Roku, Inc. ($ROKU)

So we are going to get started. I'm Laura Martin, I'm the senior media analyst at Needham & Company. I'm here to introduce Dan Jedda, who's the Chief Operating Officer and Chief Financial Officer at Roku. Well, Dan oversees the company's global financial operations and strategic growth initiatives. Dan has been instrumental at driving steady margin free cash flow growth. He previously spent 15 years in Amazon serving as the CFO of digital video, including Amazon Studios, advertising music before a brief job as the CFO of Stitch Fix.

Okay. Great. Dan, first thing I want to ask about leadership. I'm very interested in this notion that generative AI changes everything, and therefore, leaders have to manage sort of both the fear of the incumbent, let me call it, employee base, but also sort of maybe higher differently? And how do you just think about your role as a leader changing in this uncertainty of generative AI tech disruption?

Yes. Well, first of all, thanks for having me. It's my third year. I've been 3 years at Roku, third year at the Needham Conference. So thank you...

I think, you were here within a month of joining.

Probably. Probably. Yes. But I came back, so I appreciate you having me. Thanks, everyone, for coming. So Gen AI, I think the question is how does AI really impact, how we as leaders think about the operating expense side of it? Maybe -- is that really what we're getting at. So okay. I think you're kind of getting at token cost here, which is really how we look at it. So first of all, like AI for Roku is an absolute tailwind in many aspects. We don't consider it disruptive to our business. We considered a tailwind. We've got AI embedded in our ad tech. We've got AI embedded in how we do in-stream video, we'll talk about ads manager, I'm sure later on. We believe AI is going to be instrumental in both short- and long-form content, and I think that creates an incredible opportunity for Roku. But to your specific question, yes, it is on my mind in terms of how the AI cost is building up. We are in the process of looking at this ourselves. We're dashboarding everything out. We're looking at who's using AI. We're really pushing people at Roku to use AI, not just in engineering, but across the whole company. I can tell you, from my own team, what I'm seeing in AI, I love it, they're using AI agents to write analytics. We're using AI agents to analyze things. We're using AI agents to help us on our operations side of our business. It's really exciting. But of course, there's a cost to that, and that is the token cost. So the way I think about it is, I think companies are going to be dashboarding all this out. I think companies like Roku are going to be looking at the ROI of that spend, and they're going to find a -- they're going to have to find ways to measure that. We're starting to do that now. Our spend is ramping. It's very noticeable and we're watching it, and we're saying, "Hey, what's the ROI of the spend? And there does need to be an ROI. That ROI could come in an efficiency. It could come through not adding as many headcount growth, but it also can come through shipping product faster on the engineering side, which ultimately will lead to higher revenue. Like there's many different aspects of how it would drive positive ROI. The important thing is you have to measure it and you have to hold the teams accountable for those ROIs. What you can't do is just spend on tokens and hope that it provides some level of efficiency. You have to be able to measure it. And there are ways to do this, and I think companies are starting to understand smart. I know we are. So I'm pretty excited about how that impacts it. Now what does it mean for OpEx growth? That -- what it won't mean is like net-net, this is going to increase OpEx without increasing efficiency. That won't happen. I'm 100% sure of that. It's either going to increase operational efficiency or it's going to increase revenue, depending on where the AI spend.

And what do you do for like human resources or legal or something, like how do you measure their productivity. They're not directly linked to revenue. They're not really linked to cost?

That's going to be operational efficiency. I'll give you an example. So one of the things we're doing in legal is we're looking at how AI can scan contracts, and look for key issues that would take our lawyers' time to actually read the documents. That's real. That's happening right now. So you don't need as many hours focused on this because the agents can do this for you. Now you have to be careful in areas of legal because legal is you need to be tight and you need to be right 100% of the time, not 99% of the time, you need to be right 100% of the time. So legal is an area where we're starting to experiment in, but it should take less hours to review contracts, especially for the areas that we're focused on in our contract reviews, whether it's a business contract, whether it's a license contract, et cetera. So it's an area that is right to use AI and just require less legal hours. That means we won't have to grow as much in terms of account for legal, as we grow our business.

Yes. No, that makes sense to me. Okay. So you just had your 3-year anniversary as you just said, and what do you think Wall Street most misunderstands today about Roku's business model monetization potential and strategic position?

