Okay, welcome, everyone. I'm Mark Murphy, software analyst with JP Morgan. It is a great pleasure to be here with Olivier Pomel, who is the CEO and co-founder of Datadog. So first off, Olivier, thank you so much for being with us here today.
Thank you for having me here.
So most investors in this audience are pretty familiar with Datadog. You're approaching a $2.5 billion revenue run rate. For those in the audience who might not be familiar, can you describe Datadog? Can you explain the types of problems you solve? And, you know, how is it that you've emerged from this crowded field to become the leader in observability?
Yes. So what we do is we do observability and security for cloud environments. We basically sell to engineering teams and companies, big and small. Anybody from the tiniest teams of, you know, students or even two individuals who don't pay us anything, all the way up to the largest banks and telcos and software companies in the world that use our software. We started the company not really as a better mousetrap, so the idea was not to look at, you know, IBM Tivoli or BMC or all these things and build a better one. The starting point was to get teams that didn't speak to each other to actually share the same reality, work together, and solve the same problem together. So initially, it was between development and operations teams.
Today, it includes many other teams, including the security teams, for example. But really, that's what led us to start with not just a very narrow focus on the product category, but with a platform approach and trying to bring as many datasets and as many teams into the product as possible, make it as approachable as possible on the way. So the story of the company, since founding, has been one of continuously investing back into R&D, and we can talk some more about the efficient model that lets us do that, and expand the footprint we can have by covering more of those different use cases, bringing more datasets and more teams into our platform.
So Datadog's revenue growth actually accelerated in Q1, right? We saw 27%. It also accelerated for Amazon AWS, it accelerated for Azure, but there wasn't much of that in this Q1 earnings season. We think back to last November, you had actually called out a resumption in activity among the cloud natives, and in the recent Q1, you actually called out an acceleration in activity from your larger enterprise customers. So, Olivier, since you do have this unique kind of window into the broader cloud economy, why do you think it is that we saw this trend of, you, let's say, the past two or three quarters, where there's a resumption in activity? Do you think it's gonna continue or, you know, could it continue?
Could there be further acceleration, on both those sides, the cloud natives and enterprise?
Yes. So in general, I think what we're seeing right now is we see today is quite a bit better than what we saw one year ago. So one year ago, everything was slowing down, and in particular, the cloud native part of our customers, so the folks that were all in on the cloud, wall to wall already, and were spending a significant fraction of their IT budget, but also of their overall budget for overall OpEx into cloud, those were slowing down quite a bit. I think today we're in a position where the world is different. We're seeing stabilization or growth again from the most of the customer base, which is much better. If you drill down a little bit further, what you see is that on the...
If you disaggregate the cloud natives into two buckets, the AI native folks are growing gangbusters right now, and I'm sure you got some exposure to that, and so they're all growing very fast. The other, I would say, we call them a little bit older, web native or mobile native, are growing, but more cautiously. And then if you move up market to large enterprises, large enterprises are accelerating growth. They were still growing last year. Last year, when everybody was slowing down, the enterprise slowed, slowed down a little bit, but not all that much, because the enterprise is a lot earlier in its journey to the cloud.
And as a result, the cloud spend is a smaller fraction of the IT spend, and the IT spend in general also is a much smaller fraction of their total revenue than it is for the cloud natives. So we saw some re-acceleration there, though broadly in line, you know, with what we had seen before. The result is what you had called out earlier, which is a small but perfectly formed re-acceleration for us. And in general, I would say things are looking good, like the world is starting to grow again, but it's a, I would say it's a stabilization and a smaller acceleration, not the massive acceleration of all the cloud cons... consumption everywhere across the spectrum. And yeah, that's about it.
So better than a year ago when you were here with us at the Boston TMC party, as we like to call it. You met. You brought up GenAI. GenAI is explosive. We'll come back to that in a moment.
Yeah.
But it's not a V-shaped recovery.
Yeah. And in general, for the growth of the cloud, I would say, looking at investor expectations, leading into our quarter, I would say, in general, there's higher expectation of acceleration in the short term, but there's an underestimation of what the growth is going to be in the mid- to long-term, which is, you know, again, what explains some of the things you might have seen in our numbers, such as the fact that we are accelerating hiring and accelerating the investment pace.
We'll come back to that in a moment. Before we do, I wanted to think back to your recent Analyst Day, which was just a couple of months ago in February, and there was this comment that got our attention. You said, "We think observability alone holds the next 5x or 10x scale-up for us."... Given that Datadog is at a $2.5 billion run rate, right, it's a pretty fascinating comment. So we wanted to pull on that thread a little bit. What do you think, Olivier, has to happen for Datadog to scale like that, where you would become a $10 billion or $20 billion revenue situation in observability?
