Dynatrace, Inc. (DT) Earnings Call Transcript & Summary

All right. Continuing the session before lunch. I'm Sanjit Singh, I'm the infrastructure software analyst on the Morgan Stanley software research team, thrilled to have the management team from Dynatrace. Chief Executive Officer, Rick McConnell; and Chief Financial Officer, Jim Benson. Rick and Jim, thank you for joining us at the TMT conference once again.

Thanks, Sanjit. Always good to be with you.

Awesome. Before we get into the discussion, for important disclosures, please see the Morgan Stanley research disclosure website at www.morganstanley.com/researchdisclosures. I'm going to start out with my comments, I'm a big bull on observability in the category, not just for this year, but going forward, particularly in the agenetic world. We'll dive into those conversations. Just to give us some context, Dynatrace is coming off a couple of strong quarters. Net new ARR is staying at 16%, constant currency for the second quarter in a row. The ARR base is now up to $1.9 billion. You got an operating margin in the high 20s. Trailing 12-month free cash flow margin of 32%. So squarely in that rule of 40 plus territory. We are at a time in the market though, where there's a lot of revisiting of first principle thinking when it comes to software providers and how they pave value for their customers.

So Rick, with that as a context, what are the core problems that Dynatrace is helping customers solve today? And what problems will Dynatrace help customers solve going forward?

The simple way to, I think, begin, Sanjit, is number one, we deliver through observability, resilient software. And at the end of the day, there is no time in history that we can recollect where delivering software that worked perfectly was more critical. So that is a foundationally #1 in terms of priorities. And in the observability space, I would say, overall, we are at the -- we are in an era where observability generally is becoming absolutely mission-critical to essentially every company in the universe that is delivering core software. The second piece that is evolving and evolving rapidly by way of problem we're trying to solve is reliable AI. Because it isn't just about delivering software that works. It is about delivering AI-first workloads that actually are delivering the outcomes, delivering the content that you're expecting to be delivered.

Yes. That makes a ton of sense. And maybe we can dive a little bit deeper in terms of the product innovation cadence at Dynatrace. There's a lot of buzz coming out of your conference, and that conference is called Perform at the end of January. We'll get into the specific product analysis, but the big theme from my point of view was around Dynatrace Intelligence. And this representing the next major evolution of the Dynatrace platform. So can you give us a sense of what makes up the Dynatrace Intelligence platform. What capabilities and value will this unlock for customers?

So Dynatrace Intelligence, we announced at our Perform Conference about a month or so ago. And Dynatrace Intelligence fuses together the innovations of deterministic AI along with Agentic AI. Now deterministic AI, we look at it, Dynatrace, is really our super power. This is what we would argue we do better than any other observability company on the planet. And the reason is because we've had 20 years of context of building systems based on an underlying data lake house with Grail, a software top logical map using wrapping and Smartscape, elements of artificial intelligence, beginning with causal AI to predictive AI to generative AI, all of which are designed to indicate to a software developer or software provider, specifically what is happening in that software environment at any given moment. And when something breaks, what broke? With a very high degree of clarity, specificity and accuracy so that you can take immediate action. We do this one of the largest organizations on the planet that have billions of interconnected data points that we are analyzing and providing results and analytics against in real time. That deterministic AI, that foundation and what precisely is happening by way of analytics in your environment then sets up the foundation to be able to take agentic action against that set of analytics. And that's where the agents come into play. We believe that what the market will hear what customers will hear from basically every observability players. I've got agentic AI. I'm delivering agents, they can take action our supposition when we get into things like proofs of concept and others is that you have to start with reliable trustworthy input to those agents. Otherwise, those agents are going to be taking actions that are guesses. And so the result of that is we believe that Dynatrace Intelligence is really unique in the market space of delivering both what the answers are, what the analytics tell you along with the agents linked into that, that can take action. And by the way, those agents in that agentic framework are really an ecosystem of agents, not just Dynatrace agents. That can take action, for example, through hyperscalers, through ServiceNow, through Atlassian, through others to essentially enable auto prevention, auto remediation, auto optimization in your overall environment, which gets you back to where we started in your very first question, which is resilient software and reliable AI.