Yes. Great question. So I referenced this from when I came on board in May -- May, 3 years ago. I just hit 3 years at Roku. It's been such a fun ride. So first of all, I think that the understanding of how much scale we have at 100 million streaming, over $100 million. We used to say approaching $100 million, now we get to say over 100 million streaming households. We hit that metric last month, super excited about that. I've been waiting to hit that. It's great. So we're over 100 million streaming households. And in the U.S., we're over 50% of broadband penetration. That kind of scale is massive in the CTV world. So no one is even close to us in terms of that scale. And while I think people understand that because we preach a lot, I don't think it's well understood what that means and what it allows us to do. So let me give you some examples, and I'm going to talk about where we were 3 years ago as an advertising company and a subscription company and where we are now. So now I'm going to go into the monetization side of the business. So the scale is there. The team continues to focus on building scale. We're going to continue to grow from 100 million streaming households. But what's really changed over the last 3 years is how we monetize and how we are focused on this massive asset that the company has built. This is an incredible operating system and how we monetize it. We have the scale in the U.S. We have scale in other countries. I'm sure we'll talk international. But in the U.S., I'll just talk about the U.S. We have the scale and how we monetize it has changed considerably, and I don't think it's fully understood outside of Roku because -- and I'll give you some examples, like, call it, 3 years ago, we were primarily a direct sales-led company, like we had a sales force that went out and they sold to the top, call it, 100 to 200 brands. And for the -- and we had an M&E business. And that was the bulk of the advertising. So the idea was to grow M&A and to grow the top 20 [indiscernible] media entertainment. It's the content partners out there and what they spend to drive their subscriptions. That's primarily an theatrical trailers. That's M&E. So that was the majority of the ad revenue. And you could -- if an advertiser wanted to come through a demand-side platform, which I'm sure you'll ask questions on that as well. You could do that, but it had to be through our demand side platform that we own. It was a product called OneView, and it was via an acquisition that we made that we turned into a demand-side platform. So it was very limited. We were limiting the shift to programmatic, we were limiting ourselves. The advertising base outside of the top brands, we were limiting ourselves. Fast forward to today, we have completely diversified the advertiser base, the ad products that we use, how we monetize the home screen, like we've diversified across the entire advertising segment, which we now break out. So on the demand side platforms, like instead of just using -- instead of mandating that you use our demand side platform, we've integrated with every DSP out there, including our most recent integration with DV360, with Google. We've integrated with Trade Desk, Amazon, World, Yahoo!, DV360 at Google, and we've integrated all on the SSP side as well with all the demand facilitators. That's very different. And we're driving more demand. That's on the DSP side. And by the way, we're driving far more advertisers now, not just the top brands. We have the whole gamut of advertising -- advertisers from the top brands down to the SMB market, which we now have an ad product focused on SMB. So -- and then on our home screen, we've really started to monetize our home screen in a very thoughtful way. So all that wasn't happening 3 years ago. It's all really happening now, and you're starting to see the benefits of it, we grew our ad business -- 26%, is 27% in Q1. We segment that out now. Similar story on the subscription side. So we have subscriptions 3 years ago. It wasn't a focus. We had -- we continue to have a pay product called Roku Pay, but we weren't driving subscriptions. We just had it, if you signed up through our -- on our platform through a subscription partner, we got a rev share on that. But what we didn't have is a focus on it. We didn't have a leader. We didn't have product focused on it. We didn't have a home screen that really pushed subscriptions. And specifically, premium subscriptions, which is our version of similar...

IBox or AR.

Right. But premium subscriptions is similar to Amazon channels, where the content is embedded throughout the user interface, not just in the app. So we have a premium subscription product where you sign up, and now that content is embedded throughout the user experience. So we've diversified within the subscriptions business. And these -- all this diversification, all this focus have driven a lot of incremental demand our way. Supply never an issue, plenty of ad supply. We always had plenty of ad supply. Our sell-through rates were very low because of the Roku Channel being the #2 app on our platform. It was -- we used to say it was a top 5 app then we'd say it's a top 3 app. Now we've said over the last year, it's the #2 app on our platform. Everybody knows who #1 and #3 is, but we are #2. So you have the Roku Channel having a tremendous amount of supply of ad impressions. Now we're really focused on driving the demand our way. That's why advertising business is growing and then our subscriptions, a big focus on driving subscription, specifically premium subscriptions, and you're seeing that growth. Our total subscription business grew 30% in Q1. That's what the friendly comp -- friendly comp from Q1 of 2025. But even backing out friendly, it still grew 23% year-on-year.

