Magnite, Inc. (MGNI) Earnings Call Transcript & Summary

Hello, and thank you for standing by. Ladies and gentlemen, welcome to Magnite Q2 2026 Earnings Call. Please note that this call is being recorded. [Operator Instructions] I'd now like to hand the call over to Nick Kormeluk, Investor Relations. Please go ahead.

Thank you, operator, and good afternoon, everyone. Welcome to Magnite's Second Quarter 2026 Earnings Conference Call. As a reminder, this conference call is being recorded. Joining me on the call today are Michael Barrett, CEO; and David Day, our CFO, for his final earnings call prior to retiring. I would like to point out that we have posted financial highlight slides on our Investor Relations website to accompany today's presentation. Before we get started, I will remind you that our prepared remarks and answers to questions will include information that might be considered to be forward-looking statements, including, but not limited to, statements concerning our anticipated financial performance and strategic objectives, including the potential impacts of macroeconomic factors on our business. These statements are not guarantees of future performance. They reflect our current views with respect to future events and are based on assumptions and estimates and subject to known and unknown risks, uncertainties and other factors that may cause our actual results, performance or achievements to be materially different from expectations or results projected or implied by forward-looking statements. A discussion of these and other risks, uncertainties and assumptions is set forth in the company's periodic reports filed with the SEC, including our quarterly reports on Form 10-Q and our 2025 annual report on Form 10-K. We undertake no obligation to update forward-looking statements or relevant risks. Our commentary today will include non-GAAP financial measures, including contribution ex-TAC, or less traffic acquisition costs, adjusted EBITDA and non-GAAP income per share. Reconciliations between GAAP and non-GAAP metrics for our reported results can be found in our earnings press release and in our financial highlights deck that is posted on our Investor Relations website. At times, in response to your questions, we may offer additional metrics to provide greater insights into the dynamics of our business. Please be advised that this additional detail may be onetime in nature, and we may or may not provide an update on the future of these metrics. I encourage you to visit our Investor Relations website to access our press release, financial highlights deck, periodic SEC reports, and the webcast replay of today's call to learn more about Magnite. I will now turn the call over to Michael. Please go ahead, Michael.

