Datadog, Inc. (DDOG) Earnings Call Transcript & Summary

Good morning, everybody. I am Sanjit Singh. I head up the infrastructure and analytics coverage at Morgan Stanley. We are very pleased to have Olivier Pomel from Datadog. Datadog is one of the most exciting companies that have come out in quite some time in software. They are coming off a great year, having done an IPO, brought 7 new products to market and have grown the business north of 80% this year. So welcome, Olivier, and thank you for joining us.

Maybe to kick off the conversation, for those who are unfamiliar about Datadog and the problems that they solve for customers in the monitoring analytics market, can you sort of take us in the shoes of an end user? Like what are the problems -- when they come to work, what are the problems that they're dealing with and that Datadog is helping them solve?

Yes. So at a high level, we do -- what we do is monitoring analytics for infrastructure and applications in cloud environment. And we sell to the engineers for the -- in operations or in development that actually have to manage these infrastructures. To give you a sense of what we do, it's absolutely necessary when you run an infrastructure and an application to have some visibility into it. There's no choice. You absolutely have to have it. So the metaphor there is if you're flying a plane with a cockpit, so you sort of need one, really no other way. But to make things worse, in our world, one person is going to fly 1,000 planes at the same time, and then the engineers are actually going to be making changes to same flight, so it's even harder. And our job is to make sense of all of that, all of this complexity, that the engineers can actually keep flying the plane and keep everything up.

And maybe just if we think about the monitoring analytics market, it's a market that's really driven by some of the changes in the ways of compute. And I wanted to get a little bit granular in terms of where you think we are in the sort of current era when we talk to kind of 3 major forces: one, the move to cloud, first; two, the adoption of micro services and containers; and three, serverless which seems to be another sort of model that companies are using to sort of move faster in terms of rolling out new products. Where are you in each of these phases from -- when you talk to your customers?

So the first one, so the move to the cloud, we're still fairly early. Depending on who you listen to, it's going to be in a single-digit percentage of the workloads that have moved. Sometimes it's in the -- between 10% and 20%. Again, it's very hard to know, but depending on who you listen to, it is in that ballpark. So we're still fairly early. If you think of the whole s-curve or technology transition that we're tracking, we're in the very beginning part of it. I would say, though, we have signed up in the transition that we know how broad it is. We've seen that it applies across the market, to everybody, from SMBs to large enterprises and across all industries, and we've seen that in our customer base so far. So that one, we know it's happening, still early, but we sort of have a good idea of who it applies to and how big it is. The second one, so containers. I would say we're at the point where the market is starting to coalesce on the reference architecture and reference solution. So we've seen it pick up over the past, I would say, 4, 5 years. And I would say it's only in the last 2 years that we've seen the market coalesce around Kubernetes as being the solution for managing containers and orchestrating containers. So there's -- again, super early in there, even earlier than the broader cloud transition, but at least there's also a common destination there, common way of doing it that is accelerating, I would say, the transition. Then the last one, serverless. They were even earlier. It's brand-new. We're at the point where there is a ton of excitement. And actually, if you want to read a bit more about it, we published a study of what we see our customers actually using in serverless. So if you Google The State Of Serverless Datadog, you'll find the study and it has a number of data points that are very interesting, if you want to track the trends there and see what people are actually adopting. I would say, for serverless, there's not a Kubernetes yet. There's not a, hey, this is absolutely the way you should do it. So the reference architecture and the destination is a little bit in flux, but what we know is that there's a lot of interest for it.

And if you think about what these forces have meant for Datadog, could you just give us the context of what you saw when you saw these -- in terms of these trends sort of taking place within your customers? And how much of a tailwind has that been for the business itself?