Yes. And I had -- my next question was around some of the things you announced around agents. But before I get to that, if we kind of zoom out and get at least from my point of view, why I'm bullish this category, we think about the attach rate of observability to agentic deployments. Do you feel like that that's going to be as high or even higher than sort of like the cloud-native application era. I mean, these agents are going to be accessing critical business systems. They're going to be calling external tools. They're going to be interfacing with your end users. Like is it sort of obvious that this is all going to be monitored, tracked, logged from your guys' perspective?

We have a theory, which I would say is early stage at the moment, but it speaks to precisely what you're suggesting. And that is, today, we look at the preponderance of workloads that happen across enterprises, the largest enterprises that we typically would sell to. Maybe 30% plus or minus of workloads and traditional workloads are observed by Dynatrace. Why is that? Well, if you need to observe your primary set of infrastructure, you need to observe your primary mobile app, your primary website, whatever those workloads may be. But I was speaking to a customer recently down in Australia, and their comment was we have 2,000 apps we have this much -- we're not going to -- we don't need -- they don't have the same level of criticality. We don't need to observe all those workloads. In the case of agentic AI, especially, and in an LLM environment that is producing probabilistic outcomes, we believe that you really are going to need to observe darn near 100% of those workloads because it is going to be sufficiently independently operating that you are going to have to do extra work to have systems that can give you the confidence that you're delivering resilient capabilities. You're also going to have the systems that are mission critical to delivering reliable AI outcomes. And so the result of it is, we believe that in this world that is evolving rapidly to an AI-first world, 2 things happen. Number one, explosion of workloads. So you just have more raw workloads. And then the second is your very point of the question, you need to be able to rely upon those outcomes in such a way that probably drives more observability as a penetration rate against those workloads.

Yes, it makes tons of sense. So let's talk about some of the agentic capabilities you introduced at Perform, your user conference. So you have an agent-first reliability engineers, you have an agents for development, you have agents for security teams. Which of these are you most excited about from a monetization perspective? And how are these domain-specific agents priced.

Sort of like asking what's your favorite kid. It's hard to say. I mean, I think that they are all critical. And the way to think about the agents is, first, you have the foundational agents. These are agents like an SRE agent that would tell you root cause analysis. For example, you need to know specifically what's happening. And then sitting on top of that, you have sort of management and supervisory agents that can direct traffic. So do the agents tell, a, Dynatrace agent to take action to resolve a particular incident maybe to turn off a feature, for example, that we can do internally. Or do you assign that to a third-party agent to like an AWS agent that may provision more storage or whatever it might be or a ServiceNow agent and might take some workflow action. So that's sort of the next layer. Third layer would be those agents that actually could take action to resolve issues and then you have a series of ecosystem agents that integrate. So it is sort of a very thoughtful stacked map that can define what agents are taking action. But I see that as -- while critical to overall architecture and the architectural topology, foundationally, the more critical piece, I think, for investors and others take away even customers with whom we speak is that start with a deterministic foundation, you then have the confidence to take agentic action, and we, Dynatrace, have produced an architecture through Dynatrace Intelligence that enables you to take that that agentic action thoughtfully either through Dynatrace agents or third-party agents to be able to deliver against those elements of essentially an auto correcting software ecosystem, which is ultimately what we all want to be able to deliver.

Awesome. Let's talk about in terms of sticking on the theme of product innovation. You launched your next-generation real user monitoring service powered by Grail, powered by Smartscape and Advanced AI in the context of that you've gotten to like $100 million consumption run rate with logs. The question here is like, given what the traction we see today with digital experience monitoring and now with this next-gen platform on RUM, how confident are you that this can be your next $500 million, $1 billion plus business?

Well, I will tell you that we already have $400 million-plus businesses. We obviously -- one of them, obviously, most recently at Logs our DEM business is well over $100 million in our Infrastructure Monitoring is actually the second fastest-growing business, interestingly enough next to Logs business. And then there's obviously full stack for APM. So the expectation is across all these categories and some of the things that Rick talked about is more workloads, more workloads across a broader stack, the biggest sales play that we've been able to drive has been end-to-end observability. These are customers that are looking to consolidate fragmented tools onto one platform. You see in our land sizes for new logos, you see it in the expansions that we're doing. And so these are going to be primary sources of growth now and in the future.