Thank you, Nick, and thanks, everyone, for joining us today. I'm pleased to report an outstanding second quarter for Magnite. We significantly exceeded expectations across the business. Contribution ex-TAC came in well above consensus, driven by strength in both CTV and DV+, and that translated into meaningful bottom line outperformance. We expect this momentum to continue, and based on our first half results and the strength we are seeing across the business, we are raising both our full year contribution ex-TAC outlook and our expectations for margin expansion. Total contribution ex-TAC exceeded consensus by approximately $10 million. CTV contributed roughly $6 million of that outperformance and grew 36% year-over-year. DV+ contributed approximately $4 million of the beat and returned to growth, increasing 2%. Adjusted EBITDA exceeded consensus by $8 million, resulting in a margin of 37%, demonstrating the operating leverage in our business. These results were broad-based. CTV represented 51% of total contribution ex-TAC in the quarter, continuing the momentum that began in the second half of 2025. We believe the market has reached an important inflection point as programmatic becomes the desired way to transact on streaming television. We saw strong growth across many of the industry's largest media owners, including Disney and ESPN, Netflix, Roku, VIZIO, Walmart and Warner Bros. Discovery. Across our top 10 CTV accounts, growth accelerated to the mid-to high 40% range year-over-year. While the secular shift of advertising dollars towards CTV continues, we are encouraged by the improving trajectory of DV+. In particular, mobile in-app grew 17% year-over-year. We continue to believe mobile in-app is an attractive long-term growth market, supported by deeper DSP integrations, new publisher onboarding and our SDK strategy. To be clear, this is both an industry and Magnite share growth story. Our results demonstrate that we are expanding our share with growth that is outpacing the broader market as customers increasingly choose our platform. Stepping back, there are 3 structural trends driving our business today. First, SpringServe has become the operating system for CTV monetization. Second, audience enablement and decisioning are moving from the buy side to the supply side, and we believe Magnite is leading that transition. And third, as AI reshapes advertising, our newly announced Magnite Orchestration shows early signs of becoming a critical infrastructure layer for agentic advertising. Taken together, these 3 trends are improving our long-term competitive position and growth prospects. SpringServe remains our primary differentiator. What began as a best-in-class ad server has evolved into the operating system for CTV monetization. It has become the intelligent control layer for premium streaming, combining ad serving, mediation, monetization, demand facilitation and data enablement. And these functions are increasingly being enhanced by agentic tools. Publishers want to work with a trusted partner capable of maximizing yield while preserving control over their inventory, data, pricing, business rules and viewer experience. Buyers want direct, transparent and scaled access to premium streaming supply. SpringServe uniquely sits directly between those objectives. The power of SpringServe is evidenced by a string of major new wins and partner expansions. On the publisher side, we announced Samsung selected SpringServe to power ad serving for its premium smart TV home screen inventory, reaching hundreds of millions of smart TVs globally, and to open this inventory to programmatic buying for the first time through our DSP ecosystem. We are excited to add Samsung as another key home screen customer, solidifying Magnite's leadership position among OEMs in this increasingly valuable environment. On the buy side, WPP has expanded ad formats in its media supply hub enabled on SpringServe to include Pause ads and has validated Magnite's ability to seamlessly pair CTV ad formats with WPP's open audience segments via ClearLine through a custom real-time data integration. Moving to the second structural trend, the acceleration of supply-side audience enablement and decisioning. Historically, many of the most important optimization decisions in digital advertising were made on the buy side. Today, publishers and buyers have access to richer first-party data, commerce signals, AI, pricing intelligence and much more workflow flexibility. And as a result, more valuable decisions are moving toward the supply side, and the breadth of our relationships and technology position us well to capture the shift. One of the most compelling applications of supply-side audience enablement and decisioning is commerce media. Commerce media continues to scale with 21 partners now deployed and actively ramping across DV+ and CTV. Partners, including Fanatics, CVS Media Exchange, Best Buy and PayPal Ads are all using Magnite to activate valuable first-party data across owned and operated inventory and the broader open Internet. And with our partnership with Walmart Connect, we are helping combine Walmart's first-party commerce data with premium CTV inventory, including VIZIO supply, while supporting offsite execution and closed-loop measurement. Now turning to AI. Earlier this year, much of the discussion focused on whether companies like Magnite could be disintermediated. Today, the conversation has largely changed. Our customers are increasingly leaning on us to develop and deploy AI capabilities within our platform, converting AI into a tailwind. Buyer agents and seller agents will become increasingly common across digital advertising. But agents do not eliminate infrastructure; they increase the need for it. As thousands of agents from publishers, marketers, data providers and measurement companies interact simultaneously, someone must coordinate those interactions. Someone must discover inventory, interpret campaign objectives, package audiences, enforce publisher controls, protect privacy, optimize monetization, clear transactions and provide the trust required for advertising to function at scale. We believe Magnite is uniquely positioned to play that role. Last quarter, we announced our seller and buyer agents. Our seller agent allows publishers to seamlessly create custom inventory and audience packages that are discoverable and purchasable by buyer agents, while our buyer agent enables buyers to create custom media plans from simple RFIs, generate ad creatives, and activate and discover audience opportunities. This quarter, we took the next major step by introducing Magnite Orchestration. Earlier, I described SpringServe as the operating system for CTV monetization. As AI reshapes advertising, we believe Magnite Orchestration has the potential to become the critical infrastructure for agentic advertising. Rather than simply introducing another AI agent, we are building the orchestration layer that enables any agent to work together in a trusted environment across a scaled independent marketplace. Disney Advertising, Spectrum Reach, Kepler, MiQ, Publicis Media Exchange, Dentsu, and DIRECTV are already working with different components of our AI suite. These partnerships provide early but meaningful validation of agentic advertising operating across both the buy side and the sell side. As advertising evolves with advancements in AI and becomes more automated, more data-driven, and more interconnected, the value of intelligent decisioning and trusted orchestration only increases. We believe Magnite is uniquely positioned to lead in these areas. If CTV has been the defining growth story for Magnite over the past several years, we believe supply-side audience enhancement, enablement, and decisioning and AI orchestration together have the potential to define the next chapter of our growth. Before I conclude, I would like to recognize David. As previously announced, David plans to retire at the end of September after more than 13 years of outstanding leadership and service to Magnite. David has been an exceptional partner and a trusted adviser. His financial leadership helped guide Magnite through transformational acquisitions, significant industry change, and tremendous growth. Just as importantly, he has built a deep and talented finance organization that will provide an excellent foundation for his successor. Our search continues to progress well, and we are evaluating a strong group of internal and external candidates. On behalf of our Board, our leadership team and everyone at Magnite, I want to sincerely thank David for his extraordinary contributions. With that, I'll turn the call over to David for more detail on our financial results. David?

