Roku, Inc. ($ROKU)

So good afternoon, everybody. Welcome back to the conference. Rob Coolbrith from the Evercore ISI Internet research team. Really pleased to have Roku's CFO, Dan Jedda, here with us today. So Dan, good to see you, and welcome to San Francisco.

Thank you. Thanks for having me.

Great. So I guess maybe we can start with what's relatively new 6 days ago, exciting for me personally. I've now -- I've just told Conrad, I've now kitted out my entire house, not the kids' rooms with Roku Sticks, and we're excited about the new operating system. I bought them at Walmart, by the way. So I don't know if that helps, but that's where I bought. I'll have a question on that later. But, first of all, so when you talked about it in April, it sounded like it was still early, you're still sort of kicking the tires on it. It sounded like it been testing at scale. But I've gotten feedback from some people, hey, this is a little bit earlier than we expected. So people ask me, was this baked into what you're expecting for Q2? And then if you could talk about how this drives the business?

Yes. You're referring to the new home screen, I assume.

Absolutely.

Yes. So let me talk about the home screen. We're very excited to roll it out. We have been talking about it for, I would say, 2 quarters. We've been testing it a long time because we wanted to roll it out right. Our home screen hasn't changed a lot prior to 2024. And so this is a pretty big move for us. We put a content row at the top of the home screen in 2024. We put video in the ad unit earlier, and I think that also was 2024. But by and large, the home screen remained relatively static. And one of the areas that we decided to focus on was making the home screen much more personalized because when we simply put a content row at the top of the home screen, we saw engagement go up. We saw subscriptions that we monetize go up. We saw hours into the Roku Channel go up. And it was a win for streamers as well because they love having that personalized row. So we basically took that and said, "Hey, let's do -- let's look at the whole home screen." So we spent many quarters, months and quarters testing the right home screen, which is what we're rolling out now. It's about 20% rolled out, and we'll slowly roll it out over the next couple of months. And that new home screen does a couple of different things. It makes the whole top of the homepage almost all above the fold, very personalized. It also collapses the left nav. There was a left nav unit that is now collapsed. So you start your experience with the home screen, you just have to right click over. So a couple of things that I love about it, both from a streamer experience standpoint, it's much more personalized, it will drive more engagement into areas that we monetize. From a streamer experience standpoint, that's great. From my CFO hat, I love it because we can monetize the home screen better. And let me just give you a couple of examples. First of all, now -- that video ad unit now automatically plays if it is a video ad unit, sometimes it's a static display ad. Sometimes it's a video ad unit, but you no longer have to click to the right to see that. It simply auto plays. It's video right from the start. That means more ad impressions, that means bigger reach, that means more ways that we monetize that ad unit. Also the -- within the personalized home screen now, we will drive more streamers into areas that we monetize, whether that's more into the Roku Channel, more into Howdy, more into Frndly, more into subscriptions, premium subscriptions primarily, which I'm sure we'll talk about. We will drive more engagement into that. Now that's not why we did the home screen. We did a home screen literally to make it a better streaming experience, but we also are going to benefit from more monetization. And what you won't see, which is also very exciting, which I believe will eventually happen is now that we've laid out this home screen, we can add more monetization into it with like a biddable in-tile app unit, which is something I'm very excited about. And we'll eventually do that. We're not going to do it right away, but we'll eventually build in biddable ad units, whether they're guaranteed CPA, which is a cost per 1,000 impressions or an actual auctioned biddable ad unit, we will eventually do that, which is another very unique way that we can not only personalize because it will be very endemic to the streamer experience, but it will also be a big monetization, a new ad product for us that we can monetize. We don't have a pure biddable ad unit yet on the platform. And so that would -- an area like that would be perfect for our first.

Great. So the other questions that I've gotten about this are -- so where do you think you are in terms of innings on home screen monetization today? And how much runway do you see for improvement there?

