I told you, it's hard to lean all the way back.
I was told that. Just the big instruction is.
For you, don't lean back, right?
Nobody's leaning back, yeah.
Okay, great. David, wonderful to have you back to the UBS Tech Conference in Scottsdale. It was nice to see you out on the course. David played well, as I did, miraculously. So we had fun outside of the venue here.
Thanks for having us. It was a beautiful day yesterday.
Gorgeous, wasn't it? And I hope you guys had a little fun playing with Sergio, for those that did. So David, maybe we'll start. There were a number of things about your recent quarter that were really strong. Your core did better. Your AI native breadth widened out a little bit. You renewed that big contract. So a lot of good things to talk about. So let's take those in order. So I think one of the things that stood out to me was the strength in the core. So X-ing out all the AI natives, the core accelerated. Can you add a little bit more context as to where that came from? What went well in the quarter to result in that?
That's a great question, and thanks for that intro. Yeah, I think we saw a very broad-based, strong quarter. We were very gratified in that we've been expanding our sales capacity. We lagged a little bit on the back end, or maybe we're a little conservative on the back end of the bubble bursting, and then have spent the last year and a half plus scng, and we found that produced returns. And that came in the form of more new logos and larger new logos. And we also saw metrics like the maintenance of attainment, the maintenance of CAC return, and things like that. So I'd say the first thing is we were able to execute and go to market. The second thing was we have had really good success in the platform and in attaching products.
So that helped produce, excluding the AI, that helped prodeuce a stronger net retention. We also saw some good distribution. We had had, I think, a little bit of a pushback in the SMB side of the business. 1,000 employees or under, and these are not tiny companies, but they have under 1,000 employees. And we saw rebound there. We saw strength in SMB at the same time we maintained strength in enterprise. And then lastly, I think we had a constructive buying environment. It's not a bull market, but it's not a market that is obsessed with cost and is pushing back on optimization. So the headwinds that we had seen in other parts of the earlier in the market kind of abated, and all of that came together to produce a pretty broad-based quarter of strength.
David, on that last point around facing a slightly more constructive buying environment versus prior quarters this year, what do you think changed? Because just as you're saying that, and I'm reflecting on all of the software company reports that we've had to digest, not everybody has called out a more constructive environment. So could you pinpoint what you feel like turned, and maybe it's a little bit unique to your category of software?
Yeah, I think infrastructure. So I think we basically were based on the replatforming of software, modern software, and then putting that in the cloud. And I think you have to go through some infrastructure to pull that off. So I think a combination of the return or the strengthening of the modernization of software stacks, it may be, and this is all outside the AI native we're talking about right now.
Yeah, exactly.
It might be also. We've seen in other technological evolution periods, we've seen an acceleration of modernization of software stacks, and we may be seeing some of that. So I think that's probably why the infrastructure companies and the foundational companies .
Correct, that's where we're seeing it. That's where we're seeing it, yeah.
And if part of the tailwind that helped you was companies modernizing partly by migrating more workloads off-prem to the cloud, does that dynamic feel like it's a continued tailwind, David?
Yeah, we think that's going to be a long-term tailwind. And so when you look at the percentage of workloads that are in the cloud, which Gartner and others say is in their 20% or 30%, and when you see major parts of the economy in large enterprises, in regions, and in governments not having modernized, we think that this will continue for a long time. So it has a lot of sustainability to it, which is a friend of Datadog. And the signs we've seen already in the year to date, as well as we said that continued in October, we've seen a good investment cycle. So we're optimistic. We can't predict the future, but we're optimistic.
What about strength across the product suite? Datadog obviously offers a complete platform.
Yep.
But you can think about it partly as core infrastructure monitoring, APM, log management. Was that inflection you're referring to in the third quarter unique to any part of that suite?
No, I think it was pretty parallel. For the most part, our customers are using the platform in a more consolidated way. They're moving from point solutions and consolidating on the platform. So I think you have a similar type of demand development, which is in the metrics, traces, logs, the core three pillars, relatively parallel strength. And then you have, I think we announced that our digital experience crossed $300 million. So that's the next group of products that continue to be adopted. We also have, we can get into this, contribution, although smaller, from such products as Cloud SIEM in the security family, from product analytics, and from service management on call that enhance that.
Okay. That's a good discussion of the core. How about we talk about the second positive thing, which is the AI native cohort collectively was 12% of your revenues. And as you and Olivier pointed out, it's well beyond that one customer. So much more breadth.
Yeah.