Thanks for those kind words, Michael; very much appreciated. We are extremely pleased with our second quarter results. As Michael mentioned, we exceeded contribution ex-TAC and bottom line expectations across the board. Given the momentum in our business and the many catalysts driving our growth, we are raising our guidance for the remainder of the year. Total revenue for Q2 was $193 million, up 11% from Q2 2025. Contribution ex-TAC was $190 million, up 17%, well above the high end of our guidance range. CTV contribution ex-TAC was $97 million, up 36% year-over-year, well above our guide of $90 million to $92 million. DV+ contribution ex-TAC was $93 million, an increase of 2% from the second quarter last year, above the top end of our guidance range. Our contribution ex-TAC mix for Q2 was 51% CTV, 35% mobile, and 14% desktop. From a vertical perspective, health and fitness, technology and finance were the strongest performing categories, while automotive, our top declining category in Q1 2026, has returned to growth but remains depressed. Total operating expenses, which includes cost of revenue, were $162 million, up from $151 million last year. The increase was primarily due to increased personnel costs, higher tech stack-related expenses and higher facility expenses. These were offset by lower traffic acquisition costs. Adjusted EBITDA operating expense for the second quarter was $119 million, an increase from $108 million in the same period last year, with similar drivers as previously noted. Our net income was $19 million for the quarter compared to net income of $11 million for the second quarter of 2025. Adjusted EBITDA grew 30% year-over-year to $71 million, reflecting a margin of 37% compared to 34% in Q2 last year. We're seeing encouraging productivity benefits from AI across engineering, operations, sales and G&A. While we are still early, these capabilities are helping us accomplish more, improve execution and support continued margin expansion. GAAP earnings per diluted share were $0.13 for the second quarter of 2026 compared to earnings of $0.08 for the second quarter of 2025. Non-GAAP earnings per share for the second quarter of 2026 were $0.26 compared to $0.20 in Q2 last year. Our cash balance at the end of Q2 was $333 million, an increase from $185 million at the end of the first quarter. Operating cash flow, which we define as adjusted EBITDA less CapEx, was $57 million. Capital expenditures, including both purchases of property and equipment and capitalized internal use software development costs, were $13 million. Net interest expense for the quarter was $6 million and net leverage was 0.1x at quarter end. During the second quarter, we repurchased or withheld over 2.1 million shares for approximately $28 million. Year-to-date, through the second quarter, we repurchased or withheld approximately 4.4 million shares for about $57 million. As of quarter end, $165 million remained available under our current repurchase authorization, which is effective through February of 2028. I will now share our expectations for the third quarter of 2026 and our current thoughts for the full year. For the third quarter, we expect contribution ex-TAC to be in the range of $188 million to $192 million, which represents growth of 13% to 15%; contribution ex-TAC attributable to CTV to be in the range of $98 million to $100 million, which represents a growth range of 29% to 32%; DV+ contribution ex-TAC to be in the range of $90 million to $92 million, which represents a growth range of negative 1% to up 1%; and we anticipate adjusted EBITDA operating expenses to be in the range of $119 million to $121 million, which implies adjusted EBITDA margin of 36% to 38%. For the full year 2026, we are raising total contribution ex-TAC growth to be between 13% and 14%, up from at least 11% previously; raising adjusted EBITDA percentage growth to be greater than 20% from the mid-teens previously; raising adjusted EBITDA margin to be at least 37% from at least 35.5% previously; raising free cash flow growth to be in the high 40% range from the mid-30% range previously; reaffirming CapEx of approximately $60 million, a reduction from the year. On the Google ad tech trial front, we have no updates since last quarter, and our estimates do not include any market share gains that might result from potential remedies. A final note of context for our revenue guide: even with our raised full year guidance, we remain somewhat conservative in our estimates for the rest of the year to properly capture potential macro risk. Finally, on a personal note, I continue to be incredibly pleased with our performance and the robust financial position the company maintains today. I'm very proud of the durable company we've built, our winning culture and our world-class finance team. We have incredible momentum in the business and I look forward to another great quarter and closeout to the year. The best is yet to come. And with that, let's open the line for Q&A.

