dave and then one for Mike. Just on the AI tools, you know, you mentioned the moderation pipeline shifting more in that direction and then a lot of exciting use cases, you know, for those new tools. Are you still on track to build everything you're planning to build from an AI perspective on open source or from scratch? And is that from a CapEx and headcount perspective fully baked into your view from here? Yeah, I think we have so much breadth in the AI. You know, we've highlighted, we've got 70 ML pipelines running right now inside the company. That goes all the way from our safety systems, moderation systems, real-time translation systems, gets into our search and discovery systems, and then gets into more of the user-facing things, like Roblox Assistant. I think the big takeaway is we want to run hot—super high volume inference, for example, voice, as much as possible on our own hardware, so we can do this extremely efficiently. And I would highlight on our edge data centers, using as much as possible the same hardware to run inference, it gives us the ability in parts of the world that are not as active, to use that for inference. As far as the collection of tools we're using, I would say generally, yes, yes, yes, open source. The market's in a really interesting time as far as what people are building and training and tuning and distilling and optimizing internally, and we're building a full AI internal platform that many of our teams will then use for high-performance inference. Thanks. And then from Mike, I just want to revisit the goal of $800 million of DevEx that you guys mentioned last quarter. I think in order to hit that, you know, versus consensus, you'd need to do 26% of bookings as DevEx next quarter. I guess, are you reiterating that $800 million dollar target? So, I think we've talked about this, Matt, many, many times, so it would be unlikely for us to hit $800 million this year. But if you look at what we are accomplishing this year, it's a massive pool of capital that has grown at very high rates over the last few years, and we're super excited about what we are delivering to developers this year, just like we're excited with the top-line growth in the business and the operating margins of the business. You know, incredibly proud of the numbers that we're printing and incredibly proud of how the developers are sharing in that. And, you know, we're gonna continue to drive efficiencies across the business, and that's what we're gonna focus on. Great. Thank you. Your next question comes from the line of Eric Sheridan from Goldman Sachs. Please go ahead. Thanks so much for taking the questions. Maybe two that are sort of follow-ups to topics we've talked about, and hopefully not front running next week. But, to put a finer point on it, when you think about the developer community and lining up platform investments to meet the needs of the developer community, what are your highest priority items as you move out of 2023 and into 2024? And then to come back on the infrastructure investment, is there a way to think about capacity utilization in your data center network and, and how much, excess capacity you have right now to better understand engagement growth and, and filling that capacity, to where, CapEx could be, you know, coming in under expectations like now versus a more normalized trend longer term? Thanks so much. Yeah. Hey, I'm gonna go really fast on three points. On core data centers, wanna highlight just our infra platform teams ruthlessly at improving quality, efficiency, performance of all of our componentry there to really squeeze out as much as we can on the cost side. Also, on our cloud expenses, we're being very, very thoughtful there, moving as much cost internally as possible. On the edge capacity, just highlighted ability to run inference on the same grid where we run real-time simulation, so we can run more and more inference for super, super low cost. So a little bit of a vision there as well. And Eric, I'm not gonna give you a direct number on capacity, but I would suggest to you that our capacity has quite a long way to go. It's not, you know, through 2024, for sure, before we would have to start thinking about adding capacity. So we're in a really comfortable place right now. We have time for one more question. Tom Champion from Piper Sandler, please go ahead. Hi, thanks. This is Jim on for Tom. Just a quick one on cost of revenue. So we've seen this shift toward bookings through lower fee channels. I guess, can we sort of talk about has that trend continued, and how should we think about modeling, like, leverage in that line going forward? Thanks. I'll go first to really high level, and, Mike, you can talk about modeling. You're highlighting more of the complexity. Once again, consumable, durable, different mix shift. As an example, developer subscriptions on the platform over time, with some of our partners may have different COGS than other type of Robux purchases, which bodes well for COGS. Our prepaid business bodes well for COGS as well. Mike may talk more about the mix shift. Yeah. Hey, Tom. So we've definitely been able to provide some leverage over time. There are some interesting areas where we could dive in, but generally, our real focus on operating leverage right now is infrastructure and personnel, and we're doing a really good job there. And those are numbers that will, I think, drive the vast majority of margin improvement over the next few years, next couple of years, at least. We're gonna talk more about that at Investor Day. I think you're comfortable modeling COGS at about where they are today. For now, that's a safe bet. That's a fairly safe bet. And think about operating leverage coming more from our people costs and from infra costs. Great. Thank you. Those are much more in our control. Mm-hmm. Thanks. I will now turn the call over to Stefanie for closing remarks. Thank you for joining us today, and that's a wrap for us. This concludes today's conference call. Thank you for your participation, and you may now disconnect.