I want to have a discussion with both of you on Dynatrace's defensibility in the era of AI. But let's get an update first on just some of the trend lines of the business. I wanted to walk through this with you, Jim. In terms of the ARR performance, we've seen constant currency net new ARR stabilized at 16% for multiple quarters after years of deceleration. Can you talk us through the specific factors that contribute to stabilization? And do you feel like this is kind of the new baseline for growth as we look forward into fiscal year '26?

Yes. So I appreciate the question. So again, to your point, we've had 3 quarters of stabilized ARR growth at 16%. We've had 3 consecutive quarters of double-digit net new ARR growth, which obviously fuels ARR. Our guide for the fourth quarter, at the high end would suggest this continues. To your point, we haven't done this in several years. You say, well, what has caused that? Well, what has caused that was changes we said we made basically a little -- almost 2 years ago now, where we made go-to-market changes. And we were very clear when we made those go-to-market changes. These were changes to go on the offensive. So those changes were we oriented more resources around large enterprise accounts, what we call the Global 500, so the 500 largest companies or governments on the planet. We continue to fortify our partner ecosystem, in particular with the GSIs. And so everything we put in place 2 years ago, we knew year 1 was going to be a maturation year. We needed to get the resources staffed. We needed to get the alignment going. We changed some compensation plan designs. And so we knew year 1 was going to be a period of building. What you're seeing this year you're seeing execution consistency, which is exactly what we expected when we built the plan 2 years ago. So the maturation of the go-to-market model has continued to advance. I expect that will continue going into fiscal '27. And then you look -- even though we're not going to guide here, you look at fiscal '27, there's a lot of momentum in the business that we expect that if we can continue to execute like this you'll see it continue into next year.

To follow on that point, and this is an area of question that I get asked a lot about investors. The underlying consumption in the platform, you guys have purchase going north of 20% and outpacing the subscription revenue. So in terms of the lag between consumption growth and then that materializing ARR, what's the best way to think about those 2 dynamics?

Yes. I mean, I get that question a lot because one of the things we wanted to make sure we shared with investors is what's happening with the underlying growth in the business, which is how customers are consuming the platform. Obviously, our model is a subscription model. And so subscription model as revenue, is ratably recognized. That isn't always how consumption occurs. And so if ARR is growing 16%, and consumption is growing 20%, it will converge. The challenge is the timing of it. There's a lot of dynamics that go into when that will occur. We do look at something internally, we look at the consumption to ARR ratio. And the way to think about that is the headroom. What is the headroom a customer has before they need to do an expansion. And so for us, it's about continuing to drive more consumption. And one of the things that we've done is even though the model is a subscription model at its heart, the Dynatrace platform subscription model is a consumption model. Customers commit to a dollar amount, they can commit to a term. And then they consume. And it's a frictionless model, they can consume because they have the rate card for every capability on the platform. They're getting value, and we can work with them on driving more adoption of different product capabilities, they'll consume more. The more you consume, you can consume faster and it will burn through your commitment early and you'll do an expansion. And so I expect that we will continue to provide that as a metric, like how we're doing around driving consumption. We now have teams of people. They're measured on this. They're compensated on this. We have specific product strike teams for Logs measured on consumption. Product strike teams for DEM measured on consumption. Product strike teams for application security, measured on consumption. And then our CSM teams are also measured on consumption for the accounts that they support. They're not product specific. And so we've advanced a bunch of activities that are very consumption oriented. That wasn't the case 2-plus years ago. It was a very SKU-based model. Now it's a get them on the Dynatrace platform subscription and drive consumption.

That's a great context there. One of the things I've been saying all week, Rick, is that my conversations at the beginning with investors was sort of like, Sanjit congratulations, you cover infrastructure software, you cover data platforms, your companies don't have seat-based models, you're so lucky. And in the last couple of weeks, now everything sort of gets questioned, right, in terms of the defensibility. So I wanted to spend some time with you, Rick, talking about how Dynatrace is positioned in this era of AI and give you some of the scenarios that I get asked about and get your sort of perspective. One of the questions I've been getting is how does the value proposition of Dynatrace platform changes when the agents are doing the investigating and triaging versus human site reliability engineers or DevOps personnel working through dashboards?