So I guess my first question on this front, David, is how you're pulling that off. Because it sounds easy to be selling your products to this fast-growing cohort, but it's not easy for everybody. For those of you that were on the call with MongoDB last night, Dave, who you know well, Dave was admitting, and he's been quite candid about this, that Mongo hasn't penetrated those AI natives as well as they would like. And so it sounds like they're trying to take a page out of the Datadog playbook to go after them. So what have you done right such that your AI native penetration is maybe larger than almost any software company we follow?
Yeah, so I think when you think of where Datadog has won a lot, it's been in what we used to call cloud native.
That's right.
But now we're talking about a group of those companies called AI native.
Correct.
But Datadog monitors client-facing mission-critical cloud applications. And these AI-native companies are cloud-native companies, meaning they don't have legacy stacks. Their whole product is delivered digitally through the cloud. They're a modern software company. So it's a really good fit when you think of how Datadog architected its platform. It's always been a really good fit. And then how it's frictionlessly adopted, how it grows with the client. I think it's even probably a better fit now than it was in the cloud-native period of intense growth because of the breadth of the platform. So these companies that have a lot to do and they have a lot of R&D investments are finding they're able pervasively in their observability needs to use Datadog. Now, in addition, they're experiencing great success. Many of them are releasing their revenues. They're experiencing a very significant demand cycle.
Our platform has been architected, and I think this is one of the things that might be different than some of the other companies, so that it is very easy to use. You sort of set it up, and the clients can use it frictionlessly. So as their workloads increase, our revenues increase. And because it's so easy to implement for these modern companies at the kind of platform they use. So I think it's a really good product-market fit. And I think we mentioned, as you said, that this is not just about a concentrated set of customers. It's over 500 customers. It's over 100 customers that are spending more than $100,000, over 15 that are spending $1 million. And it's most of the leading customers.
We have a set of AI/cloud natives that are pervasively adopting Datadog, and we're succeeding through the way we've always kind of go-to-market and sold our product.
Are they using Datadog in any different way than a UBS or other conventional large company would across your suite? Are they quite concentrated in one particular area, or are the workload types the same as a typical large enterprise?
The workload types are very similar. They're using the metrics, traces, and logs. Essentially, once we're not talking about model training here. We're talking about production, and so when you get to the production environment and delivering the software, it's similar to other, what we'd call, cloud natives. So most of them are using the metrics, traces, and logs and then the digital experience. Some using the service management, some using the security, but very similar usage patterns to what we see in the rest of our customer base.
What about, David, to continue this thread around the analogy to traditional enterprises, what about the contract structure, the duration, renewal cycle? Does that also feel similar, or are there some differences that are noteworthy to call out to the audience?
In cloud natives, I think you generally have what has tended to be an annual commitment. It's based on what you know about the capacity planning. And then as clients grow, in both cloud natives and AI natives, you'll find that as they're growing their business, they are going past their initial potentially conservative commit. And because we price based on volume and on term, and we didn't invent this the same way the cloud providers, the hyperscalers do, what the clients find is that as they get to certain levels of usage, they can get a better deal. They can sort of get a better price point by committing. And so similar to others, they have gone through this cycle of then committing longer and at higher volumes. And that's the same thing that happened with the largest customer.
That's the same thing that happens with other rapidly growing, I think, contrast to enterprise. Enterprises may both have more predictability and more control, so they might be willing to commit longer, three-year contracts. But cloud natives generally have committed for the most part to around annual contracts, maybe some more. And another thing that is similar is we've helped them use the product. So for instance, if they are not optimizing or using too much of one or the other, we help them. And because we have all these products and they make a commit, they can still satisfy that commit even though they're controlling certain products more.
Got it. And if we take a subset of those AI natives and we talk for a quick moment about model providers, I'm just curious if their needs are a little bit different because we're all watching the compute commitment that a lot of those model providers are making. I would assume, David, that it's correct that to the extent that any of them scale up massive training compute clusters, there's less pull-through there for Datadog than when they scale up inference compute footprint. Is that correct? Is there any real pull-through on the training side that's noteworthy yet?
Most of them are doing their own analysis of their own training. There hasn't been.
They're using their own tools for that.
There is no tools. That's like their core product, right? So we're production. We're monitoring your delivery of your models. So I think when you're reading a lot about, and it's confusing, when you're reading a lot about the hyperscalers, the NeoClouds, all this GPU, the vast majority of the use cases is in model training. And for better or worse, Datadog is not monetizing that. Datadog's monetizing the production environment. And so that's why the use cases, that's why the types of products are very similar to the other applications that are in production.
Yeah. David, there was a lot of angst in the investor base prior to this renewal, the large one, about whether they would, how much. Looking back at that period of uncertainty, what do you think maybe the street didn't appreciate about the Datadog relationship such that you were able to renew and actually expand?