Yes, the home screen -- so 125 million people start their experience on our home screen. So when you have that kind of reach, there's a lot of runway ahead of you to figure out how to optimize the home screen. You can -- we're going to be very smart about it. We're not going to have it be like a plastered ad unit like a NASCAR race car ad units everywhere. We would not do that. That is not the right streamer experience. But you can make it in a way that is both positive for the streamers and positive for the monetization. So, like I said, I think we're in very early innings of the home screen monetization. Again, prior to 2024, it was very static. It was just app tiles. There was an ad unit, there was a static ad unit, and it was app tiles. It's now changing quite a bit, and there's a lot more we can do. And the flexibility of the home screen is now dynamic. So imagine like if I take it to its end degree, like imagine where everybody in this room gets a different home screen experience based on true personalization of what they're watching, what we think what our algorithms would tell us that they like to watch or should watch, whether it's areas that we can monetize more using GenAI to truly personalize like that's the -- that's down the road, like it can be very personalized. Right now, of course, this is literally like, I would say, very early innings on our focus on home screen monetization.

The other thing I've noticed about -- and my one ask on customization is maybe if I could change the color, purple is not my favorite. But yes, the one thing that I've noticed about is, okay, lightning fast actually get to watching something, which is, obviously, sometimes you'll be able to [indiscernible] and bargaining and whatever about what we're going to watch. So what have you seen in the test in terms of actually getting people more quickly into streaming content? And then what kind of impact do you think that could have on actual monetizable hours?

Well, sometimes like we may intentionally not put them right in streaming content because they don't know what to watch. We might put them into the NFL zone where they can pick from a variety of games that they want to watch. Imagine like that's the power of owning the OS. So maybe you don't know what you're going to watch. By the way, some of this is like a testament to the personalization and the algorithms learning over time. But yes, we -- whether it's a continued watching, whether it's you may want the subscription or you may want to watch this show or if it's a bit of ad unit where it's like a content partner that says, hey, I think this person should want to watch this new show like Marshals and actually shows an actual creative to go watch Marshals and sign up for Paramount+, like we can do all that with this new home screen. So -- and I do think like there are many times where streamers come in and they are not quite sure what to watch. By the way, we see this with FAST a lot. I mean that is the -- that was the reason why FAST was so successful. You go, you pick a FAST channel that you want to watch and you just start watching it because you're not quite sure. The home screen is not FAST, it's personalization, but the theory is still the same, like a lot of folks don't know what they want to watch or we can show them something that they hadn't thought of and they want to watch.

Got it. I want to switch gears a little bit and talk about some of the DSP integrations. Obviously, you've done -- you've gone from closed to open in a hurry, and you've done a fantastic job there. We often get the comment from people in the industry that this is one of the best companies out there to partner with. So I wanted to start with the Amazon DSP integration. Again, the inning question, how much further can you take that? And how do you think about the progress that they're making in terms of bringing their demand to bear against your inventory?

Right. So first of all, like all the DSP integrations we've done, and we've now integrated with all of them. We most recently announced DV360, which will launch in H2. We've signed the deal and announced that in partnership with Google. But specific to your question on Amazon, I get this question a lot, like how much is Amazon contributing. Let me just back up and remind you of our DSP strategy. You're right, we were close. This is a big difference between the Roku of 3 years ago and the Roku of today. 3 years ago, if you wanted to transact through a DSP, you had to come through our DSP, a product we call OneView, which is an acquisition that we made, which is very limiting. And we pivoted that strategy in probably late 2023, early 2024 to say, hey, let's open this up. We have so much inventory. TRC is doing so well. We have partner inventory that we can sell. Let's open up our demand-side platform strategy to more partners rather than force an advertiser who wants to run through a DSP to come through our DSP. So -- and prior to that, it's -- I'm being intentionally vague, but prior to that, like the DSPs were considered competitors. So Trade Desk would be Amazon now, Yahoo!, they were actually competitors. That's how we view them, not as partners. We've completely flipped that. And we've integrated with all the DSPs, and we've integrated with as deep as they're willing to go. So if they have a unique identifier like UID 2.0 for Trade Desk, there's an Amazon unique identifier. We will adopt those unique identifiers and go as deep on the integration as they're willing to go. The one unique difference is Amazon, which is platform-wide, not just Roku as a publisher, but from a platform side. So that is a fundamentally different shift, a very -- that is a full-on pivot on where we were. So you fast forward to where we are now. We're integrated with all of them. Amazon launched in Q4 and is ramping. And to answer your question, because I'm not shying away from it is, as the Amazon DSP ramps, so will Roku in partnership. It's a long-term deal. It's a platform-wide deal. It's -- we're fully integrated with Amazon, where we can match our customers to their purchase graphs and Amazon can do multiple things within their DSP because we allow them to from our platform for anyone that calls from our platform, we will help Amazon recognize that, and we will monetize that with Amazon. So as Amazon DSP grows, so will we -- we'll grow right along with them. So part of the answer to your question is it really depends on the success of the Amazon DSP as far as what inning we're in. I personally think Amazon will do very well in the demand-side platform. I've got some experience with that. I came from Amazon, and I think they'll do very well. But we do not tell our advertisers where to transact. We integrate with all the demand-side platforms. And wherever the advertiser wants to transact via DSP, we will be integrated and they can buy Roku Media through the DSP.

