I wanted to start off with some of the recent news that we've been getting a lot of questions about: the new Dentsu partnership in Sweden. Actually, hang on. That one is fascinating, but I did want to ask about Walmart and VIZIO. Maybe we can start there. Take us through the details there. What made Magnite stand out as the partner of choice to enable that data more broadly, and what do you think that partnership could mean for the business?
Yeah, no, great question. We're very excited about that news that broke last week. I think for a bit of time we've been pretty vocal that data, especially in CTV, is starting to move to the sell side. Right? If you park your data within one partner on the DSP side, you have a fairly narrow aperture, and it forces people to work through one party. I think to the extent that we've seen that happen with a lot of our commerce media partnerships that we've announced over time, I think was good traction and good evidence of that.
I think the fact that we've been a longstanding partner with VIZIO prior to Walmart acquiring VIZIO, both on the ad-serving side of CTV, as well as a supply partner of theirs in CTV, this is a natural extension for them to take their first-party data, move that from just being captive within one DSP, and tap us to be the gatekeepers by which to protect that data, yet utilize that first and cross VIZIO screen inventory, and then expand that to Yahoo as a next DSP.
By no means do they plan on limiting that to just their CTV inventory or just one incremental DSP. That's just the beginning. We're pretty excited. We think that's incremental. That's something that will start to have some traction to it this year. Again, I think the bigger value is that it truly is something, a very visible move by a very large commerce media player in the market that's chosen Magnite by which to park their data.
Got it. Extending over time to additional DSPs and then potentially also to additional inventory beyond the VIZIO.
Exactly correct. Yeah if you will.
What do you think is enabling, and what maybe are you enabling in terms of data that's giving people the feeling, "Okay, I can begin to extend my data, make it purchasable for a variety of DSPs, variety of partners, and make sure it's still secure, there's not going to be data leakage, and I'm going to get full value for that"?
Yeah. We've made a huge investment in our curation, what we call curation. Yeah. That whole data management. We actually acquired a couple of companies a handful of years ago to beef up our protection so that that data can't leak out to where it shouldn't belong. We have to be trusted by, not only do we manage data from Walmart and commerce media and retail, but also first-party data from all of our CTV clients, that's super critical. Yeah, we've just established a track record and product capabilities, and that has just grown significantly and is a huge part of our story going forward.
Yep. Getting back to Dentsu, I think it's emblematic of when it seems like almost every time when I open my screen here, I'm seeing news about new Magnite partnerships, right, new things that you're lighting up. Is that a function of what you just mentioned, the capabilities that you've built up over time for secure data enablement? Or is it a function of additional focus that you applied there? What is making Magnite partner of choice, if you will?
I'll start, and then you can hit it. We're really hitting a network effect, a little escape velocity. We've got exclusive access to almost every CTV provider out there, with the exception of one. We've got these relationships on the buy side, and we have exclusive relationships with running marketplaces for certain folks on the buy side. If you want access to certain demand, you need to work with us. If you want access to a certain supply, you need to work through us. It's just created this virtuous cycle, and it's super exciting.
I think you're really starting to see a change and a shift in what agencies are doing to truly differentiate themselves and create value both today and in the future, right? You see that in their data plays, obviously, with what Publicis did. You also see it from their agency marketplaces, which is essentially their tech offering, which in a fairly concentrated amount of publishers or media content owners, they can still play a very, very significant role in how CTV ad spend is controlled, deployed, audience and data against it, as well as the preferred path for how to transact.
They want the first path for their partners, customers, brand buying to come to is to their page. They want them to come to their marketplace. They don't want to relinquish that to go through a demand-side platform. They would rather keep that captive themselves, give them decisioning, allow them to pick and choose, select or deselect, what platforms they want to advertise, which ones they don't. Bring in their own data without having to pay a markup fee for transacting against it. We are the ones that are lighting up those marketplaces for them. There's several that are announced.
Dentsu, Sweden was one that was announced recently. Spain was announced last year. They tend to be a little more decentralized, so you have region-by-region buy-in versus a corporate mandate. Some operate in more of a centralized function. Some have both a decentralized and a centralized control point. We see agency marketplaces leaning into connected TV buying as well as DV+ non-CTV buying, being a trend that doesn't back off in the future. It's something that truly is an area where they think they can impart value in the ad tech ecosystem and value chain.