Yeah, I think there's a couple of things. First of all, we've tried to tell everybody that our gross retentions in large customers are in the very high 90s, really, so 98 or plus. So we basically have tried to tell everybody that it's really a fringe case when a large customer leaves Datadog.
Yeah
Yes, they may change their spend a little bit or whatever. So we were trying to tell everybody that if you look at the history, that there are customers that in-source, but it is not a good economic decision, and it doesn't happen very often.
Yeah. I got the message, David.
Okay, so that's one thing. Then, okay, once you see that, you see, okay, that may happen from time to time, so Olivier went out of his way to say, but there are some customers that really never become big Datadog customers. He says Google, Meta, et cetera, and so we said, with this one, everyone's thinking about, we don't know, but they're evidencing that they're making the decision to be with Datadog, and then, of course, they decided to re-up the contract and extend the commit, and that's evidence of what you see, which is it's not a good decision in these kinds of real-time use to do it yourself. It's very expensive. You generally want to put your R&D efforts on your product, and most companies decide not to do it. I mean, we've shown many, many companies that it's economic, so we tried to tell everybody that.
So then it happened, and we've been saying things like that, and I guess we proved it.
David, is it also the case for some of these large AI natives or cloud native companies to use multiple observability platforms as well? I think you were also messaging.
Definitely.
Calm down. There's probably room for two or three.
Yeah. So in large enterprise, in large companies, based on use case, there might be multiple observability or pieces of it. And yes, in this situation, I think there's been a lot of noise in the market that other solutions are being used. But that's fine. I mean, they may use that for this purpose and that for that purpose. So I think with everybody saying, all of these, some of these vendors saying that they have that customer, what is being shown by extending that is, yes, we can coexist. Because most IT departments and most reliability engineering companies decide to do it at a level that's below what we're all thinking about. Like, what are you doing with the logs? Are you storing them long-term? What logs? Is it IT logs? Is it production logs?
So you may have log stores that you're using in different terms or different purposes that may have nothing to do with Datadog, but they can say they're a customer as well. So that's what happens.
So this is a nice segue to talk a little bit about the competitive environment. Datadog's always had strong rivals, and obviously, you've powered through to the point where you are now. But let's talk about a couple in particular because sentiment was impacted a couple of weeks back when one of them, Chronosphere, was acquired by Palo Alto Networks. So David, what's your reaction to that deal? Obviously, the street evidently was a little bit worried, but where was perhaps that worry misplaced?
Yeah. Yeah. Well, we've had this discussion about Chronosphere for three or four years. So Chronosphere, as their basic product, is a large-scale metric store. In terms of if you define an observability platform at what Datadog has, they don't have that. So they have a piece of it. And we've had this same discussion when they were out doing some fundraising, et cetera. They were going, we've taken this business. And it didn't amount to much, right? So they didn't. And I guess we're hearing they have some of the same business that Datadog has. But yes, that's fine. So basically, for long-term metric store, you may want to have that. We also have a long-term metric store. And we've been investing relentlessly in managing the price points for what you're going to do with it, whether it's real-time, long-term storage without limits, frozen, not.
All these terms are about slicing and dicing metrics and logs, so we have that too, but customers, like you said, may want to have a certain amount of metrics stored in a different way, and that's fine. We see that in some customers, so I think when you're talking about observability platform and what Datadog's business is and the kind of what's been happening with point solutions over time, you see that the balance of trade has been heading towards Datadog, and when you actually look at the size of some of these companies, I mean, we don't work there, but I think we essentially are able to put, in terms of ARR, the same amount of ARR they produce in their lifetime on Datadog somewhere between a month and six months, depending upon what the company is, so you see what actually is happening in the market.
David, the other category besides Chronosphere is some open-source alternatives. And one or two have made noise all year and have carved out niches in things like real-time log analytics. So what's your thought on that, and to what extent does Datadog have or will soon have comparable future functionality?
Yeah. What Datadog does, so when you think about logs, real-time logs, the various storage periods, frozen logs, flex logs, Datadog has successfully gotten ahead of this by slicing and dicing the platform with good margins to be able to have that business and also capture net market share in logs that may not have the same requirements as real-time. So I think ClickHouse is one that's mentioned. And I think Datadog has a very large and growing and successful log business in observability logs, increasingly non-observability logs. So I think we've done many, many of the things in the platform. That doesn't mean that we're going to get 100% of the business, but we've done really well in this area.
Can we talk about pricing for a moment? Obviously, when you're selling a software that to some extent is correlated to the client's infrastructure spend, as they scale that, their bill can get high. And it's, I think, incumbent upon software companies like yours to find ways to be flexible such that that doesn't become a point of friction. Can you talk a little bit about that journey to arrive at a place where that's less and less a point of friction?