[Operator Instructions] Your first question comes from the line of Matt Swanson of RBC Capital Markets.

David, you will be missed. Hopefully get to see you a little bit more before you start fishing and golfing. I guess starting on the top 10 CTV accounts, you're talking about the mid-to high-40% growth. Could you just talk about those 3 buckets in which you grow CTV revenue being: increased supply, people moving up the rate card and increasing take rates, or just kind of where that growth is coming from?

Yes. Matt, it's Michael. Yes, I would see generally in 2 buckets. One is just greater adoption of programmatic. It just becomes table stakes in the upfronts to be able to offer buyers the option to buy programmatically. And so each year, we're just seeing a greater adoption of programmatic with the big streamers, the premium streamers, right? And then the second bucket is a willingness to have, instead of publisher-led programmatic, publisher-sold programmatic Magnite demand, Magnite able to come in and bring demand from DSPs that they don't have relationships with, advertisers that these premium publishers haven't had relationships with. And that obviously carries with it a different profile in terms of take rates. So I think that you can look at it in those 2 lenses, the greatest contributors to that.

And then maybe this builds on that second part of your answer there. But you're now connected to the vast majority of CTV or streaming supply out there. And so it feels like a lot of the long-term TAM expansion for Magnite comes with also increasing the amount of demand or the amount of people spending on your supply. Two of the three big themes you highlighted seem like they're at least in some ways about making it easier, if not -- obviously you're not going to be doing the DSP job, but making it easier to buy within CTV. Could you just talk about how that strategy shifts in terms of making it as easy as possible for dollars to flow to all your inventory?

Yes, it's a really good point. Listen, I think if you look at the 3 pillars we talked about, SpringServe being one, I assume you're alluding to audience decisioning and enablement and our efforts in AI with Magnite Orchestration. And I think they both fit that bill, right? The idea is there's valuable first-party data on the media owner side and on the advertiser side and what's the easiest, most frictionless way to surface that inventory. And to your point, you're right, it's not about eliminating DSPs, but you are bringing some very valuable decisioning and enablement onto the supply side that is new. And you see the success we're having with commerce media, that's really the tip of the spear there. They have great data on their side, the Disneys of the world, the Netflixes of the world, great data on their side. What's the easiest way to get that data to match and open it up so that it democratizes DSP involvement so that you just don't have to use one DSP where that data has been housed. You now are able to bring your own DSP and access the data. So that definitely brings more seamless demand into the picture. And AI, obviously it's early stages. To date we've transacted a handful of millions of dollars. And next year it'll be much bigger than that, but it's not going to represent the majority of our overall spend. But building tools that enable these advancements to work closely together in a safe way, a trusted way, is going to be very important for our buyers and sellers to be able to realize the benefits of AI. So yes, I think those 2 areas definitely fit the bill of decreasing friction and bringing demand into the ecosystem.