Got it. You've had the agency marketplaces up and running for a few years now, I think. Maybe you could talk a little bit about the shift to agentic transactions, if you will, and is that supercharging that trend in programmatic, or is that really just bringing something new to bear in terms of the market, in terms of the automation of the more traditional IO-type business?
On the agentic side, we came out with some news on kind of three product fronts about a month ago. I think everybody's trying to scramble and figure out who has what on the agentic world. It makes sense that we would have a seller agent that helps you be able to create and replace what audience graphs used to be historically to be able to, across publishers, figure out what audiences you want to reach across their different taxonomies.
You also have mediation agents that help a specific publisher or media content owner better manage their inventory, yield manage, optimize price floor latency, who's included in auction, what source of demand to include, how to mix that out together. Those are things that you'd expect a traditional supply-side partner to be able to have, especially one that's got an ad server. What's new and novel that we really announced as part of that announcement is our buyer agent, right? From a buyer agent perspective, what we announced, and we've started doing some kind of stealth demos of what we offer there.
On the buyer agent side, this is specifically targeted towards agencies to start with, and it's targeting insertion order business. On the front end today, what still happens to a very large degree for a manually placed insertion order is a brand comes to an agency, and they come with a proposal for what they would like to accomplish. You have either an intern or first- or second-year employee at an agency that then takes that and develops a media plan. A media plan is effectively what inventory they should buy, where it should run, how it should perform to accomplish the goals of what the brand came to them with against executing. Fairly simple and easy to understand. If you take a document that's a proposal and load it into an LLM, an agency, or we can create and turn that into a media plan.
What gets complicated is if you don't have that connected to supply, real-time supply, you might not know where pricing is. You might not know how it performs. The step that happens afterwards is you take that media plan, you test, you analyze, you refine, and you rerun tests. You do that a couple of different times, and that may take as many as two weeks each time you retest. If you think there's three cycles of test, analyze, and refine before you actually get to what you think a campaign should be, refine your audience, refine your placements, version your ad for that, we do all of that, the front end from proposal to versioning to test, all within about 10 minutes, including building your own creative and versioning your creative.
We think it's something that's very powerful. We are not interested in our buyer agent being the end-all be-all of buyer agents and we need to compete with others. We welcome people to use ours. We would use anybody else's to connect to our seller agent. What our goal here is, and what we think agentic does on the buying side, is dramatically reduces friction, time, and energy spent to be able to go from a proposal to actually deploying ad spend. That's what we're encouraged by, and we do very well in the economics that we have in either marketplaces or in ad spend across take rates we charge without this being something more than just an accelerant for how more and more ad spend transacts through our systems.
Got it. Maybe you could help us sort of think about the scale of that opportunity. Via programmatic guaranteed, we've thought about a lot of the business which is still sort of offline negotiated, now sitting already on the programmatic rails. How big is the remaining IO opportunity? I know the various estimates out there, what do you see out there? Do you think that's a pretty big slug of business potential?
It's huge. Yeah. One of our greatest areas of growth is really that transition from direct sold into programmatic. The efficiency gains of agentic are so interesting. Somebody just talked about when they're working with an agency and a publisher working together for a manual IO, they said on average they had 38 different touch points in that manual process of putting that IO together. The ability for an agent to collapse the mundane components of that process and allow much more focus on strategic is going to help transition much more of that shift into programmatic. It's a huge tailwind for us, we think.
The lines are blurring a lot, right? It used to be that if we rewind four years ago, five years ago, the upfronts and any PG, programmatic guaranteed deal that got done, that was the enemy of programmatic, right? The lines have blurred such that if you make a commitment at the NewFronts or the upfronts, you have the ability in any spend you do programmatically in the scatter, in a private marketplace. That counts against your upfront commitments. Now the upfronts and the NewFronts are a very good indicator of the strength and the health of the market. We've even indicated that this year, especially led by things like live TV and live sport, it's been a very strong start to the season for people indicating what spend they're going to deploy over the next year.
Those things now are amorphous and really those lines are blurred. It's really hard to say a commit is all PG, everything else then is in a scatter, it's a PMP or an open market deal. Those lines have really blurred in the last several years.