Yes. So if you look at the big categories in a top-down fashion, so the main one we play in is IT operations management, so ITOM. It's a market, so we're the leaders right now in that part. We have only 9%-10% of that market today. So there's a lot more room to be had there. That market, you know, ITOM itself is growing. I think, you know, the estimations are 10%-11% CAGR over the next few years. If you look at the next closest market, which is cloud services, is growing 20% over the next few years.
So, you know, it doesn't take a lot of imagination to project our sales, you know, five years into the future, see a total market that's twice the size than it is today. And us, having been a share taker, consistently, over the past decade, and being a share taker again in the future, it doesn't take a lot of imagination to see how we can get to 5x-10x the size we are today. Obviously, there's a lot that goes into that. There's a lot of work that is required, a lot of investment. That's what we're doing, but we think it's there.
Okay. Doesn't require a lot of imagination. I like that. So now along this path, Olivier, of a 5-10x in the business, do you think we should subscribe to the view that Datadog would be able to kind of maintain revenue growth at a higher level than when we look across the hyperscalers? So if you look across AWS, Azure, GCP, et cetera, on the basis that I believe Datadog is something like 1%-
Mm-hmm.
Today of all that spend that goes into that stack, is there a scenario where this is going to rise? And we're going to look back and say, "Well, now it's 1.5%, now it's 2%, and beyond," because the workloads themselves are becoming a lot more complex. So I wanted to ask you, what do you think about the growth algorithm? Because I think, you know, we have felt it is a bit of a fallacy to just try to look at Datadog and say, "It'll grow at exactly twice the rate of AWS every single quarter." So something about that seems a little too simplistic, and you're exposed much more broadly.
Yeah, and I would say this is the kind of thing that works well in the mid to long term, not so well in the short term, when you try to line up the numbers. One thing to keep in mind is we're not a one-to-one to the cloud providers in terms of, you know, how our numbers work. We have other products that don't relate directly to what they say in the numbers. They have other products. It's especially the case for, you know, let's call it Google and Azure have a lot of other things in there, but it's also the case for AWS. We don't recognize revenue the same way necessarily as they do, and we don't have the same exact customer or geographical mix. So the numbers are really hard to reconcile on a quarter-to-quarter basis.
That being said, we're exposed to the same trends as they are. I mean, the growth of our business is driven by cloud migration and digital transformation, and now, the curve of AI adoption. And I think this is what propels those three companies, and this is, you know, what's going to propel us as well. If you think of the ways, our share of the overall cloud market is going to increase, there's a couple of ways. The first one is that for the products we have today, we're still largely underpenetrated with our existing customers. So our customers largely don't use all of our products yet, and even when they use several of our products, there's a lot of margin for growth, you know, within those products.
So when customers adopt our APM, for example, it's going to take them years to deploy that across all of their workloads, all of their teams, everywhere they possibly can, which is why those products still have a lot of a very long runway. We still have also a lot of customers to acquire. So we're not present at the same scale as the big cloud providers in all the geographies and all the segments. You know, a good example of that will be government, where we are very, very small compared to the scale of cloud there, and there are many, many other examples across the world of segments and geographies we don't cover very well.
And even in the geographies we cover today, we have less customers than the cloud providers, so there's a lot of upside to be had there. The last way for us to increase this year is that we're still expanding the product footprint. So there are still many categories that relate to running, securing, understanding, shipping applications in cloud environments that still many of these categories we don't cover yet, and we intend to cover in the future. So you should see us provide more products and have more of a even longer runway for our customers to expand into our platform. So when you combine all of those, basically, that this is what you get from the 1% we are today to 2% later and maybe more.
When a customer is fully penetrated with Datadog today, the share of their cloud spend we represent might depend on the type of customers and what they do. But it's fair to say that when they use everything we have today, we're somewhere between 5% and 15% of their cloud bill. So as we grow and as we further penetrate the market, we can see how we can go, get up in, on that curve.
So you mentioned, you mentioned the evolution of the platform, and, you know, we have always put Datadog on a pedestal because, you know, we really hold these organic, homegrown, cohesive platforms in a very, very high regard. We think that that is a different classification of a software company, right? Especially when you compare it to the ones that are rolling up and, and, and bolting on. And it feels to us like this is the kind of platform that is gonna continually expand and, and really become a center of gravity. Now, I think back, Olivier, to the very recent earnings call.
Mm-hmm.