Got it. So I think you already touched on Google a bit, but I wanted to maybe put a finer point on that. So recently announced Confidential Publisher Match. I think you were the launch partner, maybe the only announced launch partner. So that's, I think, a credit to you and your ability to partner there. But it sounds like it's going to be ramping over the course of H2. Do you think that's going to be more or less up and running for the holidays?

It will be up and running. Again, but to that question, like the success of DV360 is predicated on how well DV360 does as a demand side platform. So I would say it this way, like there's this idea that -- there was an idea, call it, I want to say 2 years ago that the walled gardens, I lived through this at Roku, where the walled gardens were going to rule all of CTV because the walled gardens had all the inventory. And by walled gardens, I mean Amazon, I mean Netflix and YouTube for all intents and purposes. They were walled gardens, and they were the ones who are going to have all the inventory and everybody else is in tough shape. And we knew that, that was not the case. We knew that our scale would give us an advantage over -- to partner with the DSPs rather than compete with the DSPs. And so Amazon did announce a DSP, and they don't just focus on Prime Video. They have a whole DSP strategy, and we're partnered with them. Trade Desk, we've been partnered with them for a while under UID 2.0, which is their hashed e-mail that we've adopted. YouTube is no longer going to be complete -- going to be a walled garden in so much as DV360 will incorporate YouTube, but they'll incorporate all of the publishers as well. We're integrated with that. You could go to Yahoo!, you can go to Wurl. You can go to all the DSPs. We're integrated with all of them. And so I say this in so much as there's no -- the walled garden approach is really -- is no longer accurate to say the walled gardens will win in CTV. What will win in CTV are -- is the most performant ad inventory. And that is something we strive to do is have the most performant ad inventory because every streamer, all 100 million households plus, 100 million-plus households are logged in. We know who they are. We know what they watch. And we can integrate our -- we have amazing first-party data. We integrate with other outside data and measurement companies to make our platform very performant. So if an advertiser wants to transact through Amazon or through DV360 or through Trade Desk, we will be a performant platform for which they can bid on our inventory on the programmatic pipes, and we'll do very well on that.

Yes. I think that -- as I was setting up all these slides -- I, of course, got the question about my ACR data. And I like -- I'm a big supporter of the company. But maybe you can talk a little bit about that, some of the unique advantages of operating the platform and in terms of what you're able to offer advertisers in terms of reach and global reach and frequency control and things like that.

Yes. So it's a great point, and it's one I think is really important if I take a step back because this is also something that was talked about in 2023 is like, hey, there's companies out there that have massive IP, call it, Luca and Pixar or Stranger Things. There's -- all the NFL, which is owned by 3 or 4 players with Prime Video coming in with Thursday Night Football and you have Fox and you have Paramount and you have ESPN and now you have Netflix. And so -- and there was like all this massive IP out there. And I got this question, what is Roku's advantage? Well, we have 100-plus million streaming households. And they start their experience every day, watching multiple hours a day with our UI. So we control the user interface. That is our competitive advantage. That is our version of the NFL, if you will. And I would say, over the last 3 years, we've gotten very good at monetizing that -- those 100 million streaming households in a way that's beneficial to the streamer and a way that's beneficial to us from a platform monetization standpoint because we have -- again, we have the data, we have the ACR data. We have, of course, what you watch on our platform within our owned and operated apps. We can integrate with other measurement companies. We can integrate other third-party data. And so we're in this unique experience to control the UI and therefore, increase areas that we monetize. I get this -- and as a litmus test, just so everyone understands like how powerful that is. The Roku Channel is the #2 app on platform, so -- in the U.S. So we used to say it was a top 5 app. Then we said it was a top 3 app. Now we say it is a top 2 app. People can -- people know who the #1 app is, but the Roku Channel without big sports budgets, without huge content budgets is the #2 app by streaming hours on the platform, and that is because of controlling the UI and nudging people where to go and where to watch content. It turns out that when you nudge streamers into certain areas, they'll go watch content if they know it's available.