Your next question comes from the line of Shyam Patil of SIG.

Congrats on the very strong results and some of the industry-leading growth rates there. And David, congrats again, all the best with the next chapter. Michael, I had a couple of questions. One was on CTV, one on agentic. On CTV, again, very strong growth rate. I think it might be the highest in the industry right now. I know you talked about strength over the past couple of quarters just being broad-based across the board. Just wondering, for this past quarter, was there anything that really surprised you to the upside, like 1 or 2 things that really, really surprised you? And then second, on agentic, it sounds like this could be a pretty significant opportunity. I heard you talk about it for a while now, just in terms of the momentum there. I was just wondering if you could talk a bit more about what customer conversations are like right now? And then when this starts to ramp, I know timing is always tough, but when it starts to ramp, do you think this is something that can really inflect the growth rate?

Yes, sure. So on the CTV front, as we cited in the script, it was quite broad-based. So it wasn't certainly led by one publisher. I think the 2 things drove it. Number one, just increasing adoption of programmatic by the buyers and media owners, particularly with our premium accounts. And secondly, international growth. When these big streamers expand internationally and global, we go along for the ride and they really lean heavily on us in the programmatic channel to activate demand in these markets because they don't have necessarily boots on the ground to sell direct. And so I think you saw it in Disney's earnings, they cited international growing faster in the programmatic bucket, and we can attest that we see that as well. So I think you're just seeing broader-based adoption of programmatic and international expansion, which comes almost as a programmatic-first expansion. And as far as agentic is concerned, customer conversations, it's the topic. We recently were at an industry event in France and there wasn't one conversation you had with customers that didn't involve agentic. It's early stages. I think we feel really pleased with the level of investment we made in it. I don't think we've overinvested, but we're ready for it when it comes. I think we've been leading the discussion in the industry, so we feel good about that. It's really hard, as you pointed out, to pinpoint when the tipping point occurs. We end every big conversation with customers asking them what their prediction is for 2027 and you get a range of 0 to $1 billion in terms of for the whole industry. So for a company that's going to do $9-plus-billion in ad spend, if $1 billion for the whole industry gets transacted, it's not all that meaningful in terms of impact to our financials in the near term. But I definitely think mid-to long term, it will definitely be a growth driver, largely because I think you'll have more money put to work. So the working media will be larger. And I think that you're going to have TAM expansion because a lot of the experiments that we've seen have been direct IOs that have been converted to the programmatic channel. So that will bring in new dollars into the TAM of programmatic and that's a positive too.

Your next question comes from the line of Jason Kreyer of Craig-Hallum.

I couldn't draw up a better quarter for David's last quarter and for investors to see the David Day effect. So congratulations there. I'll start with a question for you, David. So just wanted to get an updated view on the political environment. We're hearing positive trends there. I'm curious if you're thinking any differently about what you were embedding into the guide previously versus the updated guide today.

Yes, that's a good question. And just to level set again, 4 years ago in midterms, we had about $11 million in contribution ex-TAC. Presidential, we had about $19 million. We entered this year targeting something in between those. We've continued to include that level in our forecast. That said, I think the primaries ended up a little stronger than we might have anticipated. And so we're cautiously optimistic that there could be some additional upside there. It's so hard to handicap given the volatility in these races, candidates in and out and how competitive they're going to be. But yes, we do think there's some -- hopefully some upside in the political realm.

And a follow-up for you, Michael, sticking with the agentic topic, I want to ask about just orchestration and your seller agent. Can you just talk about the strategy to get publishers to utilize the seller agent, how that adoption progresses and then how that catalyzes orchestration to be this critical infrastructure layer that you had called out?