Yes. So of course, I mean, the catalyst for us has been the cloud transition, right? So there's been infrastructure monitoring and monitoring in general, systems and computers to monitor, and this is a category that has been very, very sticky, so it was always very difficult to rip and replace whatever was there. The reason why customers are adopting us and the reason why we're being so successful with customers are competitive replatforming and moving to the cloud. So that's the initial catalyst for us. I would say there's 2 other factors that combined to with that. The first one is that this cloud transition, this replatforming, takes place in the context of digital transformation. So companies are becoming software companies, which means that not only are they changing the way they're implementing their infrastructure and their applications, but the definition in terms of the footprint of the impact of these applications and the infrastructure is a lot larger than what they used to have. They're turning into software companies. Then the last thing that we should consider there is that when you talk about the move from just cloud to container to serverless, what we see there is a continuum that is taking us from a few things that are changing very slowly, so physical hardware, that was before the cloud, to tons and tons and tons of tangible things that we combine all the time in real time, which is serverless. And so what you see with that is that there's an explosion of complexity. And that explosion of complexity is actually the problem we solve for our customers. It's how do you actually get an infrastructure that is 1,000x or 10,000x as big, that is much more important to you because you are a software company, that is going to have 10x more engineers on it because you're also turning into a software company, and that is going to change all the time. That's the problem we solve.

Yes. Such a great point because when I talk to investors, one of the things that I hone in on is that while this performance monitoring has been around for quite some time. It's probably harder than ever to do today versus 5, 10, 15 years ago when we were dealing with on-premise applications and data centers. If we think about why you guys have the leadership position that you do currently in the market, you've been ahead of all of these trends, right? And some of your competitors have been caught more by surprise in terms of the pace of how the shift is coming. And I guess one of the key investor debates on Datadog is around that durability of growth because there are going to be other ways that are going to come, right? Like maybe serverless, maybe there's something after that. As you think about how to sort of sustain growth and keep growing market share, what is about the business that's going to keep ahead of the curve? Or said another way, what happened to some of your competitors who used to be leaders, right, in the category? What happens -- what happened then in the last couple of years, happen to you 2 or 3 years down the road?

Well, so there's a couple of things about it. So first, if you go back to our differentiation, our differentiation is that -- so we start from the infrastructure, which gives us a way to get deployed absolutely everywhere at our customers from day 1, basically, in the cloud environment. So we're cloud native, we start with the cloud. So we're going to grow with this cloud environment. We did a product that is extremely easy to adopt. The reason why my good friend and I started the company was to get teams that were separate and didn't talk to each other to actually use the same platform. So we wanted to help for that to reach across different teams and was accessible to everyone. So when you combine all of that, you see that we're -- the product that were going to be used by everyone. We have a lot of serverless contact with our customers, which gives us an ability to keep solving more and more problems for them. So as we go, the way we've been growing, the way we've been developing the product is we saw a bigger, bigger problem in the middle of this serverless contact. And we basically keep up with the trends that our customers see.

Is there anything to the story about the profile of your customer base? You have, I think, roughly about 1,000 customers who are spending more than $100,000. But then you have thousands of customers that are maybe not spending huge amounts to Datadog but are probably maybe the most modern, sophisticated, quickest to adopt to some of these trends. Are they providing -- to what extent are they providing a feedback with respect to the product engineering organization?

Yes. Our customer base covers a gamut. So we have large enterprises. We have mid-market enterprises. We have small and medium-sized businesses, and then we have individuals. We even have a free tier for individuals and small companies and start-ups and students and whatnot. What we see is that the cloud transformation actually covers this whole broad spectrum of customers. When you look at our business, it's very interesting because the trends actually are very similar between a lot of the different segments. We see the same level of growth, same level of multiproduct adoption across all the different business segments. When you think of the low end of it, which is either customers or users that don't pay us or customers that pay us very little, so the low end of SMB, what they bring to us is that they give us some network effect when it comes to seeing every single combination of the different software and platforms and basically battle-testing all of our integration and also several work scenarios. So we see that happen. They help us develop this integration and battle-test them. And that actually feeds into the success of the product. In addition to that, our product is successful because it is very easy to adopt. It doesn't have professional services or it doesn't require professional services to be adopted. And the way we maintain that is we don't just hire designers and tell them, hey, please make a simple product. The way we maintain that is we actually serve all of these individuals and SMBs. And if they are able to self-provision, self-deploy and adopt the product on their own, it means that the [ mere modules ] in large enterprises can do it as well, which is core in being successful there. If we break something in the product to serve the large enterprise customers, we lose this -- or we'll hear about it from these individuals and low-end customers. And that's for this whole feedback loop works for us.