Got it. I want to go back to the ability to partner and how that's influenced by sort of CTV industry supply dynamics. Now those have made changes over the last couple of years. There was a sense that when APV went to flip to free that, that had an impact on the market. It seems like that's largely digested now and the market is looking for incremental sources of supply. How do you think that plays out for Roku just given where you are in terms of fill rates, in terms of CPMs and so forth?

Yes. So again, I think you're referring to the fact that we had a lot of supply over the course of the last many years and...

In the industry, yes.

Yes. And that is absolutely true. It turns out as the hours shifted from digital -- from linear to digital, the ad budget started to shift, and therefore, there was a lot of monetizable impressions. If you take a step back and just look at it from an industry perspective, first and foremost, the industry itself, it's about $90 billion in the U.S. Just talk to the U.S., it's about $90 billion. A little over 1/3 of that has moved to CTV. The hour is much more than 1/3. It's closer to 2/3 of the hours have moved to digital. So that's just supply. The demand is shifting faster and faster because now everything is available on CTV, whereas before, sports were the last holdout of linear. A lot of sports was not available on digital. Now that's not the case. As a matter of fact, now you have a lot of different sports, specifically in the NFL that's only available on digital. You have exclusive -- you have Prime Video on Thursday Night Games. You have exclusive games on Peacock. You have Netflix having games. So some of the sports are only available on digital. My point is all the hours shifted in advance of the dollar shifting. And now that's catching up. But yes, because of all those hours shifted, there was a tremendous amount of ad inventory on the market. I do think that continues to normalize because the dollars are now shifting quite fast from linear into digital. There's also new TAMs, if you will, that are moving into CTV, specifically the SMB market, which is highly performant. Maybe we'll talk about that. That's a whole new bucket of dollars that are moving into CTV from an advertiser perspective, and it's a huge opportunity. We're very excited about it. So you do have the secular tailwinds of the TAM expanding to help out with the supply. But what happened when that supply was more than the demand is CPMs did compress. That didn't scare us. We were just fine with that because we had a lot of impressions that we had to fill. But what I think ultimately happens now as more demand comes is I think the most performant ad inventory is what's going to be in demand. So what does that mean? It means the inventory that is -- whatever you're trying to do from a performance standpoint, whether it's reach, whether it's a KPI, whether it's an action, that inventory is going to be in high demand, and there's also lower demand ad inventory out there that I think will just generate very low CPMs. So I think you're going to have the CPM demand curve. The good news for Roku is we play along the entire CPM demand curve. We have ultra-premium CPMs like the Roku home screen, which is not available through programmatic. Roku City, which is not available through programmatic. We have mid-tier CPMs because we have great first-party data, which will have CPMs, call it, anywhere from $15 to $22. And then we can play in the low CPM space on some of our other inventory, if it's like something like an app install, which generally has lower CPMs. So my point is we have -- we play along the entire CPM demand curve. And so we're in a good spot from that.

It seems almost as though if you -- not just your commentary, but also more broadly, if you talk -- listen to the SSPs, like there was some kind of inflection point maybe around Q4, where it wasn't just the linear inventory coming -- or the linear demand coming to bear, but it was also some mix shift from traditional digital and so forth, display and video, that kind of stuff that was normally running in web or in mobile app that was suddenly coming to the market. Is that something that you've observed as well?

We have observed a lot of demand coming our way for video, which I would say is more performance-based. Again, it's performance means different things. From a brand perspective, that's going to be very different than an action-based KPI from an SMB. But we are experiencing a lot of demand when it comes to having high reach, high-performing ad inventory. Take the video on our home screen. We added video to our home screen about 18 months ago. That used to be a static display ad. That static display ad was really good for M&E advertisers, media and entertainment advertisers, but not great for any other vertical, if you will, simply because those other areas like health and wellness, like insurance, like retail, they wanted video, like auto. They wanted to see video in that. So what we did is we added video to that ad unit. And it had -- what it did is it opened up that specific ad unit, which was 99% plus M&E to an entire set of new verticals. Now we have Walmart in there. We have car dealerships, autos in there. We have insurance in there. We have health and wellness in that ad unit, and it's doing very well. Some of that was incremental demand. Some of it was a shift from in-stream video to this new video ad unit, but that's -- we're fine by that because that home screen video ad unit is very high margin. There's no cost of goods sold for us on that. So it's very, very high margin. It's a very high-margin ad unit. So we're happy even if it does shift from an in-stream video ad to this particular ad unit. But where I'm going with this is the demand that we see is for the performant-based inventory, which again is something Roku over-indexes on because, again, we've got such broad reach at over 100 million streaming households. Over half of broadband households in the U.S. have a Roku TV in their household. Again, they are logged in. We know who they are. And therefore, we can utilize that from an advertising perspective.