Got it. May I play devil's advocate for a moment here? When you think about maybe the starting point motivation for AdCP and where it came from, right? People were talking about GHG reduction and reduction in energy use and why don't we blow up the entire programmatic supply chain and create sort of an automation of, again, what that intern or that young media planner does, which used to be, "Okay, I've got a ranker. I'll pick things off the ranker until I've got so many total rating points." When you think about that, what would be your sort of counter to that? Why do we still need the programmatic supply chain in all its glory as a critical part of this sort of new agentic workflow?
Yeah, you still need a trusted system of record. If you think about it, you've got on the buy side and the sell side. You need someone to validate that there isn't fraudulent traffic that's there. To our earlier discussion, that data is being protected and not leaked out. You've got to have someone collect the money, and you got a many to one challenge in all of this. I think about maybe a Nasdaq or a stock exchange or something, you still have a centralized system of record that is going to be critical to this process. We see agents tapping in on the buy side, tapping in on the sell side, but running through our ecosystem, and we think we're really well-positioned for that future.
Got it.
How about data protection side? If you think about all the privacy laws that have gone into place globally over the past decade, if a publisher, all of a sudden, millions of websites, millions of apps, every CTV publisher allowed unfettered access from a buyer agent that they don't know and trust, having access to personal identifiable information. The publisher are the ones that are responsible for those fines and for those violations. For that reason, they want to make sure they reach demand, they want to make sure they reach it at scale, and preserve all the chain of command and the authorizations and the consents that are required to uphold all privacy regulation.
Not to mention they want to get paid for that inventory. If they work with thousands of buyer agents, they have no idea whether they're going to get paid or not. Right? Those are very important factors for why anytime somebody's running a test using agentic transactions, buyer and seller, we're the first ones they call because they have to transact somewhere.
Got it. All right, last devil's advocate question. When you see somebody like Netflix developing a CAPI, right, do you think that's a first move toward potentially the proliferation of walled gardens, if you will, sort of mini walled gardens where agents can transact across multiples of those? Just to sort of play this out, is that a potential vision of the future that you're thinking through and thinking about Magnite's role in?
What do you refer to by CAPI?
Their Conversions API, Netflix. This idea of potentially becoming a sort of mini walled garden, if you will. Not mini, but maxi walled garden, maybe.
Yeah. I think that from the perspective, this is kind of thematic for a lot of the CTV players moving more down the path of more performance, right? They understand that they need to move down funnel, they need to prove more results. They really want DR more so than just branding sitting at the top of the funnel. I think all of them are moving in that way. There are elements of that that they highly lean on and require of their demand, their ad serving partners, their yield management partners, their data encryption and audience creation partners, as well as connecting to the world's demand. We don't think that's an element that displaces us or is an element that hurts us. If anything, that's further along a very technical, difficult, programmatic journey where we think we can add value.
Anytime things get more difficult, more complex, more expanded, more programmatic, those are music to our ears because those are our skill sets, and that's really the strength that we play from. We like living on the edge of constantly being tested. There are things that we started doing in this space 10 years ago that certain partners of ours today use. If you look at the life cycle that we've been in, ever since we entered into CTV in 2019, we have not lost a single partner. In fact, we've expanded every single relationship that we have with any partner out there. That's a testament to we're constantly innovating, we're constantly developing, whether it's on the live sports side, whether it's marketplaces that they can white label and take to market on their own, whether it's onboarding SMBs and reducing friction there, giving them more features that they had to go to three or four vendors before on their own, or ad serving capabilities.
All of those things are why we think we continue to win, and it's not static.
Got it. Backing away from these alternate realities and getting back to the actual reality of the business, which is going pretty well. I want to unpack a little bit the recent acceleration in CTV contribution ad tech. It seems like we've seen an inflection point in terms of advertiser behavior. Maybe it's because of the changes in the NewFronts, up fronts that you just talked about. What do you think is driving that shift where obviously you've seen a little bit of moderation and display and DV+, but more than offset for the acceleration in CTV?
No, it's just a continued explosion of transition to programmatic, and it's sustainable. As we mentioned, there's still a vast majority of activity that is direct sold and does not run across programmatic channels. You're seeing that adoption. Disney has a stated goal of, I think, 50% of their activity being run programmatically by the end of this year, if not higher. There's better targeting, better data usage. As Nick mentioned, performance was kind of the buzzword at this year's up fronts. You can't get the performance and the targeting granularity that you want unless you're running programmatically. I think all those factors are kicking in. It's an exciting time for us.
That growth rate that you're seeing of 25%-30% year-over-year in the last handful of quarters in CTV, we think is very sustainable and is the new normal for us.