You touched on the point around essentially what's kind of like a product-led growth model. And you see a lot of -- or a few modern software companies really optimizing for this model, Atlassian, in particular. When I look at the profile of your business growing 80% year-over-year and you have positive operating cash flow, really highlights kind of the power and the efficiency of that model. Can you talk a little bit about some of the key tenets of the product [ like gross margin ] allows you to grow so efficiently?

Yes. So the -- so first, the -- landing a customer is efficient because we don't look at our professional services. We're not the ones that are actually selling the cloud transformation. We hop on it at the right time for these customers. So the sales cycles are short, and typically, I would say, cheaper than what you'd expect for standard enterprise sales. In addition to that, we're very heavily land and expand. So we're always going to land fairly small. We're early in our customers' migration to the cloud. As a consequence, our customers are going to be relatively small in terms of their footprint in this cloud environment, and then we're going to grow with them. Because we start with an infrastructure product that has to be deployed everywhere it makes sense, we're going to be covering the whole infrastructure, which is small to start, and then we're going to grow it as customers move more and more workloads to this cloud environment. Then the last part of it is we've been adding more products to our platform. These products actually belong together in a way that makes sense to our customers. What I'd like to say is that the problems that our customers have to solve won't stop at the boundary between the different products. So they actually have to use them all together. And the adoption of these products is largely frictionless and that customers can just turn them on, developers can start using the APIs or they can start sending data and they can start manipulating this data. We don't need to call them. They don't need to call us. That's just part of the platform.

Got it. When we think about this model around product-led growth -- and it's a little bit of an unfair question because you guys are doing great. You're at $400 million plus a day. But we think about getting to $1 billion and beyond the $1 billion, sort of the long-term story, do you think this model sort of scales beyond those revenue thresholds? We've seen other companies in software, they do great, they get up to $500 million, get to $1 billion, and then there's a little bit of a -- sort of some adjustment period to get beyond that. What do you think about sort of the long term about the model that you're running today and how it may or may not have to change when you think about getting beyond $1 billion?

Yes. So we -- the model we have today actually isn't very different from the model we had 2 or 3 years ago. And we've seen a lot of stability in the economics of the business. And when you think of the -- so one metric we look at internally is the tax at that time. And it's been constant at or below a year for a very long time with the company. It doesn't really change over time. So it doesn't really change depending on the segment of customers. If you look at SMB or mid-market or larger enterprises, it's about the same. Of course, the mix of spend is different in these different segments. We spend more on marketing, for example, for SMB, and we'll spend more in sales to enterprise. But the overall mix hasn't really changed and is fairly constant. So with that, we think -- we don't foresee a massive change in the future. But at the same time, we're very efficient today. We don't want to be prisoners of that efficiency. We want to keep bringing more products to market. We want to cover the whole market. There's a number of things we want to do. So if we see a way to grow certain parts of the business or to enter new markets and new categories with slightly different payback, we go forward.

Understood. You mentioned earlier about how the product portfolio has been expanding. It's been -- it is a focus and will continue to be a focus of the company going forward. You were the first to get into -- or sort of expand from infrastructure monitoring to APM, I think, in 2017, last couple of years, went to logging, you added the network. There's a security story coming on. Can you talk about the forces that are sort of consolidating that market? It used to be a best-of-breed point solution market for APM. There would be one vendor for a customer, logging a different vendor. But as we're seeing more of that consolidation, what are the forces that are driving that consolidation? Because it seems like there's a more fundamental reason than, oh, you just want to expand your TAM. It seems like there's a customer motivation around here as well.