Got it. I just want to switch gears a little bit to the political environment, what your expectations are, maybe political and cyclical events, we can lump those together. But are you seeing any sort of early indications of what political could shape up to be for this cycle?

Yes. Our indications are we think it's going to be similar to the general election of 2024. It's a little hard to judge that right now. It's a little early. We're seeing bookings on par with what we saw in the general, but political comes very late because, of course, they need to figure out where to spend specifically, comes very late, hence, why a lot of it is run through programmatic, which is great for us because that's where we really do shine is on the programmatic side. A lot of political is hyper geo targeted. Again, that's something because everyone's logged in. It's great for us. We're very good at geotargeting. So we feel like we're in a good position for this year's political. And I would be pretty surprised if it didn't equal general election in this midterm cycle, but it's a little early for us to tell.

And the last cycle, how is that covered for you in terms of the pacing? Some people say it's Q3 weighted, some people say last 6 weeks...

It's end of Q3. It's definitely into Q4. October is a very big political month. It really does depend on where -- that one, I'm not quite sure how that is with the general versus the midterms. I think the dollars are going to be similar. How they're spent up until November, early November is to be seen. And I'm not quite sure on that. It depends on what the competitive elections are and what are the outcomes that all the spend goes against. But it does hit -- October and the first part of November are quite heavy. By far, the heaviest month is October. Now you do have a full Q3 spend because the spend does really start in, call it, the early part of Q3 and is in pretty heavy throttle at the end of Q3. So as you exit September, but October is a big month for political.

Okay. I want to switch gears to subscription, talk about that for a bit. You've had some sort of unique tailwinds over the past 12 months. You've obviously added Frndly, launched Howdy. You've added some more premium subscriptions. Just how are you thinking about the sort of sustainable growth trajectory of the subscription business? And just given some of those particular tailwinds and maybe some of those things are things that can sort of experience an S-curve, if you will.

Yes. So similar to how I talked about our DSP business going from 2023 to now premium subscriptions -- sorry, subscriptions, of which premium subscriptions is a major part of subscriptions has had a similar, call it, ramp, if you will, of focus. So we've had a subscription business for a long time. We have a payments product. A lot of what we call direct-to-consumer or D2C subscriptions, which is where you sign up with the partner, but on our payments platform through Roku Pay. And we've said before, we've monetized tens of millions of subscriptions. But it had not -- prior to 2024, it had not been a big focus. It didn't have a leader. There was not a lot of product, not a lot of innovation being pushed to it. There was no personalization at that point in time on Roku. So subscriptions wasn't a, call it, a top investment initiative. Now it is. We changed that. We said it very clearly exiting 2023 that subscriptions was going to be a high priority because it is -- because overall platform monetization was a high priority. So you fast forward to today, where premium subscriptions now is doing extraordinarily well, and it's doing well because of the way the premium subscriptions is really embedded in our UI. So if you have a premium subscription, the content is ingested on the Roku -- on Roku, and you see that content throughout the user interface, whether it could be on the personalized home screen, it will definitely be in the Roku Channel, all that premium subscription content is part of the unique experience. As opposed to a D2C subscription, the real only way into that is directly through an app. You have to go into the partner's app. Maybe you have a button, it could be the button. It's an app or a button. You're not going to see a lot of it on the content row. We might see a little bit, but most of the ingress into the partner is going to be through the app tile. That's not the case for premium subscriptions. The ingress is very little of it is through the app tile. Most of it is through the user interface, the user experience throughout the UI that we control. So we've done -- we have multiple Tier 1 partners sign up for premium subscriptions. We just announced FOX One. We had Peacock sign up. We had Apple TV sign up. In last year, we had HBO sign up for Tier 1. We have a very, very strong torso and tail of Tier 2, Tier 3 partners, all in premium subscriptions, all driving that premium subscription initiative. Also, what premium subscriptions does allow us to do is have more product focus on the subscription itself. I get a lot of questions like, hey, has it reduced churn? We haven't had a noticeable impact in churn yet, although I think that will change over time. But what it does allow us to do is do bundles, for example, which we're testing right now. It allows us to, again, put the experience into the checkout process, for example. If you're on another app and you just want to add a new app like Howdy, which we do now, where you're checking out on something, hey, you want to add Howdy for $2.99, you can do that. Premium subscriptions allows us to have much more unique features in the subscription business, more unique experience within the Roku OS, and that's driving a lot of the subscribers in premium subscriptions. It's been a big win for us. We also launched it in Mexico. We launched Howdy, which -- Howdy is a premium subscription partner. Frndly isn't there yet, but we're working on that. It will be soon. There's just some technical things we're still working through on the Frndly side, but Howdy is a PS partner. We launched PS in Mexico -- we launched Howdy in Mexico. We'll launch more countries with premium subscriptions. It's a huge positive for us.