On that topic, Sanjit, specifically, I would say that we expect exactly that evolution we expect over the coming years. And by the way, what the investment community is talking about at the moment, in most cases, is not what customers are talking about with us at the moment. There's a broad disconnect, I would say. But if we look out over the course of the coming years, we do expect that humans, end users, if you will, will become a relatively lower consumer of the Dynatrace platform and agents will become a relatively higher consumer of the Dynatrace platform. That does not, in any way, translate in my view to any disintermediation of observability by AI. Rather, we see observability as being mission-critical to these AI workloads where agents are taking action that have to result in reliable outcomes. And you're simply, in our view, not going to have a probabilistic system providing input to a probabilistic system on delivering an outcome that is trustworthy or the kinds of organizations with whom we do business. So it is that sort of notion that we believe that observability and in particular with some bias, Dynatrace, can and should become the control plane or reliable AI. And that really is based on an architectural moat of, as I mentioned earlier, Grail, Smartscape, Dynatrace Intelligence, the various technologies that we have built into the platform to deliver both deterministic AI and certainty of answers that can be trustworthy and that can be acted upon in a reliable way to deliver the AI outcomes of the future. Whether those outcomes are delivered by an end user or by an agent, in some sense, is not that impacting to our overall business model. In fact, to the extent agents are a bit chatty and they're going to consume more analytics than an end user would then, if anything, we believe that that's a tailwind to Dynatrace just as AI broadly as a tailwind to absorbability and Dynatrace by virtue of generating more workloads.

And one thing I'd add on that, Sanjit, is that we're not a seat-based model. And so because we're not a seat-based model or a consumption-based model, so we monetize through consumption. So get them on the platform again through the Dynatrace platform subscription. We don't even have to change our monetization model for the way we go to market with product packages. It's already in place.

Yes. I take advantage of those elements, yes. And correct me if I'm wrong, one of the marketing messages that you've had for years even before the [indiscernible] was answers -- like literally, answers not dashboards, right? So...

The answers are not guesses. That actually came from Rick.

Yes. answers, I guess, is one of them. And the other thing we would say is, you're right, the answer is not dashboards. It is it is really critical for our customer base to get precisely to answer, not try to ascertain what's happening in the environment through a dashboard and through alerts in that environment.

Awesome. Let's go through the second kind of flavor in terms of investors' concerns on the category more broadly, but also with specific to Dynatrace. So this other angle is that the ability to combine open-source tooling to collect metrics, chases and logs, and combine that with an agent, either for one of the model labs to reason over the data and execute an incident response. And so I guess what investors are getting at is the potential for customers to manage observability themselves at theoretically lower cost or even more nuance, negotiate better pricing when it comes to their Dynatrace renewals and bills. Why is this line of thinking off base?

My response to that is that, look, the primary deployment of observability throughout history of observability has been DIY. It is, relatively speaking, quite recent that companies like Dynatrace and observability companies have come into the fray. And so DIY continues to be feasible that you could use open-source tools, you can use, OTel, Open Telemetry, you can bring in these sorts of elements and you can manage it on your own. The fact of the matter is that is getting more and more difficult to do each and every day. Now might an LLM decide to do that for their own infrastructure, maybe, why? Because that is core to their business, delivering a resilient LLM that has reliable AI output, as you can imagine, for the LLMs is quite core. In the case, you've an enterprise, the largest banks, the largest health care organizations, the largest airlines, delivering a dynamic end-to-end observability solution that can process billions of interconnected data points contextually in real time. That is super complicated, and it takes this sort of broad-based platform architectural moat that we've described to have constructed that. That is not, in our view, a likelihood of an outcome, certainly for the vast majority of large enterprises.

If we call this category monitoring, and that goes back into the late -- mid- to late 1990s. One of the things about this category monitoring observability has been highly tied to changes in compute cycle. And the history is being that the leader in one cycle doesn't typically stay the leader in the next cycle. I actually think in this category, Dynatrace is one of the true success stories. You guys were a leader, multiple cycles to go when we were building on-prem job applications. You guys innovated, rewrote the platform from a clean sheet of paper, looking like aces today, but in terms of this broader AI debate, what are the ingredients of the business that allows Dynatrace to stay on top of the innovation frontier with potentially a new platform shift ahead of us.

Well, we've talked about some of it. I think the Dynatrace Intelligence is a core part of that. This notion of it's not just about deterministic AI and agentic AI. It is about using deterministic AI and agentic AI outcomes becomes particularly critical and doing so in real time. That dynamic in our view, and we've talked about the shift toward agents as consumers, if you will, observability of observability data. And in that environment, I would say that structure, that architectural context is even more critical to deliver. So that becomes sort of the next generation, even as AI-first sorts of models evolve, those models are going to evolve in a way in which observability foundationally becomes more critical. But you really do have to have the deterministic piece. And I'd say that is where Dynatrace differs from others in the market to be able to provide that underlying foundation for success.