Got it.
I think that we did see a shift of DV+ moving into CTV that's kind of resetting the investment calendar as you entered into 2026. You saw a little bit of testing of that in Q4. Those allocations as they were reset are holding. What we're seeing now is the improvement we're noticing in our DV+ business relative to where we started the year isn't a reversal of that. It isn't a re-acceleration of that. What's happening is there's areas of our DV+ business that are very healthy and growing. Think of it as the mobile app side of the business, our commerce media side of the business, streaming over mobile and desktop, whether it's audio or video streaming over mobile and desktop. There's some very healthy areas to that.
Our number of commerce media partners continues to grow. Those that are in market scaling and ramping continue to grow. I want to make sure that we're clear that there's, call it half or more of the non-CTV, DV+ business that's now less than half of the total business. There are some very healthy elements to that that should continue to be growth drivers for us going forward.
Got it. I think you've said before that despite some of the traffic challenges that traditional display publishers are experiencing, your QPS in that part of the business in general, DV+ QPS continues to basically reach record highs. I mean, that trend continues?
Yes.
Yeah. We'll process almost 2 trillion ad requests a day. There's a lot of volume out there still.
Yeah, I think another thing that maybe surprised us, maybe some other people too, was just the maybe diminution of managed service, that you could see in the reduction in the TAC rate in the business. I think that probably speaks to even more so the strength of CTV spend is probably not even fully reflected in the contribution ex TAC. Maybe you could speak to that a little bit and, then, particularly in a political year, do you think that we could see a little bit of trend reversal there where more of that business has come in through managed service traditionally?
Yeah. Good question. Managed service, just to expand on that a little bit, is a business that SpotX had. It focuses kind of mid-market on independent ad agencies, advertisers, think regional car dealerships, tourist boards, healthcare systems, and typically, those players have not had in-house programmatic expertise, and so we've basically acted as their in-house programmatic expertise. We would sling kind of a normal IO, but then we would run that programmatically, and that had kind of higher take rates, 40%+ . What you've seen is a rapid transition for those businesses to bringing their own programmatic businesses in-house, which is what we've expected. This managed service business has actually had a longer shelf life than we even expected.
To put it in context, a year ago, Q1 of 2025, we had $7 million of contribution ex TAC in managed service. In Q1 of 2026, we had $2 million. It's a de minimis business for us now, and I don't expect I'll actually talk about it in the future. To your point, if you actually strip that out on a pro forma basis, our CTV business, the programmatic CTV business, actually would've grown 38% year-over-year in Q1 rather than 30%. You can see the health of the underlying programmatic CTV business in those numbers. Shifting to political, actually, we don't run a lot of that political through that mid-market group. Most of our political actually just runs directly programmatically. These midterms are going to be really interesting. I've read estimates that total spend will grow from 10%-20% in this midterm versus last midterm. For us, we had about $11 million of contribution ex TAC in 2022.
We expect that to be 20%, 25% greater this year, or at least that's what we built into our formal expectations and our guidance. That said, CTV and programmatic are the highest growing component of political, and so I could see that growing even more than what we've incorporated into our results right now. We'll be a little conservative until we actually see that happening, but there's certainly some potential upside there.
Got it. I know it's early, but we've asked some people and they say, "Okay, we've seen a little bit of booking so far." Any indication from the early bookings as to the strength? Is that basically consistent with the views?
Yeah. We have a dedicated team in Washington, D.C. that focuses, it's a very unique ecosystem, on this political spend, and so we've invested a lot in those relationships. The actual bookings, 90% of that political spend will run in September and October. At the end of the year. You don't have the specific bookings, but from an interest perspective, relationship perspective, getting our agreements in place to work with folks who have money to spend, it's all looking very positive.
Okay, great. Yeah, I think that as you've seen the mix shift to CTV, which has been, to your point, very strong. There historically has been a concern, okay, you'll see a reduction in blended take rate in the business. It doesn't seem to be playing out strongly. Are the offsets there in terms of, okay, you're doing more data enablement, you're doing more with the publishers. You're going deeper on the ad serving side. Just wondering if you could maybe sort of unpack some of those dynamics of what appears from, we're guessing, supports the CTV take rate.