Yes. So it makes sense. That's the thing. It wouldn't work this efficiently if it didn't make sense to the customers. You can look at it a couple of different ways. The first one is, if you look back at the world 10, 15 years ago, developers used to be fairly separate from the IT operations people, and so they would buy different tools and live in different worlds, which is, again, the reason why we started the company. We actually wanted to bridge the different worlds and break down the silos between them. So that has changed in the cloud world. The cloud is all about disintermediating IT, giving developers some amount of self-service and basically pushing those teams together so they can build faster, iterate faster, change your mind over time, that sort of thing. That's one. The other way to look at it is that even if it actually sort of made sense already to have those different teams talk to each other, the world was dominated by the products that were already there and really hard to replace. So ever since the first server was done then, there were software sold to monitor that server. And infrastructure monitoring is an extremely sticky category. That's the reason why most of the money still goes to an IBM, who used to be HP, who used to be CA and BMC. Those categories are super, super sticky. So when new vendors came by and had to add new functionality, they actually had to package it as something that was completely separate and sold differently and layer on top of it as opposed to requiring a rip and replace. Again, today, everything is coming up for grabs at the same time as we're replatforming to the cloud, an additional opportunity to bring that to customers. And we've seen internally -- from the moment we could tell all customers we had all their needs in one platform, it makes so much sense to them. It accelerated the sales cycle. It was actually a lot easier for us to start selling 3 products with infrastructure, APM and logs than just 2 products with infrastructure and APM. It makes so much more sense when you integrate everything.

And you sort of mentioned that the sort of replatforming is the major catalyst going on right now today. If we look at the pace of consolidation by SMB, mid-market, large enterprise, any way to characterize how that's evolving?

Yes. It's about the same. We see the exact same trend between those different segments. The number we shared was that about 60% of our customers are using 2 or more of our products. About 25% are using 3 or more of our products. And we see about the same numbers in each of the segments.

In terms of those metrics, the sort of like my next question about the big acceleration you're seeing in sort of multiproduct adoption. How much of the product adoption is happening more or less organically, you're not spending too many calories to get certain product categories adopted? Which are some of the most naturally attaching products? And which ones require a little bit more of a directed sales effort?

Okay. So it's largely organic. Again, it makes sense because the problems actually span the different categories, so the solutions have to and customers are going to naturally adopt them. Internally, we have the same sales teams and customer success team service all of the various products. We don't have any other ways -- we don't have any specific incentives for sales teams to sell multiple products or specific products. So all of it is very organic today.

And in terms of any change in how customers are landing. So infrastructure monitoring has been your core landing product, but as you brought on logging and APM, are you seeing any changes in terms of which products are landing now? And just the overall deal sizes, are customers landing initially with 2 or 3 products versus last year?

Yes. So we see -- the number we shared with that last year, about 65% of our new logo deals involve 2 or more products, which was actually a bit of a surprise for us because we did think we'd still land with the infrastructure product and expand to the others. We've seen a lot more adoption from multiple products from the get go than we thought and much faster than we thought. The reasons for that, we think that, first of all, there's tons of white space in the cloud world. We -- mostly, customers who are new to the cloud, they want to instrument the applications across 2 or 3 other categories at once when they do so. The other thing that we saw there was that there was no big difference between the adoption of APM and logs in addition to infrastructure. We still lead with infrastructure as one of the 2 or 3 products we typically land with. And the reason for that is that the go-to-market for it is so clear for us. We are the first, call it, cloud provider. It's immediate. It works. So there's no need to mess with that.

Right. Let's talk a little bit about the competitive environment. And so you've had a first-mover advantage when it comes to this current cycle, this current wave. And now you see a lot of the players in the space sort of expand their own product portfolios or at least working on that right now. Everybody is using the term servability. And so I think from a customer standpoint, if I'm a customer, I'm thinking, well, why these guys sound the same? How are they differentiated? What are the sort of the implications either from a customer messaging or product differentiation that you're focused on to continue to differentiate yourself from the pack as all of your competitors sort of adopt a similar playbook with maybe a twist here and there?