I think the other thing just covering the space, in the context of investors debating with cost of code going to 0, that customers can now build anything, right? I think what you guys have been doing, what some of your peers have been doing. These aren't tools. These are distributed compute platforms that processes billions, trillions of data points in real time. It's not like we can go out and build this easily. This is pretty hardcore stuff.

So some of what you're suggesting, I think, is exactly right, which is -- it is -- at least in our view, it is much easier for an LLM as you vibe code something to rebuild something that has a standard workflow. Our workflow at any particular customer, at any given moment is highly dynamic, highly variable depending on what's happening at that moment in time based on a data plane and contextual data as input to that system that is inherently different than it was seconds ago, let alone minutes or hours ago. And that dynamic element means that the platform always needs to be learning. And that is a shift that doesn't result in, I'm producing a piece of code for a moment of time, and it takes that sort of domain expertise of the individual environment in the context of the overall platform and its generation that delivers meaningful value.

Yes. That's great context. So we've talked about the secular debate. Let's get down to the field level and talk about some of the things that are going on in the ground in the business. Starting with kind of market to market in terms of the go-to-market progress. We're about -- I think, Jim, you mentioned almost 2 years into the go-to-market changes. You stated that visibility and confidence is greater now than a year ago with pipeline also accelerating. Where are you seeing the most success and which elements are still maturing.

So I'd say it's playing out about as we expected. So again, when we outlined this almost 2 years ago, where we are in our journey is about what we expected. I'd say the #1 sales play, I think I may have mentioned it earlier, is end-to-end observability. So we have 3 sales plays. We have end-to-end observability. We have an APM land play where you land, you do a POC and you expand from there. And then we kind of have a cloud-native play as well. I would say, universally, the most successful sales play has been end-to-end observability. That we have a sales organization that knows how to sell it. The value proposition is very clear. We can actually allow them to save money and they consolidate tools and they can get a better outcome. So I'd say we're still -- even though I talked about this 2 years ago, Sanjit, that this was an emerging trend. This is a prevalent trend now. This is more and more enterprises are looking to consolidate fragmented tools onto one platform, not unique to our particular industry. You're seeing it in security in other places as well. So I'd say it has been a source of growth, and I think it will be a continued source of growth because it's still many companies are not doing this. I'd say the area that we're making great traction, and we're very proud of the fact that we hit our $100 million milestone for Logs. But that's just the tip of the iceberg. So to your point, we have a lot more growth to be had within Logs. 40% of our customers leverage our Log solution, continue to grow the cohort classes. And so they start on the platform smaller, and have grown significantly. So between end-to-end observability, Logs in this growing use cases, that you will see this also with AI-native workloads that these are all areas that will continue to mature. And the good news is the go-to-market motion is we're year 2 which is where we thought the productivity improvements would begin. We expect those productivity improvements to even accelerate further in fiscal '27.

I mean, to your point on like the consolidation buying behavior, there's -- if I just kind of look at this market, it's been crowded. There's a lot of players. You can probably name a dozen different players. But there's probably not that many they can pull off the consolidation deal, right? And so when we think to that and then I look at the success you've had in winning large deals on the pipeline being constituted with deals over $500 million, kind of speaks to your ability to really win those consolidation opportunities. When we think about managing the timing of variability inherent with these large enterprise deals and while maintaining guidance accuracy, what rate of success are you seeing converting the pipeline needed to see, not only your goals for fiscal year '27, but just making sure that you get more confidence into your guide?