Yeah. Well, the other important side of that equation on take rate is volume. We got to keep that in mind. Volume has been significant. From a sort of, if we want to de-average our take rate, the downward pressure right now actually comes from the reduction in managed service. If you think about managed service, with a 40%+ take rate, and that dropped 70% year-over-year. That has a slight downward pressure. If you look at the rest of the business, it's actually stabilized pretty significantly. You have a baseline lowest take rate business where we're the pipes, and we don't really play a role in helping to bring demand into that process. What you're seeing is still continued significant volumes at that layer.
You're seeing a growing component where we're playing a bigger role in bringing demand, and that's where we earn that higher take rate. The mix effect, we'd have a lower mix effect from the higher volume at the lowest take rate is being offset to a large degree by some of that growth in playing a bigger role in demand. It's kind of holding equal right now, from a mixed perspective. As you can see, with 38% growth ex managed service, it's super healthy and take rate changes are not really impacting the trajectory of the CTV business.
Got it. I want to go back to commerce media. You had some comments there earlier. Just again, wanted to maybe ask you to unpack some of the drivers of the wins that you've had there, and if you maybe talk about the growth path with those partners after you sort of light them up and launch them.
Yeah, I think we're very excited about the fact that more and more folks are coming to us on the commerce media side and signing exclusive arrangements with us. Historically, when we'd add a publisher, we would add them, Google would get their fair share of it, or unfair share of it, plus a whole bunch of other players that they would tag to be able to sell their inventory. What a lot of our commerce media partners have done is decided that they want one partner to be able to use their first-party data, share their data, and either use it across their screen inventory, where they hire us to sell ad units that they own themselves, or use that to be able to buy in other ad formats, but apply very, very rich first-party data.
Think of a Pinterest wanting to transact in things like connected television as a good example. You've seen that playbook really come out of people that saw our Netflix relationship and said, "Hey, wait a second. You're working for a walled garden. They have phenomenal data they don't want to share. They want one party to be able to manage it for them." We've started playing that role, and having ad-serving capabilities has been a very important distinguishment, even in non-CTV, to be able to help service, run, manage ads for them, yield manage, and do a lot of things where multiple parties, very tough relationships to try to put together and structure. We're able to do it as a one-stop shop for them and execute and be the most scaled player to access the most demand in the market.
We've gone from 7 to 10 to 15 to 23 partners now on the commerce media side. We've got 15+ that are now ramping or have already launched. I think as you go forward, could that number double two, three, four years from now in terms of our total number of partners? Certainly, there's more and more coming. I think you saw Expedia and PayPal being some of the recent ones that we've recently signed. I think that as you continue to move forward, you're going to have some of these that are very substantial, that are fully leaned in, that are very meaningful growers for us.
I think there's half of those that either don't get it right, don't lean in, don't really fully participate, may not have the right resources or the tech in place, may not necessarily develop the businesses, but there will be a robust business if even our batting average is only 50% in the marketplace. We feel really good about that business five years from now. That will be a significant part of our DV+ business.
Got it. Do you see any indications, well, may you open the door with offsite or audience extension, then you begin, there's potentially onsite opportunity as well with some of these partners?
Yeah, it's usually both, right? They usually want both help on their screen inventory as well as how they can leverage and use their data to be able to reach and advertise using very, very good purchase intent.
Got it. You mentioned Google a moment ago, that reminds me of something important. We're in June right now. I think we had closing arguments back in November. Pending, but hopefully soon we'll get some sort of remedies decision. You guys have sized the sort of point of share, I think, and talked about the incrementality on the margin side. Happy to have you sort of summarize that again for us. What's a reasonable way to think about the catalyst in total? How much share, depending upon the scale of the remedy, and how quickly does that emerge?
I guess the final overlay is Google's competitive response. They've traditionally charged a super competitive, supra-competitive take rate in that part of their business. How do you think they respond? Do they try to become more competitive? How are you guys sort of war gaming this, or scenario planning this right now?
I'm happy to.
Why don't you start and I'll?
Yeah, I think you're right. We're in overtime on a decision. The rocket docket, we would have expected that one year past the verdict, guilty verdict decisions last April. You're being generous by saying closing arguments November, and this remedy hearings in September of last year. We're one year past the verdict of guilty on Google on the ad server side and the exchange side of the business. That verdict will not change. It's really now just about what remedies get put in place structurally and behaviorally to be able to address and undo or cure what a guilty verdict looks like in those markets. I think people have quickly tried to make correlations to search. In search, they weren't found guilty.