You used this word consolidation-
Mm-hmm.
many times. You mentioned a slew of these consolidation wins, where a company will take a whole bunch of, of point products, right? And they'll consolidate it all onto this Datadog single pane of glass. Are we reaching some kind of a tipping point, here today where, you know, you have this fragmented market out there, and it starts consolidating on Datadog?
Well, I mean, that's definitely the plan, is to make sure that customers consolidate on us. That's why we expand. We cover so many of the different categories they might need to cover, and we turn customers from, you know, being system integrators on their own to having a unified view of the world across teams and across data sets. So that's the strategy, you know, so to speak, there. We're starting to see tipping points. The tipping points have more to do with how far along individual customers are in their cloud journey. So basically, as the cloud side of their business represent more and more or enough of their overall business and of their overall mind share, they're going to consolidate a lot of their existing stuff on us as well.
And so we mentioned a lot of those on our earnings calls. We see, we expect to see a lot more of that moving forward. We've seen some tipping points already, one of which being the fact that last year, customers were a lot more cost-conscious, so taking out existing products or existing tools when they were adopting more was more important to them. But we expect to see, to have more tipping points in the future as we cover more and as they get further into their migration.
Okay. Looking forward to more tipping points. So, okay, let me explore one other possible tipping point.
Mm-hmm.
Splunk and New Relic were acquired both in the last year. That's a real change to the competitive landscape in a span of about nine months. Can you comment on that to some extent? I think we're curious if that's opening up any more of these consolidation opportunities a little faster.
Yeah. So one thing I will say is that, in general, we tend to obsess over customers and not over competitors, so we don't overfocus on whatever happens to them. I will say, though, that it is true, there's going to be opportunities just because those two companies have been acquired. And so New Relic is a relatively small business, but Splunk is a sizable business. We think there's definitely opportunities there, and I know that internally, we're accelerating some work on the product side that might be helpful for that. So we're—it's not lost on us, and we think there's some things we can do to help.
Olivier, one of my favorite quotes came from a major partner of yours that we were speaking with. And I had asked this person, I said, "What is the marketing message, you know, that is coming out of the Datadog sales teams?" And he thought about it, and he said, "Datadog lets the product do the talking." And we don't hear that too often. So, it feels like it's not—it's not marketing propaganda. It is a product that you have built that is demonstrating mastery in this domain of observability and beyond. Can you kind of speak to that engineering ethos? You know, what is it you've done that's allowing the product to speak for itself?
Yeah. I don't think that's necessarily just engineering ethos. That's about, you know, the way we structure and we build a company and the kind of feedback loops we have within the company. So there's one thing, you know, I'll mention right away, which is that we're very careful about platform purity, which means that when we acquire companies, teams, products, we re-platform them. So what we deliver to our customers is a really unified platform, not, you know, a set of different products bolted together, and that's very important to us, and we'll keep doing that. You know, so we won't stop doing it any minute, and we'll keep doing that. Another thing we're doing is that we're structuring the company in a way that we serve a very wide array of customers.
So I said in my intro that we serve everyone from, you know, the students and people who don't pay us anything, tiny companies, all the way up to the largest enterprises. But our money mostly comes from the larger side of the spectrum. So the bottom half of our customers represent just, I think, a couple% of our revenue. So from a purely monetary standpoint, it doesn't really make sense to serve the bottom half. Why would we bother? We bother because it gives us very broad exposure to absolutely the whole set of possible setups and tech stacks and new technologies and everything that is being done, and also it gives us a lot of awareness with developers and, you know, the folks that might work in smaller companies.
That awareness we have of everything that's going on, as well as that feedback loop we create with all those, you know, smaller companies and individual users, helps us build a better product in the long run, a product that doesn't get overfitted to the needs of a few, very large, very sophisticated customers. In the long run, it's a big advantage. Another thing we're doing, again, in terms of the feedback loops and the way we structure the company, is that we are very careful about having very clear metrics and revenue metrics associated with every single product we develop. So we avoid bundling whenever we can, and as a result, we get very clear signal on whether any given product is useful and valuable.
What can happen very often is that customers are very excited about a product, but they just, they just don't want to pay for it, or they don't want to pay what's needed to actually sustain the product. I think having these very clear, honest feedback loops in the company helps us build the right thing and not just accumulate, you know, good stuff that in the end is not all that useful.
Yeah, and not being used and not generating revenue.
Yes, yes.
So, okay, so you're describing to us the breadth of customers that you're able to take on.
Yep.
It is impressive. What about the breadth of the product scope? How ambitious do you wanna be here? Because you've gone from infrastructure-
Yep.
You've gone to app monitoring, you've gone into logging. They've all scaled to very large size, but you know, quietly, you've been planting the seeds. You've been planting them in security and elsewhere. And, you know, so should we think about Datadog as somehow becoming... I don't know if best of breed is the right word or, but becoming a market share leader in security, in service management, in all these other areas outside of observability?