Yes. Great question, Jason. So all of the experiments that we have transacted, all the buys, have involved the publishers using our seller agent. I don't think success for us looks like everyone has to adopt Magnite's seller agent. The idea of the orchestration layer is to allow people to bring whatever tool they have and be able to have it work seamlessly with the other side of the fence. So if you're a buyer working with seller, seller working with buyer, data provider, et cetera. So we really think the future is being able to be this trusted partner that allows inventory discovery, execution, clearance, brand safety and I think that only can be accomplished by someone as scaled as we are. I think you're going to see far fewer competitors of ours in an agentic world. You don't need multiple orchestration layers and that's why I think we feel so bullish about the prospects of Magnite Orchestration. Just like SpringServe is that operating system for CTV, we think we have a real fighting shot to be that operating system for the agentic-enabled advertising world and feel very good about the level of investment we've made here.

Your next question comes from the line of Laura Martin of Needham.

Michael, I want to stay with agentic. And the minute you tell me that you want the agentic layer to be similar to SpringServe being the OS, SpringServe has a horrible margin and a horrible take rate. So I'd like for you first to address, is the orchestration layer going to be free and/or is it going to have a better take rate than SpringServe, A? B, when we were talking about agentic in Cannes, you were really quick to say that it was a new total addressable market because it was basically a workforce automation tool for linear TV moving into CTV, automating CTV, which to me the upside there was a $50 billion TAM you guys have never touched. So to me, that's the primary agentic benefit so far having nothing to do with agentic. So am I just thinking about the 2 ideas not integrated enough?

No, a great question, Laura, or questions. So SpringServe as you know, plays a myriad of roles in our technology suite. There is SpringServe, as you pointed out, the ad server. And ad serving takes a different take rate than mediation or demand facilitation. But SpringServe has now become embedded in all of our platforms, so there's an instance of SpringServe in everything that we do. So to characterize SpringServe as a very low take rate product might refer to it from an ad serving standpoint, but it's certainly not the case for SpringServe enabled across the Magnite technology suite. So SpringServe is not a low take rate. In Orchestration, we intend to charge for it. It won't represent SpringServe, the ad server. Orchestration will do many, many, things that folks will value and we'll be able to charge appropriately for it. And to date, all of our transactions that we've done agentically have carried with it very -- a similar take rate structure to our normal suite of products. And you are absolutely right and did note that in one of the questions and answers that we see TAM expansion with agentic because what we have seen to date has been one-to-one deals that normally would have been processed outside the programmatic ecosystem as direct-sold deals and now we're seeing it brought into the programmatic ecosystem. And so you're absolutely right, there's a TAM expansion involved in agentic that will take direct dollars and bring it into programmatic.

Okay. And then my follow-up question -- thank you for that, that's helpful, especially on the take rate stuff. My other thing, Michael, is one of the ways Magnite is different and not better from my point of view is you're adding FTEs at the speed of light at a time when we're saying that technology should be replacing employees. So you clearly disagree with me. So could you please tell me why we have to be adding all this headcount at a time when I think tech should be replacing people?

And we definitely share that worldview. The people that we are adding are mission-critical. They're generally engineers. We didn't over hire during the pandemic, which a lot of our tech peers did. And so a lot of the folks that are shedding bodies are shedding extra bodies. We see this opportunity as being so rich and the path for Magnite so clear that adding 100 people over the course of a year, we don't think is counter to the notion that AI is making us more efficient. If you look at what AI has done for us internally, I'll give you 2 examples of big cost savings for us that involved headcount in one instance. And that is we no longer are working with any contractors in our ops organization. We've built agents that do that work for us and we've been able to let go all those -- they're not FTEs, but it's real cost for the company and you've seen that in the margin expansion. The second piece is on the engineering side, we've been able to build our own load balancer and not have to use Amazon's. And that's resulted in $20,000 of savings on a daily basis. So we are experienced in enjoying AI from a margin expansion. The people that we hired, mission-critical, superstars and they're going to help us get there faster.

Your next question comes from the line of Robert Coolbrith of Evercore ISI.