One of the sentences you said was we're starting to monetize our home page in a thoughtful way and just for a way of background to level set everybody up to our level is they used to only do essentially theatrical advertising, but you could also -- if there was a new show on Netflix or something they might buy an ad unit on that home page. So when you made the comment that you're starting to monetize your homepage more thoughtfully, I still don't have video on my home page. I still don't have an ad unit for you guys that's like a house ad unit on the home page. And I still don't have scrolling, which Verizon had even before they got bought by Walmart. So defend the sentence, we're starting to -- things monetize our homepage more thoughtfully.

You have video on your hope page of that I'm 100% certain. You have video, auto playing video in the right marquee ad unit on your homepage. That is a significant number of impressions. It tells me you're not spending time on your home page, you're just jumping to your content. I'm 100% sure you have video on your home page. Secondly, the content row at the top, which is a personalized row, that's new. That's driving incremental engagement into hours that we monetize. It's driving incremental subscription sign-ups. And by -- not by a small amount, like it's very intentional, how we personalize that content row at the top. And then lastly, I'll say this, like we have a new home screen coming out that we've talked about. We've rolled that out to a material percent of our group. It's still not fully rolled out. It's not even close to being fully rolled out. It's, call it, low double-digit percentages have rolled out as we continue to test and reiterate the new home screen, I've been using it for 2 months, it's amazing. It's even more personalized and the ad unit is even more prevalent because you start right on the home screen rather than starting on the left map. So you've got even more impressions now being generated for that ad unit that is -- that can be video. It's not always sold video, but there's a video ad unit, an autoplay, video adding unit. All that is -- and now you mentioned ad units -- other ad units on the home screen. Stay tuned for that. As we roll the new home screen out, I'm very excited. I think there'll be more ad units on the home screen. I think there can be ad units embedded in the content titles of the home screen. Biddable or not, we'll figure that out. We've got an amazing product team that's focused on us. I'm super excited about it. So there is even more opportunity in terms of engagement that we monetize in terms of subscriptions that we monetize coming to the home screen.

Right. Because I mean what I would say is that most of streaming is targeted. And the nice thing about the Roku home page with your density of like installed base, it could be a reach product. If every single person sees that at, it is a -- in my opinion, a competitor or linear in terms of its reach.

It could be a reach product, and we have that because 125 million people start the home screen, and they like -- I mean that's like above Super Bowl level reach every single day. So we have that, but we also could have performance-based products. Because again, the whole top of the home screen will be personal -- is personalized. It's personalized now, but the newer home screen is even more personalized. And what you can do is you can put, again, an ad unit in there that's highly endemic to the personalization. So again, think of it. It can be both a broad-based reach unit, but it also can be some type of a targeted performance-based unit as well.

Okay. So when I go to Roku, it says, "Dad, Mom, then my 3 kids names. Is the home screen that page? Or is a home screen once I hit mom or once I hit my 23-year-old daughter?

The home screen, the home, again, we have logged in user information as a household. We do not have profiles up. So your home screen is where you start. Now again, to be fair...

And how do you personalize it [indiscernible] as 4 people?

We can personalize at that level. It's not personalized per se at who's watching at what point in time. But we have a pretty good idea. We know who's watching what when they start and we can personalize it accordingly.

Okay. Because I was going to say it must be a household personal position, like zip code or by viewing. So the fact I watch BritBox, you think I'm a brit. So you're giving me a personalized teeth care products or something. All right. Fair enough. How should Roku -- how should investors think about Roku's long-term margin structure if the business continues shifting towards software, advertising, subscription and home screen monetization?