Yes. So the differentiation doesn't come from the messaging so much as it comes from the product and the company that we built around the product. I think -- when you think of all the things that differentiate us, the fact that we are platform-first, cloud-native, that was infrastructure that we deploy everywhere and also going to be used by every single engineer without any professional services. It's actually really, really hard for companies that are -- that start from a very different point. They don't have a platform to start with. They are not in the cloud to start with. They're not deployed everywhere because they're not -- they don't start with infrastructure. And they also have products that tend to be products that are going to appeal to 5, 10, 15 power users and be deployed with heavy professional services. It's actually really, really hard for them to change from that to what we have. So the way we keep differentiating every day is by being adopted this way by customers, bottom-up with a broad platform approach in the cloud.

In terms of just the dynamics of the competitive environment, to the extent there is a competitive bake-off, who do you see most, let's say, in the mid-market versus large enterprise? And then how much of the market is greenfield for new workloads? So contextualize that for us.

Yes. The majority of it is greenfield. We come in when our customers move to the cloud, and they absolutely have to get visibility. And they give us a call right away, and we instrument the environment. What we see the most is the intent or the temptation to build it themselves by customers by coupling together via open source technologies. That's where we see the most. And of course, when we sell to large enterprises, in their on-prem environment, they're going to -- they all have an APM, they all have log management. So most of them are going to have either an AppDynamics or Dynatrace on-prem. They're going to have Splunk on-prem. So it actually doesn't preclude them from adopting us right away for APM and -- or logs in their cloud environment.

Right. We're talking about the pervasiveness of infrastructure monitoring in the environment. When we think about -- I'm going to call it 4 pillars of reason. But if I think of infrastructure, logging, APM network, which of these capabilities, which of these pillars represent the largest areas of spend and are seen -- and which of them is sort of seen as the most strategic to your customers?

So it's a bit hard for us to think in terms of relative weight between the products because we intimately think that this is one big unified category and that the difference between the products won't actually make sense in the long term and that there will actually be quite a good amount of fungibility between them. And we see that today already. We see customers that are maybe on standby sending us logs for certain use cases. And then a few months later, they realize it's actually more efficient for them to send some of those as APM traces or as information metrics. And so same use case, same customers, but the spend shifts from one product to the other. If you look at the historical categories, the largest was infrastructure monitoring. And APM and logs were, together, about the same size as infrastructure monitoring. But again, in the future, this might look very different. The way we think about our TAM is that there's one big category today that we're a leader on, which is observability. And that includes everything you mentioned and also our Synthetics product and Real User Monitoring product, that's all observability. There's a couple more categories that we're entering or will enter in the future. One is security. We have the very first version of a product there. And the other 2 are ITFM and real-time BI.

Can we touch on the security? It was not a question I have, but since you mentioned it, is -- what's going to be the security value proposition? Is this going to be similar to what Splunk is doing, sort of a modern SIM, or what sort of Elastic is doing around security analytics? Explain what your vision is in security.

So it's interesting because the problem we see in security is very similar to the problem we still have to solve initially of the development teams and operations teams being separate and not talking to each other and not having the same view of the world. Today, what we see at our customers but also what we've seen in the past and that one company including Datadog is that it's very hard for security teams to work hand-in-hand with development teams and IT operations teams. They live in separate worlds and separate silos. They have interest in the -- getting the applications instrumented, but they don't actually have access at the right time to begin instrumentations. So we think that by bringing all of 3 in -- under the same roof, we can really allow our customers to operationalize security, which they have -- broadly haven't been able to do so far. And then when we look at the way we entered the observability market, we started with the cloud environment, which are new and which are almost entirely white space. And we see the same being true in security. The cloud environment are white space. There's no dominant solution there. Customers mostly have nothing or they have some open sort of trend to pull together. So there's a door that is wide open there for us. That being said, the product we announced in security is still very early. We now see it because we wanted customers to self-select and come to us and help us design and build that product and we don't want to build it in isolation, which is going to be a long investment. We expect to spend multiple years to fully flesh out that product vision in security.