Yeah, definitely. So we price based on volume. So the unit price goes down. Now, some of you may be wondering, okay, does that mean your margins are compressed or your unit pricing is lower? It isn't because actually we have smaller customers coming in. So the weighted average does not change substantially. We also have services. This is one of the areas where we've gotten better. We have good transparency of usage. We actually have SKUs and also services that help clients optimize and use the product. The fact that we sell credits or commitments and they can use a variety of products means this is something that I think is much more valuable than it was when the bubble burst, that clients can see where they're using and they can actually move to other products if they're overusing.
We also have always helped our clients, meaning if they have an accident or if they are misusing it and they go over, we actually don't charge it for them. And then we fix it. And then if they continue to use it, we do. So I think we're doing a lot of work to try to help clients. In addition, I think we're expanding, as I mentioned, the value of the platform so that clients are able to spend with a straight face more money with us from our functionality. In addition, I think we've gotten better about saving them money by consolidating point solutions. That includes proving to them that they're better off from dollars, not to have 10 things, but one. And we're getting better at migration credits to help them through that process.
Got it. You talked a little bit earlier, David, about some of the product suite extensions to drive further revenue growth. Which couple stand out as getting closer to being needle-moving? You've got many that are early stage like LLM observability. You mentioned a minute ago Cloud SIEM. Are there one or two worth highlighting that next year and 2027 could move the needle for you?
Definitely, I think the three pillars plus the DEM are really good. And in logs, and these are related, the fact that we have observability pipelines that can get access to logs that are not observability, that's interesting. Now, that has enabled one of the big potential opportunities, which is Cloud SIEM. So we're starting to be successful within large enterprises of extending or displacing their SIEM for cloud workloads. That is, we also have a large competitor, Splunk, has been acquired. Maybe they did aggregate pricing. And so there's a lot of synergy between our log business and our Cloud SIEM. So that's an opportunity. Another opportunity that we're seeing is service management. Today, we went GA on our Bits SRE.
Congratulations.
It's now posted. Even though we have a lot of customers using it and we have some revenues, we hadn't gone GA. So we are $500 per 20 incidents. So that's an opportunity compared with the other parts, complemented by the other parts, on-call, et cetera. So that's a nice opportunity. And then I think we have made some good acquisitions in product analytics, which was Eppo, and in data monitoring, which was Metaplane, which are small but are significant growth areas in the future potentially when we look at the market and point solutions that are out there.
Maybe I'll close and we'll leave a couple of minutes for questions with one on margins. David, I know you've got mid-20s long-term margin guidance. You're basically almost there. So when you're thinking about the next couple of years, are there puts and takes that you would remind us to keep in mind as we model out margins?
Yeah, it's a good question. So we've been above that 25 for a period, and we said we need to invest more, so I think what we're trying to do as a company is, while staying within this band, you can look at our guidance, our long-term, we're leaning in towards investments that are going to maintain as high as possible long-term growth rate and compound our revenue, so that's what we're trying to do, but we're doing it in an analytical way. We're looking at, in product, are we getting good return? Meaning, are we getting revenues in the $50 million or more? Are we allocating our R&D resources? Are we working on our own optimization and our own platform? And in the go-to-market, are we able to expand the go-to-market yet maintain the CAC returns that are really strong and the attainment?
Right now, we think there's lots of opportunities. We've been able to do both. We've been able to invest as well as do this margin delivery, and we think that becomes more and more powerful as you get bigger and bigger because you can invest a lot of dollars in R&D, we're by far the leader, and still respect the margins and still distance yourself, so we like our position in that way.
Okay. Good. We've got a couple of minutes for questions. If anybody in the audience would like to raise their hand. No? David, any final words about things that are exciting for you as you approach 2026 besides getting your goal score down?
Oh, I'd rather definitely have to continue to being able to partner regularly. I think that we're doing our planning now, and I think that Ali's always been someone who sort of looked ahead in what you have to do to create the distance and maintain the distance. So when I think about things like Bits SRE, when I think about what we're doing, having all the integrations with the AI providers and the kinds of information we're delivering about the number of those that are sending the data. When I think about the cohort here as well as the opportunities within Datadog, I think the winners will be the ones that realize the most out of AI in their product, and so that you won't have companies come in and be point solutions that offer something. You'll have that already.
Yeah
And I think he's done a really good job thinking about that.
Yeah
And so watch out for some of the metrics that we deliver in some of these AI areas as a way of, are we increasing market share? Are we winning because of it? Are we getting additional SKUs, additional workloads? And all of those can be ways that we monetize this opportunity.
Yeah. That's an exciting story. David , thanks so much for coming to our event. Thanks, everybody.
Thanks.
Yep.
Take care. Thank you.