Yes. So again, we broke out our 2 segments, subscription and advertising. And one of the reasons we wanted to break that out is we wanted to give investors more insight into how the platform revenue built up between subscriptions in advertising. And we wanted to give the margin structure as well. So we segmented it out starting in Q1. I'm very excited about that. We've been working on that for several quarters. And let me just talk about margins of each business, and then I'll tell you how they are going to mix out from the platform side. So on the subscription side, we did just over 40% margins for Q1. That's down on a year-over-year basis from about 44%, 45% prior Q1. And what we're seeing there is the premium subscriptions just driving a lot of the subscription business. Now we're still growing nonpremium subscription business, but premium subscriptions, which has a slightly lower gross margin is the biggest driver of that. I do think -- and I said it, I do think that we will stay at or above this 40% for the rest of this year. We have other activities within subscriptions that are higher margin that I believe will start to grow, be a higher percent starting in Q2. And I think this 40%, maybe slightly higher, will hold for the rest of this year, still a very profitable business for us. On the ad side, there was this thought, this notion, and as much as I tried to dispel it, there was this thought that, hey, since you're integrating with DSPs or since CPMs are coming down in the industry, that's hurting your advertising gross margin. Well, I constantly said that's not accurate. There's a mix impact within platform. Advertising is doing well. And now that we've broken out, you see that. So advertising margins were just over 60% in Q1. That moves up 450 basis points from Q1 of last year. And I said that we believe that we are going to maintain 60% or just north of 60% for the rest of this year as we continue to focus not just on growing our advertising revenue, which we are doing, but also on optimizing our gross profit in advertising. So we're seeing good -- and there's a lot of reasons why gross margins will continue to stay at 60%. We've got the home screen, very high gross margin. We're very -- getting very good at optimizing our campaign performance. So we optimize for, of course, campaign, completion campaign metrics, but we also optimize for gross margin within our in-stream video, and we have ways of doing this, not all impressions are created equal from a gross profit perspective, so we can optimize to fill our highest gross margin impressions first, and our lowest gross margin impression last. Of course, this just makes sense. And so we're getting really good at this optimization. So I'm pretty happy with where we are in advertising gross margins. I think that there's a chance for them to grow. We'll wait and see how we do. But for the rest of this year, I think that 60 -- just north of 60% is the right target. Now what this all mixes out to is this 51% to 52% platform gross margins, probably closer to 52%, which is what I stated during the call.

Right. Okay. So the 2 things that Wall Street Things have pricing power in the Gen AI world or unique content and unique data, Roku has both. Can you talk about why those become? As we get more synthetic content being created by generative, why do those assets become structurally more valuable?

Yes. That's a great question. And we believe that Gen AI content, specifically long form and then also within video, call it, 50 30-second advertising videos. Like Gen AI content is -- has a chance to be very disruptive in our space. And we're very excited about that. So we think that we will be -- actually confident that we will be a beneficiary of AI content in the form of long form, we monetize long-form content very well. Again, if short form comes to CTV, we'll monetize that as well. We actually have some short-form content. We have clips on Saturday Night Live. We have shoulder content on sports like we have versions of short-form content. It's not massive, but it's there. But even if short-form content does come on our -- on to CTV, we're very good at monetizing hours. We'll monetize that as well. Long form has the opportunity to be very disruptive in areas like for Howdy, which is a low-priced ad-free content offering. So imagine if Gen AI in long form impacts Howdy, like we can continue to put a lot of AI -- really good AI long-form content and still charge a relatively low price. And there is a market for that. Howdy is doing very well. We just launched it in Mexico. We just took it off Roku. It's in Amazon channels. It's doing well in all 3, on Roku, off Roku and in Mexico where we just launched. So there's that opportunity. And then I think the real disruptive can and will be AI for even shorter form content, which is like 30-second and 15-second videos. That's where our ads manager product comes in, where Gen AI has created the opportunity for small and medium-sized businesses to participate in CTV. That was the reason, one of the 2, in my opinion, primary reasons why SMBs did not advertise on CTV was one they were too small for a DSP or an agency. And two, they would have to -- they'd have to create video, which can cost $20,000, $30,000, $40,000, $60,000 for an in-stream video. And the ads manager, which is a self-service SMB-focused product has taken both those impediments away. Now it's really easy to generate -- to sign up to have a campaign that's based on performance. And then more as importantly, you can click on the -- you can create videos from a Gen AI perspective that will help you do to immediately upload your 15- or 30-second video for in-stream video. So now there's no reason why performance, advertisers and SMBs can't come to CTV and they are coming to CTV.