Yeah. I mean, look, at the end of the day, we want our products in each and every one of these categories to be best of breed. So even though our strategy is to land with observability and land in general with infrastructure monitoring and then expand to the other products, each and every one of those individual products should grow to be a market leader in its own right. And what makes them even more special is that the platform integrates everything, takes the headache out of managing and paying for data multiple times for customers, and, you know, integrating everything internally and having teams in different silos. So we solve, you know, all that for customers.
The way we make that happen on the back end of it, you know, in addition to all the good stuff I said earlier about the way we structure or the signals we have for, you know, running engineering and running product, we also are very disciplined about reinvesting into building product. You know, so one key way to look at our business is that. So we built a very efficient business when it comes to G&A and go-to-market, and I would say also very efficient when it comes to gross margins. As a result, we can reinvest about 30% of our top line in R&D, and that is what helps us stay ahead with the products we already have, but also keep expanding the footprint and building more.
And then the end result with that, you know, if you project out us, you know, five, 10 years into the future, is that we get to a level where it should be irrational for customers to use anything else, because we cover the full spectrum, and we cover it, we cover it better than the others, and we keep improving it at a way that the rest of the field cannot follow.
Irrational not to use. So, let's go from there, Olivier, into Security.
Mm-hmm.
And maybe, maybe you can give us a window into why that would become irrational not to use. Because, you know, we constantly wanna understand why is security going to be, gonna be logical for Datadog. We'll talk to investors, and they'll, they will say, "Look, you're facing established competitors in these markets. They focus on one core product. It might be endpoint, it might be firewall, it might be Zero Trust. And it's like they have a battering ram, and they're kind of breaking down these doors." And then we look at Datadog, it's like you're trying to surround the whole castle-
Yeah.
Right at once. You're trying to take on all this. Can you explain how you see that working? How do you end up winning the security war?
Yeah. So, I would say the proof of the opportunity in security is that the outcomes are not there today. There's no shortage of security products being sold and bought, but the problem is still running away from most of the customers. Obviously, this is not working, and that's why there is a gigantic opportunity to be seized. Customers also are, you know, more and more aware of that. I mean, I'm not even the one to say it. I think another CEO talked about, you know, their customers having buying fatigue on security products. I think the reason for that is they don't necessarily feel that they're seeing the value for everything they're buying. So there's a gigantic opportunity.
We think we have a good shot at delivering, because precisely, we're not going after arbitrarily narrow slices of the problem. We're going after a substantial part of it to be minimally useful to our customers. We can use levers that most of the others can't, you know, such as the fact that we have the data already. There's no friction involved in getting the data in. There's no need to pay a second vendor for data you're capturing already once, and we can get security used on a day-to-day basis and implemented on a day-to-day basis, not just by the security teams, but by the development teams and the operation teams as well. So we think when you combine all that, we have a really interesting shot at it.
In the meantime, I mean, you, you're right, though, that, you know, right now, security mostly works on heavy sales and marketing, which is easier to do when you have very narrow, very narrow slices. But again, that's, that's not how we build a business. I would say also that when we started with observability, I think it's we were fundraising maybe 12 years ago or 11 years ago.
Mm-hmm.
We were consistently told by investors who are very good at their jobs, that we had no chance at going wide because so many companies were there already, and they were going narrow, and this is exactly what we've done for that field.
And so you're saying that within observability.
Within observability.
When you started it, when you started in monitoring, right?
Yeah.
Yeah. Yeah. Okay. So, we've also noticed, Olivier, when, you know, we look across the software industry, I've been pretty outspoken that software companies are really not hiring as if they see, you know, this real type of a strong recovery. You know, headcount growth rates are sluggish, they're slow, and that is pretty unusual at the moment. And then, again, we look at Datadog, you're a complete outlier, with. You have this plan to grow OpEx in the mid-20s with accelerated hiring. What do you think of this phenomenon, you know, that's happening out there? And when we look at your accelerated hiring plans, is it some kind of an indication of a positive view on demand?
... well, I mean, look, first of all, I agree with what you're saying. In general, across the field and across our customer base, we don't see hiring picking up extremely fast. Again, with the exception of companies that are AI first-
Yep.
So quite a different universe right now. Internally, though, I mean, if you replay what we said earlier, we see a 5-10x growth just in observability. We have a number of other categories we want to play in that we think will increase our TAM quite a bit. To get to this opportunity, we'll need significantly more capacity, whether that's in product development or in go-to-market. And to start seeing results on those investments, you know, to see results on go-to-market investments, it takes a year or more. To see return on investment on investments in product hiring or engineering hiring, it takes two years or more, so we need to get started now to get a lot of that done.