And the streamers themselves are incentivized, it sounds like because they get more of their content displayed across the UI that drives more viewership, it's more ad...

It drives more hours into that content. It drives more experience. We'll have -- on the left nav, we'll have zones, we'll have sport zones, so we can have these zones have whether it's an NFL zone or as World Cup comes or soccer zone, et cetera, we'll drive subscriptions into those premium subscriptions.

Got it. And on Howdy, also -- maybe your newest Howdy subscriber right here?

It's doing very well. We're very happy.

It seems to be doing great based on the antenna data. Do you think that, that's a unique enough swim lane where you want to say, okay, more people need to know about this. Yes, we promoted on -- obviously, on the platform, on the home screen and so forth. And it sounds like you're also doing some to attach it to your other premium subscription sales. Do you think about putting more marketing dollars more generally behind it?

Yes, that's all the time. We launched it off Roku. It's actually available on Prime Video, doing very well on Prime Video. Why is it doing well on Prime Video? Well, it's a very inexpensive ad-free subscription with great content. So it does very well off Roku as well. We launched it in Mexico. We'll launch it in more countries. So stay tuned on that. So yes, we -- and we're putting more marketing beyond the owned and operated marketing that we currently have. We are absolutely doing that.

Got it. So I want to switch over to devices a bit. So I was just in my neighborhood, in Walmart, like I told you. And there were absolutely Roku-powered onn devices still very much on sale. But that's a question we get from investors, what the runway is there for that -- the continuation of the Roku partnership with onn? Is there an opportunity to maintain some SKU share there over time?

Well, certainly, the partnership with Walmart will continue to be ongoing, whether it's with the private label brand, which is what you're referring to and onn and it is to be seen. But definitely, we will be at Walmart. We will sell millions of units at Walmart. Through our OEM partners, we also have our own first-party TV where we are the hardware and the software provider that's doing very well. And of course, we still have players. So I would say that our overall footprint in Walmart has shrunk because of the onn transition, but our footprint in other retailers continues to grow. Our expansion with our OEM partners continues to grow and is doing very well because of a variety of reasons. One, we're investing more in our OEM partners. Two, there's a memory cost advantage with the Roku OS that's really important right now, and we're seeing a lot of benefit from that. I can touch on that. But our overall distribution strategy is still working, notwithstanding the onn transition.

Got it. And you've talked about some of the incentives that you may bring to bear, particularly in the back half of this year. And then, of course, the BOM cost advantage that you have to, obviously, just structurally, but then in addition, given you the memory cost issues that are obviously taking place across the space. So you could talk about maybe your ability across partners, broadly speaking, to gain SKU share?