And so one of the things that this ads manager sort of a new product for you, but like Mountain has been doing this performance CTV targeting SMBs for 3 years. Are you benefiting from the fact you're sort of a little late here because it's already been an established market, somebody else had to educate people and create?

Yes, we're not late. And let me tell you why we're not late. So I believe that we've got a big advantage relative to others. Because everyone -- first of all, as a publisher, as a platform level publisher, we're quite large. So we do not have to go and negotiate deeply discounted impressions to go back and sell because we are a publisher, we have impression. So -- and by the way, we have massive reach across the platform. We don't have to stay on the platform. We could go off the platform, we could be, but we don't need to because -- and we can go off Tier C because we can do what's called our run of network. We can go off the Roku apps and yes, and just get inventory that way. So we have massive reach that no one else has. We are in a great position to do this. What we needed to do and what we have done is built a very easy-to-use self-service sign-up and then make sure that you can upload a Gen AI video in an easy way. And then we just have to tell people about it through marketing, and that's what we're doing. So we are at -- and by the way, our first-party data allows us to do amazing targeting, get amazing performance. We're integrated with measurement companies will API sites to do site visits or whatever KPI the advertiser wants to get at. We're integrated with Shopify, so we can get conversion data, which in that view, you can get true ROA or return on ad sales. So there's a lot of opportunity for this SMB market, which is $600 billion according to sources are like spent by the SMB market in total. Of that, a significant amount is spent on search and social. And I think those dollars are going to migrate. So you have the tailwind of linear TV moving to CTV and you have a tailwind of search and social or performance-based moving to CTV. So this TAM that we play in is growing very well. So not only is our diversification and ad products growing, but the TAM we play in is growing well. It's a great position to be in.

And one of the things that I would say other companies are talking a lot about is you have to be omnichannel. You can't just be in 1 sector. Do you disagree with that because you guys are only in CTV?

Yes. I think that -- first of all, like we have ways of -- we're working on this as well, like, for example, Howdy has a mobile app, CRC does have a mobile app. And there is -- I do agree that omnichannel is a further opportunity, and we'll see where we go with this. But right now, we are seeing the performance and the KPIs play out in the CTV channel. And whether we get that data, whether we buy that data, there are ways to incorporate that perspective into our own CTV. Again, like let's see where this goes. But I don't believe it is a must have. And by the way, all indications are from all the KPIs I look at, and I look at a lot of them, all indications are, we don't need them because the ads business -- as ads manager business is doing extraordinarily well.

Okay. Yes. Ads manager opened up a whole new TAM for you.

100% whole new TAM and as importantly, an entire set of millions of advertisers that want to try something different, like who wouldn't want to see a video of their SMB business on TV. They all want to try it, but they also want to make sure it's working. So they want the performance capabilities. Now with products like ads manager, those capabilities exist.

Yes. Okay. So after spending 15 years at Amazon, does Roku have a structural advantage over Amazon and other large walled gardens or does their scale and sort of breadth of product to help them?

Okay. That's a great question. And again, the difference between my first 18 months at Roku and my second 18 months of Roku, like night and day because for my first 18 months, I had started Roku and I think literally 6 months after I started Roku, Amazon announced their serving ads in Prime, which is a major announcement, very bold -- very bold move, major...

Overnight.

Literally overnight, and opted everybody in and said you're going to have.

Was opted, command performance...

You're going to -- you're going to have to work to opt out.

Yes, exactly and pay.

So -- and during that time, I heard a lot and I saw a lot written on, hey, it's over the walled gardens have won the CTV ad business. And I could not have disagreed more because I had this -- we had this view on where this would go. One, our first-party data is unmatched.

That's true.