It sounds very interesting. Getting back to the revenue opportunity within the customer base. This might be a harder question to frame out. But if you think -- you sort of hold the host environment constant, through a customer that started with infrastructure monitoring and you sort of index that at 1 like -- where does that spending go as they bring on Real User Monitoring, Synthetics, APM, logging? Is that 2x, 3x, 5x? Is there any way to sort of frame out that opportunity within the customer as they bring on more and more product into the organization?

Yes. So it's a bit hard to tell, as I said, because we're still early in this transition. Today, the one number we've seen is customers that have 2 or more products spend about twice with customers that just have infrastructure spend. And -- but at the same time, our infrastructure is growing very fast. And our new products, APM and logs, are where we're underpenetrated in the accounts that are using them today. So very early moving target. There's going to be a lot to look at there.

Can I ask a question about logging and the particular big board logging? I think customers would like to log a lot more than they do. Sometimes pricing becomes an issue. Can you talk about your logging product and the pricing strategy that you went to market with versus some of the incumbents in the space?

Yes. So the issue there is if you just charge by volume, in many cases, you can become unaffordable for customers. So what we did is we have a differentiated pricing where customers can pay separately for just sending us the data and just pay a very small amount for that. And if there was more for indexing it and making it available for analytics. And we had customers actually change that in real time, so there can decide in real time, hey, I actually want to drop that product or index that product, or I want to be the index that very specific bid for the next hour, and I want to keep it for a month. We can do all of that. And also bring back data out of the -- just send to us an archive and bring it back to life for analysis if they need to. What it does is that it gives them the lever. So they can adjust how much they pay and that -- with the value they get from it. In the short term, it means maybe we charge less than we could, right, because they can decide, hey, we actually don't need all this, so let's just pay less for it. But in the long run, we think that it means we'll get many more use cases because customers are going to send all of the data they'll have. They won't have to ask themselves, should I really send it? They'll have it all available, and we'll get more and more use cases.

So essentially, you sort of bifurcated the pricing between -- a separate price for ingesting and then a more significant monetization story once they sort of actually start to run the analytics on...

Yes, exactly. And they're in control, which is very important to them. Because the one thing you can't do is have a bill that grows much faster than our customers' businesses over the long term. So it's good for them to actually have those controls.

The next thing I wanted to hit on in terms of just kind of the debate, not just for Datadog but kind of across software, is sort of open source. And there's a movement in the space around open telemetry. There's some open source tool like Grafana, et cetera. When you think about, let's say, open telemetry, for example, which is kind of a standardized way to -- agents to collect data, what does that mean for Datadog as you see users sort of use these tools, incorporate them into their dashboards? Is that sort of threat for Datadog? Or how do you sort of view the open telemetry movement?

Yes. So first of all, it's always been there, right? The -- today, it's open telemetry, but there's been a number of different open source libraries and standards before to gather data and to submit it. We support open telemetry. We invest in it. We have more customers send data to us through open telemetry. At the end of the day, we think the value comes from bringing all this data together from a number of different sources. And we're agnostic in terms of how we get it. It could be through open standards. It could be through a very specific integration or specific pieces of infrastructure. It could be through our own proprietary technology. It doesn't actually really matter. What matters is holding everything together in the end. And when you look at customers, to them, what wins every single time is to make it easy so that whatever they're using, whatever combination of all the new software and -- the shortest path from the big math to actually understanding it is Datadog, and that's what we invest in. Sometimes, it means going through open source sometimes, not sometimes, it means a combination of both.

Right. It's always seemed like just another data source into the platform.

Yes, exactly.

Right. I'm going to turn to the audience and see if they have any questions. Let me get to one question before, and then if anyone has any questions, just raise your hand and we'll get to you. So looking to 2020, what are sort of the major investment areas that you are sort of targeting and budgeting for? With such a large growth opportunity, is your focus on building out sales coverage, the sales organization, more investment in product portfolio? What's sort of the mission statement going into 2020?

So both, actually. We invest both in go-to-market and self-coverage as well as engineering growth to produce more product. And the way we think about it is if we hire enough of the right salespeople and onboard them and retain them, it's going to be the best predictor of success a year from now. And if we hire enough of the right engineers and onboard them and retain them it's going to be the best predictor of success 2 years from now. But we're -- right now, we're so early that we're investing both in go-to-market and heavily in product.