They were found to have used monopolistic practices. It wasn't as black and white clear with a guilty verdict that they were monopolistic, that they did run a monopoly and advantage themselves. A lot of people have asked us the question, have they started making some changes? There's some things in Europe that were changed with UPR and Unified Pricing Rules for those that don't know the acronym. Have things started to change? Have you started seeing a nicer Google because they started playing ball, make some changes voluntarily? No.
We've seen no impact in Europe. The fact that it is a monopoly and they treat their exchange with different data, different rules, different latency, completely differently than they do any other exchange that competes, that is the monopoly, that if it is unwound, will yield percentage points of market share ceding to the rest of the industry and the market becoming more fair. Even the win rates that were quoted during the trial. In the case of all other non-Google exchanges, they were in the very, very low single-digit range. You could think it has a one handle in front of it, but there's nothing to the backside of the handle.
On their side, it was in the 20s in terms of win rate. You don't get there by naturally competing in an open and fair marketplace the way that market performs. We think there's some elegant, simple solutions by which to be able to play and to open up that inventory. It's called Prebid. You run it through a certain piece of code that then distribute it to all parties, including Google, fairly and evenly. We think that would move the market. Again, we're, I guess, impatiently waiting just as you are and checking our phones even multiple times. I should probably check if the news broke during this meeting, because we're literally could happen at any minute.
Yeah. To reiterate some of the sizing on that, as you mentioned, Google and DV+ has 50% or 60% market share compared to, we're the next largest with 6% or 8%, there's quite a divergence there. Just a shift of 1% of 100 basis points of market share to us could represent $50 million in revenue. Because it's about win rate, as Nick has mentioned, we basically have sunk costs with all of our activity. We're already processing those several trillion of ad requests a day, to the extent that your win rate goes up, there's no incremental cost. 90%+ of that incremental revenue would flow through to adjusted EBITDA margin and to free cash flow. It doesn't take a lot of market share gain to have a pretty significant impact for us financially.
Got it. The EU has its say on this, and of course, what the EU says can impact what happens globally and what the U.S. says can impact what happens globally. Have you actually seen some changes on the ground? Have they actually implemented the change to UPR which was proposed, or have you seen any movement there one way or the other?
Not anything that's been significant. Those that UPR takes, those publishers have to react to that and there's not been a significant impact from that at this point.
Okay. Now, just given the potential, maybe the EU goes a little bit further over time, or maybe they do actually make some more significant changes. I think they've had that market mostly on sort of lockdown for a long time. Does that potentially become a more attractive investment target for you, more international growth?
It's interesting. Because we're already processing all of those ad requests, we really don't need incremental investment into the market. Given their market share in Europe, any shift in share would fall. We'd get a nice take from that, but we don't really have incremental investment that would be needed. If you pan back globally, internationally, it's 25% of our business today, so it's not small. Where I think the real international growth opportunity lies is in CTV. We are actually making some investment internationally on the CTV side. What you've got going on there is, from a legacy perspective, you've had an often state-funded legacy broadcasters who've been semi-monopolistic in Europe, Australia, and whole bunch of places in Asia.
Now you've had competition that's come in from all the global streamers, and so that has put those legacy broadcasters on the defensive. They have not been big adopters of programmatic nor of streaming, and so now they've had to react to the Disney+, the Netflix expansion internationally and so forth. We're getting great traction with those legacy broadcasters wanting to transition to programmatic, and they're also all launching their own streaming services that are advertising-supported. We're seeing quite a bit of traction on that front.
Got it.
I would add that this follows a couple of if/then statements that would have to kind of prove true in order to put investment dollars behind it. One of the reasons that this monopoly from Google exists in non-CTV is that they've effectively taken their ad server, given it away for free, taken 99.9% market share, and made it uneconomical for anybody to compete in the ad-serving business. They then, without negotiating fees as an exchange, charge whatever they want, as well as on the DSP side, can charge what they want without having to have a contractual relationship with a publisher to do so. We've been asked repeatedly by publisher partners, would we be an ad server?
At these economic levels, it makes no sense to become and get into the non-CTV ad-serving business as a standalone entity. If you follow another if/then statement saying, "Hey, something changes," and now the ad server is starting to price fairly for what value that they create, and that is a good, healthy, standalone business, and that charges a reasonable nominal fee for what that business is on its own, and the exchange has to then become competitive. To the other question you asked, do they become more competitive? Do they charge prices differently? Now there's a rebalancing of the types of fees charged. Could the ad-serving business become something that now looks attractive to us if the pricing in the market environment has changed, where people have, for a long time, been looking for an alternative to Google, yet it hasn't made sense of how they've run the business,
Got it. I want to spend a little time on some numbers.