100 million streaming households. You start the experience with Roku on our home screen. We know who you are because you are -- we have 100 million-plus logged in users, granted streaming house. So our first-party data is unmatched. Secondly, and again, I saw this, but I was kind of like preaching this. Think about it this way, I mentioned that the Roku Channel is the #2 app on our platform. And I know -- and I basically said that we are over half of broadband households and no one's even close to us from that. People have -- everyone knows who #1 is, everyone believes they know who #3 is. We're #2. -- who's not #1, #2, #3, #4, like these walled gardens. And so even on our platform, like we have massive reach and it would be better for these companies to partner with us rather than go it alone, despite what everybody thought these are smart companies. And what they're doing is they're partnering with us because they know our first-party data plus their first-party data, our scale plus their scale, we're far better off together. So what does that mean? That means that as Amazon launches a DSP, we're quite certain they were going to do this. They wanted to partner with us. We wanted to partner with them. We signed an agreement on a platform level wide to integrate into Amazon. DV360, starting DSP, yes, Google is not -- they're basically -- YouTube will be part of that DSP, but it's not just going to be YouTube. They just partnered with us. We just adopted as one of the first -- I think we were the first to adopt their unique identifier, which is a -- I think it's a version of a hashed e-mail, which is similar to other identifiers. And so now we're up that. So our media will be on top of the overall demand within DV360 because they see the value of integrating with Roku, given [indiscernible] our 1P data. But my point on all this is the walled garden approach is, yes, it might be there in some aspects, but the partnership approach, I think that we've solved that working together is far better than going alone.

Isn't it that just they're trying to become omnichannel. And they started from just a different place. They're coming into your world to become omnichannel?

They believe that more reach is better.

I would agree.

1P data, when you match 1P data across massive reach that Roku has is better. Remember, like it's not just the Roku Channel, it's a platform-wide agreement. So the matching that we give, which is all done in an anonymized clean room and safe environments, the 1P data match that we do really does give like Amazon far more reach than they otherwise would get even if they integrated with us just as a publisher. And then where we are integrated as a publisher because TRC, plus remember, we have reach beyond TRC. We get share of inventory from other partners. So we have a -- we have reach across the whole platform that is unmatched, so any DSP, whether it's -- it's not really omnichannel. It's just like we -- the reach and the data together were just going to be better. And both companies win in that environment.

Okay. Okay. But it doesn't give them competitive advantage over you because they also have these off CTV assets they're marrying with your CTV asset?

If they want their performance to work well, it does not. They're better off integrating with us and being -- letting the advertisers go across and get scale...

I get that, that's beneficial for them. My question is, are they putting you out of business with your own data because they are now omnichannel.

Our data is very safe and very secured, and we have ways to protect that and so on. So no, it's not a worry.

Okay. But they don't have a better ad mouse trap than you.

Not a worry.

Okay. Okay. Fair enough. So you now have DSP integrations with the Trade Desk, Yahoo, Freewheel, DV360 and Amazon. Can you talk about the impact that those DSP agreements have had on fill rate, CPMs and total Roku margin and revenue growth?

Right. So again, go back 3 years ago, 0 DSP integration. Fast forward now, everyone is integrated. We're trying to go as deeply as we can with all the demand side platforms as deep as they want to go will go as long as we're protected and we are doing that. So what does that mean? It means we've opened up a lot of incremental demand to flow to us as a publisher to us as a platform, to Roku as a platform, and we are now getting a diversified area of advertisers, diversified across the spectrum of CPMs. So where -- think about it this way, like different DSPs and different impressions will have different CPMs. I think this is really important to know. And it's probably not well understood. Like everyone thinks CPMs are coming down, that's bad. Not necessarily, like if we're adding -- if we -- like not every impression is created equal, if we have an impression in news, that's very different than an impression in some of our highly rated content like our original content or our direct license content. But my point is we don't pay these take rates with the DSPs. We may have payments with the supply side, the SSPs, but that's all incremental demand facilitation as well. That's where we're just integrated with an SSP as a publisher. And we're saying, hey, as long as you can hit these metrics, just keep [indiscernible]

Yes, filling.

You're filling the inventory. It's all incremental. My point on all this is, again, like that is very different where we are now. So it's been a huge positive to integrate with the DSPs, to set the right floor pricing to bring incremental demand because as I said, because we're the #2 app on Roku, the TRC is because we have run-of-network inventory or inventory across the whole platform, supply is not a problem. We have plenty of supply, and we can create more supply relatively inexpensively.

Just having your ad load go up?

Well, that would be the last thing we want to do. We can steer more people into TRC. We can -- there's lots of things that we can do to increase our supply. But yes, we could go. We have a very low 7-minute ad load right now. It's very low. I don't want to do that because it's just -- that is a slightly negative streamer experience, but we could, we could do that. We don't need to do that. Our fill rates are not close to being sold out. We're still generating more impressions on the supply side. But CPM is coming down, not a bad thing for us. I welcome it.