Do we have any questions from the audience? Looks like we're covering them all. Good. In terms of international, like that's been inflecting as a percentage of the business in growth overall. What are the difference are you seeing in terms of adoption internationally versus in the U.S.? I think, historically, people have talked about cloud adoption being a year or 2 behind the U.S. Are they at their inflection point now? What are sort of the trends that you're seeing across Europe and Asia?

Yes. So we clearly see inflection points internationally, both in APAC and in EMEA. You're completely right that, that was -- I think, I would say, maybe 2, 3 years behind the U.S. We saw -- 3 years ago when we saw large enterprises in the U.S. really accelerate and get to some scale in some environments. And we -- I think we started seeing that happen in Europe over the past year and in Asia over the past year. So we're investing. We're also fairly early in our coverage of Europe and Asia. We started those teams also over the past 2 years. So that's fine. We're concerned with -- we would be busy building up those teams to take advantage of the opportunity.

Got it. M&A, there's some consolidation in the market. I think Splunk made a couple of acquisitions in the space. How do you think of M&A as part of -- how does that fit into the overall growth strategy? What sort of -- how do you think about the build versus buy decision?

So we've used M&A before. We made a number of acquisitions. They tended to be mostly payment product acquisition or, I would say, payment technology acquisition, and we'll do more of that. It's possible we'll do bigger acquisitions. We -- the question is what's going to be the best accelerator for the business, and sometimes it can be something bigger. So we actually want to use those tools. At the end of the day, we have such a great opportunity because we -- as I said before, we did [ categorize ] our customers, and use by all the engineers. So we have a very, very large serverless contact with customers. We have the opportunity to cover more of the space in between. There's many directions in which we can and want to go. And some of those, we think, will be accelerated if we apply our way into them.

Understood. In an environment where there seems to be a little more focus on margins and profitability, I mean, at the time you guys were growing, I think to set up a long-term target of 25% on non-GAAP margins. How should investors think about the trajectory to that target? And any sort of time line to get there?

So we haven't shared the time line on the long-term model. What I'll say is we're investing today. We're investing as fast as we think we can without being counterproductive. So I think the question is, are we growing the business over the past couple of quarters at more than 80% year-over-year? We're investing at a rate that is commensurate with that. I think a little bit below, which is why we've been profitable or breakeven over the past couple of quarters. And the intent right now is to keep investing for the foreseeable future. I think what we've proven, though, is that we can get to a point where the business is profitable. I think right now, we'll have so much growth ahead of ourselves that, that's other time for that.

So much opportunity ahead. So maybe we sort of end the conversation in terms of where we began, which, as we talked about, sort of the market forces, cloud, serverless, microservices. What are you sort of getting in front of mix? Like what do you see sort of the future of the market? Where is the market headed? And what are you sort of investing to get ahead of that sort of next trend?

Yes. So I mean, look, we're still in the early days of the cloud transformation. We're still at the early days of companies turning into software companies. So there's a ton more value to create, to capture there. So we're busy working on that. As I mentioned before, there's a number of other categories we want to enter. And we think we can play a role in these categories because we see our customers using our platform for that today. I mentioned real-time BI, for example. We see our customers that are largely digital get real-time insights on their businesses in Datadog. We don't have any fully productized offering for it. So they just build that themselves on top of our platform. But that's an avenue for growth in the future.

And with a minute left, I have to ask the obligatory coronavirus question. How are you thinking about that in terms of potential disruption of the business? Have you seen any impact yet? How are you sort of gaming out...

And I see I'm the last on stage today. So...

I appreciate that.

The -- so look, we don't have a supply chain. We don't have significant business in China. I would say our exposure is not very specific to the coronavirus. But as everybody else, we're trying to figure out how we're going to run our own operation, maybe we can employ them to come to work and that sort of thing.

Got it. Well, with that, we'll leave it there. Thank you so much, Olivier. Appreciate it. Thank you.

Thank you.