In general, we have really high return on investment. I mean, we've given two tidbits recently. One is, if you look at the products we shipped, between 2020 and 2023, I think is the number we disclosed last quarter, those products add up to more than $200 million in ARR today. So we see very high return there. And on the go-to-market side, we have retention rates that are extremely high, you know, in the mid-90s%, overall, which includes SMB, and on the enterprise, we're in the high 90s%. Which means that it is definitely worth it for us to invest in sales and marketing and acquire new customers. These are going to be very long-term relationships that are going to be very profitable for us.
In the very long run, we know it's a very good, very profitable business with a high return on investment, so we'd be foolish not to invest more.
Okay. It's wonderful to have a little window into your thought process there. Let's try to cover the topic of Generative AI in five minutes, Olivier. So we've been positioned as extreme bullish outliers on the importance of Generative AI. That has been the case for us since the fall of 2022. We have been saying that it would happen first at the infrastructure layer, near where Datadog is operating, and we have been saying it would require quite a bit of patience to see that monetization occurring up at the application layer. And as we look at it today, there are very few software companies now who will actually tell you how much of their revenue is coming from AI. It's really just Microsoft and Datadog at this moment.
And for Datadog, we're getting this 3.5% of ARR. And you said that that's coming from next- gen AI customers. Can you help us peel back the layers? What, what's the nature, what's the composition of that 3.5%?
Yes. So what we put in there, I mean, it's a little bit of an arbitrary group, but we put in there the customers that we... As far as we can tell from what we know, are substantially all built on next- gen AI. So these are not existing companies that are shipping AI features, but these are typically the AI, the model vendors, or the, the companies that ship infrastructure components that are, are useful in the gen AI stack, or companies that are for, for whom the whole functionality is built on gen AI, basically.
So where do you see that 3.5% figure going? So it's a very pure definition.
Yes. I mean, the-
It's extremely pure.
Yeah, the rest of the customers might use AI, too, or they might have features, but we, we don't consider them AI native in the same way.
Yeah. You're, you're not gonna take the JP Morgan monitoring of AI, and-
You know, JPMorgan is early on AI, but not an AI native by any means.
Hopefully, that's in a process of changing. I believe it is. So, that 3.5% figure, where do you see that going, Olivier?
So first of all, I think it's not a measure that's—I mean, I think it's going to lose its relevance after a while, because right now, the bulk of AI adoption still happens through a relatively small number of model vendors. And so that looking at this small set of vendors is a good way to get an understanding or to see a bellwether of what AI adoption is going to look like down the road. But I think as AI adoption broadens and most of the applications get into production at non-AI native companies, and more of the models are run outside of these core main AI model vendors, I think that number will become less interesting in general.
Okay. Now I'm thinking back a year, Olivier, on this stage. You said that as a result of GenAI, what would happen is developers will write code faster, therefore, they're gonna deploy apps faster, and then those apps need to be, need to be monitored. Datadog, we assume, is getting more than its fair share. Have we seen that happening yet? Has it been an accelerant to app development?
We see a little bit. I think it's still anecdotal in general. But, you know, the way to think of it is, it's. The benefits are real. Nobody who's actually trying the AI writing copilots, sorry, code writing copilots, nobody's taking them back and removing them from service. Everybody keeps using them and using more of them. So I think the way to think about it is that it's, we are at the beginning of an S-curve there in terms of the adoption and the benefits we're going to see from that.
Okay, one last question for you. Probably many of the investors in the audience have to wonder, at certain times, are we in a GPU bubble? Are we in kind of a GPU bubble that could bust at some point? And the reason is the LLM start with training, it's very GPU intensive, and then they transition to inferencing. And so because your Datadog is handling the inferencing phase of AI, maybe you have some insights. What do you think the transition from training to inferencing might look like?
I think the transition happens when applications move to production in general, and when they start scaling with their end users. I think we see that happening already with again some of those core vendors that we have in that 3.5% of our revenue group. So this is definitely happening. We want to see that as a sign of what's going to happen later to the rest of the industry, but for everybody else, it's still early. I think the vast majority of the industry is still largely training or prototyping models, as opposed to running them in inference. But again, for it to make sense in the end, there needs to be a lot more inference than training. Nobody makes money on the training. It costs a lot of money.
For all of it to make sense, there needs to be a lot more inference than training.
Wonderful note to end on. Olivier, I can't thank you enough for taking the time to be here with us.
Thank you very much.
Very much appreciate it.