Okay. Great. No, that's super helpful. Okay. What is Roku's content aggregation strategy is Netflix and other OTT competitors and short-form video vertical videos, clips and AI generated. I know I asked this on the call, and you just mentioned it now. You guys do have some short form. Typically, it's like a highlight reel of something that's around like to promote a longer-form content. So I think the form of the question is more like stand-alone like vertical videos where it's specifically a series or short-form video work specifically like a series not an adjacency to something that's sitting long form?

Yes. So we have -- again, we have short form. That does well because we're very smart, but we just don't have a lot of short-form video. As I said this...

Because ad loads are tricky.

Well, no, it's just -- CTV is just -- hasn't moved to short form. It just isn't there yet. Now if it goes there, if the big beyond YouTube if others come that's great. They're going to want to be part of Roku, and we'll monetize that because again, we have massive reach. We have massive scale. We have -- the OS is in over half of broadband households. If more short form, if CTV were to shift, to short term. We will be a beneficiary of that because that's going to lead to more hours and we know how to monetize ours. So I think more importantly is this concept of where Gen AI can improve video and ad video specifically, within ads manager, our sign-up flow, I was just looking at this the other day, our sign-up flow is so impressive on how many people start, they sign up. They say, "I want to do this. They go to the next, they choose their campaign, their KPIs on what they want to target, how they want to target. They choose where they want to advertise across the Roku Channel and other publishers. Then you get down to the Gen AI form video. By the way, it's great we create -- we can -- we have tools that do us. We have outsourcing tools that help you create a video on Gen AI, but it's still the biggest falloff in the pipeline. And...

90% go there now.

There is so much opportunity like people take all the time to fill it out and they get down to that piece and it's still the biggest falloff that we see. And we're still doing amazing. There's so much opportunity for Gen AI and agentic AI to like create the video from scratch via prompt.

Like instead of asking them to say, here's 3 vides, so you want to choose one of these.

Yes, and then improve from there. I actually want to make this person over here. I want to add some more people into this video, I want to have some more creative...

Or I want it to be a forest, not a desert.

100%. And I think that's going to come. And I think that's going to help in this product even more. Also, if you fast forward even a little bit more like having the AI run it where creative is being created on the fly learning...

Like a dynamic.

Yes, so it's running 1,000 creatives over 1 million impressions and picking the right creative [indiscernible] optimized for the -- whatever KPI they're optimizing for -- and by the way, this isn't specific to ads manager, but that is going to happen. It's just a question of when. So we are -- Gen AI is there. It's working. It just has some ways to go, and it's going to get better.

So when I think about proprietary content, having pricing power, is there some reason we aren't hiring 3 kids from USC film school just use Gen AI tools to make proprietary...

That's happening.

But on Roku for Roku. Like a 10-minute episodes...

So again, if it becomes like that's happening, that's going on. It's just early times, and we will play in this space. We will play in this space and we'll see where it goes. I can't say when it's going to happen. I just know it's going to happen, and we will be a beneficiary. We're well aware of this. We're very focused on it. We're...

And you can follow. I mean it's not urgent to be first. You can just follow.

But we'll lead. We'll be upfront on this. Like this is a big deal for us, like we will be out front this.

Okay. Okay. Let's go to devices. How should investors think about Roku's hardware segment based on device revenue, down 16% with negative 14% gross margins in the first quarter?

Right. So this is always a challenge. I just want to take a step back and explain how our device business works and how unit sales and device works. So the way our accounting works is we recognize revenue when a first-party player or -- which is all our players and a first-party TV where we are the manufacturer of the hardware are sold, we recognize revenue and we recognize gross profit. When we are the OS on our OEM partners like TCL and Hisense, and Philips, et cetera. When we are the software on someone else's hardware, that does not show up as device revenue, even though it's a unit sale for us. Where that shows up in terms of distribution cost is on sales and marketing. So what you're seeing now in device revenue is you're seeing a lot more -- more volume flowing through our OEM partners. We just signed 2 multiyear agreements with our top OEM partners. We -- our OEM partners are partnering with this very well because we have a BOM cost, a bill of materials cost advantage given our low memory footprint. So in this rising...