All right, sure.
Yeah, we'll finally get to that.
Bring it on.
Yeah, a little less theoretical. I want to talk about the margin guidance for the full year. You raised that just a bit. I think it was about 50 bps.
Yeah.
Just wondering, how is that coming a bit better than you expected? Any key savings or leverage drivers that surprised you versus your initial view? Maybe we'll start there.
Yeah. Two drivers. A primary driver is in our tech stack costs, and so that margin expansion is all on the cost side. That's assuming the revenue that we talked about at the beginning of the year for this year. We have a significant initiative to move some of our CTV activity from the cloud to on-prem. We made some extra CapEx investment last year, and some of those on-prem data centers are coming online now. To put it in perspective, you can run the same activity for about a third the cost in your own data center, versus running it on the cloud. What we're seeing is that our transition has come in on time or a little bit ahead of schedule. It's working the way we thought it would work. Proof of concept is actually working out.
That's been super exciting. That's in the face of some additional headcount investment that we've made this year. We're really well set up for additional margin expansion, I think, 2027 and beyond. Just from a cost basis, you could expect even at current revenue growth levels, I think 100, 150 basis points a year at a minimum of increase in the future. Then the second important point there is internal productivity. That's probably been my biggest surprise is it's not the primary driver, but a secondary driver, and using agentic AI tools, we've been able to reduce a lot of offshore contractors and other internal costs. That's been surprising to see the uptake in productivity gains that we've seen from that. More to come on that front, but that's been fun to see.
You have a token leaderboard.
Yeah. No. Because token is cost. We have a productivity factoring in cost of agentic use that we're focused on.
Got it. We just got a few minutes left. Just want to check and see if we have any questions in the room. No, I guess we're good. Yeah, just maybe finally touch on capital allocation. It seems like that move to on-premises sounds like it's been successful and I think maybe not nearing quite completion, but--
We got some more. Wood to chop on that.
Yeah. With that, obviously still some to go there, but probably we think transition to a lower gear, if you will.
Yeah.
How should we be thinking about CapEx or maybe maintenance CapEx, if you will, following that transition. With the moderation going forward, infrastructure requirements of the business, how are you thinking about your capital allocation priorities?
Yeah. On the CapEx front, we had a extra pop of an extra $20 million last year. I think going forward, we'll have a run rate of about $60 million a year, which will include additional transition to on-prem, but also replacement. You have a normal replacement rate of all of our other on-prem activity. With free cash flow now that we're generating, currently we've dedicated about 50% of that to share repurchases. We've announced a $200 million program for share repurchases over the next couple of years. There's still $180 million remaining on that program. In addition to when we have shares at vest, we have a withhold to cover process. We're effectively repurchasing shares, but we're keeping shares from coming on the market from that perspective.
Yeah, we're focused. We think we're really undervalued and want to take advantage of that right now. We're setting aside some dry powder. We acquired, last fall, streamr.ai, which was a smaller acquisition, but it's had a huge impact on us. We'll be on the lookout for additional tuck-ins or smaller acquisitions that could accelerate our product roadmap. Just keep some dry powder if something interesting comes along from an M&A perspective.
Okay. Sounds great. David, congratulations on the retirement announcement. Well, number one, what are you planning to do?
Yeah.
What are you planning to do next? Any update on the CFO search?
Yeah. Well, I appreciate that. I'm still actively in the seat until September. We have an active search going on. We have some great internal candidates as part of that process. Todd's meeting to some great external candidates as well. We'll move forward with that over the next couple of months. Yeah, that's why I'm just walking around with a big smile on my face and we got our third grandkid in the family and looking forward to some family time, a lot of mediocre golf and traveling. Excited about that.
Okay. That's great. This will be your final investor conference?
Yeah. Well, I might have some in August and September.
Okay. It's your final.
Yes. We'll just go out with a bang with you.
Totally.
Yeah.
Yeah. We'll go with that. Well, anyway, thanks again. Congratulations, and thanks to both of you for being here. Really appreciate it.
Thanks, Rob. Appreciate it.
Thanks, Rob.