So I would say 2 years ago that we were growing pipeline or closing pipeline 2-plus years ago, with an or. And so I'd say the consistency around what we drove for net new ARR quarter-to-quarter varied a bit. I'd say what we have now is we are growing pipeline and closing pipeline at the same time. So pipeline growth is very strong. I'd say the quality of the pipeline is also very strong. And we measure that by just inspecting the pipeline and looking at kind of where we're at from a sales stage perspective. So -- and I think that comes from the go-to-market changes that we made. Again, we may go-to-market changes to get closer to customers that one of the big changes we made around the enterprise accounts was to go from maybe 10 accounts per rep to say, 4 to 5, which means they're a lot closer to the specifics of their customer base, which means when you're looking at pipeline and deal flow, there's a lot more intimacy around what exactly is happening with that brings confidence now around closing. Now your timing might vary a little bit, but I would say even though we continue to see end-to-end deals keep growing in number and growing in size. I'd say our confidence in our ability to not have to land all of them. Some are going to maybe close one quarter, some will close the next quarter. I'd say we have building confidence that the consistency in the go-to-market execution just continues to advance, and I expect it will continue to.

That's definitely encouraging. I think from my seat, when I think about the last 12 to 18 months. It felt like you guys would execute on quarters, knock down some of those bigger deals, we'd see good ARR results, but because the pipeline was weighted toward those deals. We had to be more conservative on the forward quarter. Is that also part of the solution maybe getting a higher velocity, more transactional blocking and tackling business that generates some of that ARR and that revenue as you guys go and penetrate your larger enterprise customers? Maybe Rick, you can speak to the potential to build a higher velocity sales motion, maybe it's mid-market or upper mid-market, just your thoughts on that.

I actually think, Sanjit, you're going to see in a cloud-native-AI-first world is you're going to see not a pivot, not a transition, but an evolution toward more departmental selling, in particular, the development audiences associated with cloud-native deployments. So I wouldn't walk away from the session thinking, wow, Dynatrace is going to go to SMB because our value proposition really wins most often in the larger enterprise. But it doesn't always have to win in centralized IT. I think it is just as pervasive, just as impactful in smaller groups within that larger enterprise, which then get aggregated. So I think that is some of the transactional volume that we can imagine evolving into the future.

Awesome. Let's talk a little bit about capital allocation. So you announced a new $1 billion share repurchase program. Can you maybe give us why re-up for $1 billion? What's the signal that you're looking to send to the market?

So I'll take that. So we were -- I think it was probably -- I think it was May of '24 that we put the $500 million authorization in place for the first -- in the first go. We tripled what we spent on our buyback in the third quarter because we actually thought it was a significant value dislocation for the company based on what we believe the prospect of the company were and what we were valued at. And so we exhausted it. We doubled the authorization. Again, going back to our philosophy. Our philosophy is, one, invest in the business. So that is the first use of capital. But because we have $1 billion of cash, and we generate $500 million of free cash flow, we're in an advantageous position where we can both return capital through a buyback and opportunistically grow and look for opportunities from an M&A perspective. And so you can expect that we will be buyers at current prices, probably at an increasing level to even more so than what we did in Q3. So that's what we're going to do. I'd say from an M&A perspective, we are active shoppers, but disciplined buyers. And there are things that need to kind of fortify the platform and fortify what we think is an opportunity to kind of grow and broaden the observability use cases.

Maybe last question to wrap up. In terms of stock-based compensation, how you're managing that, what level of share dilution should investors expect? And in terms of getting is more GAAP profitable? How would you sort of rank those priorities?

Well, we got a unique company that is GAAP profitable. So we're not becoming GAAP profitable. We are GAAP profitable, which is unique in kind of the software space. But relative to stock-based compensation, we've always been a very appropriate user of stock-based compensation. This year will be around 15% or 16% of revenue. And my expectation is that we'll probably -- we'll drive some more scale from that going forward. And so while others are trying to catch up relative to the profitability, look at the profile of the company, the company has almost 30% operating margins. Because we're a cash taxpayer, where most tech companies are not, we generate, on an equivalent basis, call it, 32% free cash flows on a pretax basis. So it's been a focus on the things that we're doing have all been about how do we accelerate growth in the business, accelerate growth in the business. You'll find that it's -- again, you will actually drive more leverage when you accelerate growth in the business. You'll actually get it at the same time. And so we're quite optimistic about the opportunities ahead for the company, where we're at, where we're positioned, both on the go-to-market side, where we're positioned on the product side and the platform side. And so I'd say this is an exciting time for Dynatrace because I think a lot of what we've put in place is actually kind of something that we're going to see this momentum continue into fiscal '27.

With that, we're out of time, Rick, Jim, thank you for giving us the update on Dynatrace. Best of luck going in Q4, and thank you for coming to the TMT conference.

Thanks, Sanjit. Thank you all.