Yes. So part of it is -- there's a couple of things going on. One, of course, as Walmart has transitioned, we've taken our hundreds of millions of dollars that we invest in distribution and we invested in other areas. That's just part of it. So we've invested with other retailers. We have invested it more with our OEM partners, which is helpful to just gain more of their share on the third-party side. And our own first-party TV continues to grow very well across the retail distribution channels. We've also come up with some unique SKUs. For example, we have a product called Hiro at Target, which is branded, which is a private label for Target, but it is our hardware and software, doing very well at Target. But to your point on memory, memory has gone up exponentially, and that's only now starting to hit the market. It's gone up. It went up, call it, 6 months ago, but those -- that new memory is just now hitting the market through the manufacturing cycle. The benefit of Roku is it's a purpose-built operating system for TVs. It's not the same memory footprint as a phone or a laptop. It's a much lower memory footprint. And because it's a lower memory footprint, and we're the only ones out there, at least on the TV side that has that low memory footprint because we have a much lower memory footprint, the BOM cost for our operating system is significantly less than our competitors and peers. And that matters a lot to the OEMs, to the third-party OEMs because now in this memory -- in this type of environment where memory is up 7 to 15x, when you think about a 45 or a 55 or even a 75-inch TV, but specifically on the smaller models, the percent of the BOM cost that memory -- bill of materials cost, BOM cost stands for bill of materials. The percent of the bill of materials cost that memory represents has gone up tremendously. And we can help -- we do help our partners offset that in a very material way. So not only are we investing more in terms of dollars into our OEMs, but we're also giving them an opportunity to significantly lower their BOM cost. And that is working very well. OEMs are contacting us saying, hey, we want to do more with Roku. We want to do more in the U.S. We want to do more outside the U.S. because not only is your operating system amazing and streamers love it, but guess what, it has a far lower BOM cost as well.

Maybe I'm getting over my skis a little bit on this, but I think Sony and TCL just did a tie up. I perceive maybe an opportunity to sort of march higher into the sort of more premium range of the market. Maybe that has a follow-on...

Yes, that's always on our mind. We're always working that. I think that will happen. I do think that operating systems will consolidate. I think Roku is going to be the winner in there, of course, because of our -- not just our overall penetration and our market share, but simply because we have the best operating system from a user experience standpoint. I mean the streamers just love it. They love the Roku Channel. They love the latency of it. They love the home screen, they love the remote. There's so much going for it. Clearly, still price is the biggest difference -- is the biggest input into buying a TV. I fully agree with that. But in addition, as long as we can maintain price competitiveness and we can, the operating system is just unbeatable from a streamer experience standpoint. So like there's a lot of opportunity to go upscale into other areas, higher -- call it, the higher end. We're always looking at that. But for right now, TCL and Hisense are 2 very large OEM partners. We just signed multiyear agreements with them. We have many other mid-tier OEM partners that we're signing more. I could list them, but there's so many of them. We're doing -- we're growing our share in the non-Walmart retail outlets, specifically Target, Amazon is doing very well, Best Buy and a lot of the regional areas, we're doing very well in, and we continue to gain share from that perspective in those specific areas.

Yes. And so I just wanted to talk a little bit about the -- or touch on the OpEx trajectory and just -- yes, any thoughts there on sort of margin directions, margin ceilings and how we should be thinking about...

Yes, it wouldn't be a fireside chat if we didn't talk about OpEx.

Yes, talk a little bit about numbers.

Yes. Can I talk a little bit about free cash flow, too?

Yes, that too.

That's also really important. Excellent. So to answer your question on OpEx. We said for several years now, we are going to grow our OpEx at mid-single digits. We've actually come in lower than that. We said for this year, mid-single digits is what we're going to grow our OpEx while we grow our platform business double digits. And our growth rate is doing very well. We've accelerated our growth rate coming into Q1 and into Q2 relative to prior quarters while actually maintaining that mid-single-digit OpEx growth for the full year of 2026 per our guidance. So let me just take a step back. In 2025, we did $421 million of EBITDA. We guided to $675 million of EBITDA in 2026. That's -- we grew 260 basis points of margin in 2025. We're going to go get the 330 basis points. I believe that's what the $675 million equates to 330 basis points of improvement in EBITDA margin for 2026. We've said that free cash flow is going to grow more -- is going to be more than EBITDA. We said that in 2024, our free cash flow was, I believe, $480 million, if I'm not mistaken. In 2025, free cash flow, again, will be higher than EBITDA for 2026. The guidance is $675 million. We'll do over $700 million of free cash flow in 2026. Our SBC, which is a very real expense, something I track very closely, is going down. We were down $25 million in SBC in '25. Our guide implies another $25 million to about $325 million on a run rate for SBC, which all this means our dilution is essentially -- was negative dilution for the first time in the company's history in Q1. We were -- our gross dilution was 2% to 3%. We're now negative dilution. So all this is to say that our North Star is free cash flow and free cash flow per share. We're doing very well. And that -- I think I publicly stated that we will hit $1 billion in free cash flow by the end of 2028, if not sooner. I'm optimistic on the if sooner part, on that $1 billion. So all in all, OpEx was your question, but I look at the holistic picture with our North Star of free cash flow and free cash flow per share, and I think that we can continue to grow. I think that margin will continue to improve. I think lastly, one of the important parts I mentioned is that free cash flow is higher than EBITDA. We're one of the very few companies that can say that. We're CapEx light. Our investment is in our R&D team, which is fully expensed, is not capitalized. And we have a deferred tax asset that will eventually write up probably in Q3 or Q4. It's right now on our books on a net of 0. It's going to be about $1.2 billion when we write it up. So we have a deferred tax asset that will help from a cash taxes perspective. So I expect that free cash flow for the next several years will be above adjusted EBITDA. And again, that's our North Star free cash flow and free cash flow per share.