David, congratulations once again on a great run. Michael, we love you too. And just wanted to ask, maybe related to Laura's question. But some of the early work that people are doing on agentic, it seems like it's less real-time decisions. But wanted to ask you, in the fullness of time, do you believe the agentic infrastructure stack or workflow stack or however you want to talk about this, will that include a robust decisioning and auction component to it that maybe addresses some of the questions Laura had? And then also, I wanted to ask a little bit about just any -- not onetimers, but cyclical events that may have contributed to some of the Q2 strength. Any callout on World Cup in particular? Anything you can tell us about that?

Yes, Rob, good questions. Yes. So on the agentic side, you're very accurate in pointing out that most of the agentic that has been ballyhooed has been one-to-one, publisher to buyer. And a lot of folks question, can these agents do one-to-many. And that goes right to the heart of our argument for Magnite Orchestration, that you're not blowing up the infrastructure because you're right, you can't do this with [ just ] agents. You need the infrastructure that exists today. And so our scale, our server farms, our cloud capabilities, most definitely, agents can do one-to-many and do RTB, but we will be the processor, we will do the transaction, we will run the auction. So it's going to be done on our rails and the interfaces will be agentic. And that's the world we believe in and that's the reason behind Magnite Orchestration. As far as one-timers in Q2, there really weren't any to speak of. This is broad-based. World Cup, we marginally participated. Most of that was linear. Most of that was linear pass-through to even when it was streaming. And so World Cup didn't turn out to be that huge sporting event for us that we're going to have to worry about comps going forward. So yes, Q2 is pretty clean. You're not going to hear us worrying about any onetime nonrecurring comp problems in 2027. Of course, David won't be around to worry about that, but that's...

Your next question comes from the line of Tyler DiMatteo of BTIG.

I guess at a higher level, guys, how do you think about the sustainability of that CTV growth rate? Obviously things have accelerated and they're very good and they continue to outperform. But when you take a step back and look at the multiyear view on that growth rate, I guess, how do you think about that sustainability? That's my first question. And then secondarily, obviously margin upside. I'm curious from here, where's the opportunity to continue to pull costs out of the business and see greater operating leverage from here as you take a step back on that front as well?

Sure. I'll talk about the growth and David can address the operating leverage. So our stated goal has always been to outpace the market in terms of growth when it relates to CTV. And presently, by any estimate, we're 2x, 2.5x, 3x the market growth rate. Is that sustainable? I think you'll have ups and downs on that. But we fully believe that looking out several years, that a 25% growth rate for CTV is something that we not only aspire to, but we think is achievable. And that, of course, then translates into margin expansion at that front. But we think there's a -- if you look at the industry estimates, low-teen to mid-teen growth right now at CTV. And will we always be 3x that? Probably not. But you can, I think, consistently see us as someone being multiples of the industry growth rate.

Yes. And on the margin front, I think a couple of factors to think about. As we have revenue growth that just gets into the double digits, you see incremental flow-through to EBITDA and to free cash flow at pretty high rates. And so you'll see natural margin expansion even with some of our current cost levels and cost growth levels. That said, on the cost side, I think you're going to see continuing gains as we continue to work on our tech stack costs. So those tech stack cost gains come from 2 fronts: one is as we continue to get more efficient in working in the cloud; and then second -- and Michael mentioned this load balancer project that we had recently was just one example of that. And then second is as we move more and more of our activities from the cloud to on-prem, which can be up to 3x more cost efficient over time. And so I think you'll see those factors. And then third, from a headcount perspective, we have added a few heads, but we think that's been the right thing to do. There's so much opportunity. We're getting more productivity, but we want to double down because of the opportunity ahead of us. But that will also not stay the same and we're very cognizant of headcount and headcount-related costs. And I think you'll see that turning a different direction at some point in the future and that's another additional bucket of cost savings, all of which point to -- you've seen the tremendous increase in our margin just in the last quarter and through the rest of the year. And I think that margin will continue to expand. Historically, we've talked about long-term margin ranges of the 35% to 40% range, and we're going to start bumping up against the top end of that. But there's no reason to think that 40% is a cap on our potential margin and we have opportunity to certainly exceed that down the road.

Your next question comes from the line of Shweta Khajuria of Wolfe Research.