These are all really important and explain that...

Yes, this rising memory environment is creating a lot of issues for a lot of different industries and the CTV is one of them because it costs -- there's memory involved. Our operating system was built by design to have a lower memory footprint than everyone else's operating systems. And that was with something Anthony felt very strongly on that he built from the start. And it was really smart to do because it's always allowed our BOM cost, our bill of materials cost to be lower than everybody else's out there. So partners would want to work with us. Now you fast forward in a high memory -- when memory cost is going up, that BOM cost advantage just grows.

Let me give you some numbers. We think every TV they make is under 2 gigs of memory cost. We think Amazon and Google are 4 to 6 memory.

So there is a significant advantage to having the Roku OS on your hardware. So what that has meant is more OEMs are coming to us saying, "Hey, we want the Roku, not only is it a great operating system not only is the Roku channel have, not only is it simple to use, not only is the home screen and amazing, but really importantly, it's significantly cheaper and basically reduces the entire cost of the TV for the OEM. So we are seeing more volume than we originally expected at the start of the year going through our OEMs. So -- and also memory -- the increasing memory cost does impact our own player margin. So you're seeing that go through as well. My point on this is between Q1 and Q2, nothing has changed from our total unit forecast. Nothing has changed on our device investment, what you have is some mix accounting between our 1P and our 3P units.

Okay. Questions from the audience for Roku, Yes, sir?

You mentioned something about taking share from [indiscernible] so can you just expand on what's driving that?

Yes. I can. So [indiscernible] Social has primarily focused, a lot of it on SMB and those SMBs have certain KPIs. They're not just going to run -- they're not running advertising for reach. They may, but they don't. They want to know that they can review site visits or clicks or conversion data. That has always been an impediment of TV because when you're on linear TV, you had no way of actually tracking those KPIs. Now with CTV, you can actually do that. I'll give you an example. Like if there is a small business who has 5 restaurants or auto dealerships in the Austin area, and they want to run a very specific geo targeted ad and track site visits. How many people visited the auto site, you cannot do that on linear TV. Now on CTV, you can do that. You come in through ads manager. You do like 5 clicks. You geo-target it, you say, go, you upload a video, which we can help create, you hit go publish and you are off targeting that geo-targeted space via APIs. We can track site visits. We're integrated with measurement companies. So we can track that KPI that, that advertiser wanted to see, and they see does it perform? What's really important is it has to perform. And that's on us as a publisher to make sure it performs and we're really good at this. So it will perform. If it doesn't perform right away, we'll get it to perform. My point on all this is that impediment, it was that tracking those KPIs and the ability to create a video, which would have been, again, $10,000, $30,000, $40,000, they don't know how to do it. They have to hire someone to do it. Now it's done through Gen AI. You've got the 2 biggest roadblocks, performance-related KPIs and Gen AI created video now gone. It's bringing more of these SMB advertisers over into CTV.

Okay.

Can you expand on that performance related to broad performance and how you're planning [indiscernible]?

Yes. Yes, it's a great question. So we'll never -- it's not there like last click attribution is there. Like I'm not suggesting that we now have last click attribution and you can track everything specific. I think we'll get there at some point. I have thoughts on this. I don't want to get into that because we're not even -- I don't think anyone is close to like last click attribution. But we are there on things like, again, tracking site visits. We're integrated with many companies that help track measurements and like Shopify, we're integrated with Shopify. So if you have an SMB with Shopify, we can through to integration to the shop by track conversion data that we can get [indiscernible] as long as the advertiser allows us to do that. We're integrated with measurement companies like incremental, which will do causal based lift analysis for the advertisers. So you can literally get via data feed, you can understand what the impact of the ads were. And there's other KPIs. So again, I mentioned site visits. There's other KPIs. By the way, here's a great 1 that's actually done really well. Insurance, putting an insurance number on a CTV video and being able to track how many people saw that video that also called the number. It's lean gen in that particular case. But it's working. Insurance companies love that because they can track the cost per every lead. This is not possible in linear TV, not even in CTV many years ago. It's possible now. It's a great product, working very well for us.

Okay. I'm going to call it there. Thank you.

Thanks, everyone.