Okay. Last one just on -- given all that, which sounds great, the capital allocation priorities for the business.

Yes. We have just -- I believe it's about $2.4 billion, $2.5 billion of cash on our balance sheet. We have no debt. We're in a great position. It's a very enviable position. I feel honored as a CFO to be in this position. And of course, figuring out the utilization of that cash is a top -- is always a top priority. Right now, we've invested in our share buyback. We've strategically done acquisitions. We're very selective in our acquisitions, and we'll continue to look at them. There's nothing imminent. We're very selective. We're very thoughtful on acquisitions. But in the meantime, we're going to continue to generate free cash flow. I suspect we'll continue on our share buyback, all else pending from a capital allocation perspective.

Okay. We've got a grand total of 3 minutes left. I'll open it up and see if there are any questions in the audience. And seeing none, I guess we'll leave it -- there is one.

There's a misperception years ago who you were actually competing with. Are you -- just on the time engagement [indiscernible]?

Yes. Great question. Actually, thank you for bringing up some of the performance advertising. It's an area we didn't cover and one I feel very important to address. So the entire industry is both competitors, peers and partners, and that's how we approach it. So there was this idea, I mentioned that the walled garden of Amazon was a pure competitor 2 years ago, maybe 2.5 years ago. Guess what, Amazon is a great partner of ours. Trade Desk was a competitor 3 years ago. They're a great partner with us. YouTube, competitor in the form of ours. DV360 is a DSP, they're partner with us. All of the content companies, Paramount, HBO, Disney, all of them, they're all partners with us, and we help them drive subscriptions. They also spend money on advertising to drive subscriptions, but we're all competing with ours between the Roku Channel and the partners. So it's a very -- like we feel very good. Our partnership with Amazon is extraordinarily strong. And so from that standpoint, I just -- I feel like we're in a very good spot. To your point on -- you brought up some of the performance-based advertisers, that's interesting. Now they used to not be a competitor, and I'm going to hold that they are now a competitor because CTV is an excellent area now to focus on performance advertising. We have a product called Ads Manager, which is growing extraordinarily well. It's getting to be a sizable business for us. We just launched it 18 months ago. That is a self-service ad product where as an SMB, you can come in, pick your budget, pick your KPI, maybe you want site visits, maybe you want conversion, maybe you want a lead gen for calls to a 1-800 number, you pick your performance metric on the self-service, you upload a Gen AI created video or any video. Most of them now are Gen AI related because SMBs didn't know how to produce a video ad. Now we can with Gen AI within seconds. And within 5 to 6 -- you put in a budget within the 5 to 6 clicks, you are up and running a performance-based ad campaign across our OS, our TRC, our partners. We can go off Roku if we need to. That is a unique and relatively new feature where performance ad budgets, I believe, are going to shift because CTV -- why wouldn't you want to try out CTV if you can get certain performance metrics. And we are seeing that within Ads Manager. We're doing very well from an ads manager perspective. And that -- now that's a whole new TAM, call it, $600 billion of SMBs. That's what they spend. There's $200 billion of true performance-based advertising out there. That TAM, some of that TAM is going to come into CTV, and that's a huge opportunity for Roku.

Great. We'll leave it there. Thank you so much.

Thanks, everyone. Appreciate you coming.