I understand what you're saying. Yeah, I understand what you're saying. Yeah, thanks. What a delight to host two companies from the great New York City. Yeah. Headquartered in New York City, and my favorite outside of San Francisco ever. I think I spent most of my time outside of San Francisco in New York. I love New York and welcome David Obstler, the CFO of Datadog, which is based in New York. Yes. MongoDB had Mongo based in New York City and Datadog. It's very appropriate that we're back to back. Dave is on our board. No way. A long time, Director of Datadog, and we learn a lot from each other. It's a very appropriate pairing. Kindred spirits. Kindred spirits. Two great companies back to back. Welcome back to the Goldman Sachs conference here. Thank you. It's great to be here again. Fourth year in a row we're doing this fireside chat. Yeah. I think we've done this at multiple ballrooms, whatnot. Yeah. A delight to have you back. I keep asking you the same question. What is the vision for Datadog five years out? If we are to come back, Communicopia, I think it'll be 2031. What does your company, Datadog, look like? Just as Dave was asked the same question about MongoDB, what does Datadog look like five years from now? Yeah, I think we want to look at our customer, which is sort of the production engineer, the reliability engineer, the DevOps. We want to be the platform they turn on in the morning and never turn off or perhaps never turn off. When you think about how the world's moving, we'll talk about it towards more and more complexity of applications, more and more migration. There are many more sort of use cases or breadth of use case that we can satisfy with that customer base. We've already, I think, had a strategy well articulated in our platform in metrics, traces, logs, observability, but anything that touches the function of that application when it comes to, and we'll talk about a database, network, LLMs, service management, we want to own. We want to spread out our use cases to things like security or DevSecOps and sort of coding tools. That's the vision. That's what Olivier, our Founder and CEO, and his partner, Alexis, have been doing relentlessly since the founding of the company and want to continue doing. How do you operationalize the vision? What are the things you're doing to put this in action and help actualize the vision of the company? Yeah, that's right. That's where I come in. There are many, many types of product enhancements and go-to-market enhancements. I think we're in a very good position given the size of our customer base and platform and the fact that we get real-time feedback from customers. We, as many of you know, are a consumption-based model with underlying subscriptions or credits. We actually can see what our clients are doing. The philosophy has been to look at what they're doing in their day-to-day operations and have a list of things where we can enhance value or develop the platform and then get that feedback from customers. We've been, and I know it's one of your questions, announcing various milestones, $50 million, $100 million, $750 million of parts of the platform that are going to be adopted. What we do day to day is think about how important a use case is and can that be evolving from that $50 million to $750 million and beyond. That's how we do it. It's mainly from the customers. The compounding S curve. You have got a product. Yeah. There are disclosures for logs. Right. A whopping number a few years ago. Yeah. Yeah. I want to go back to the vision question a little bit. You threw in LLMs, networking, et cetera. In the cloud world, it was easy to understand how that network topology and the infrastructure layer got to be much more complicated. Yeah. was much more massive scale and how Datadog kind of rode that wave. Right. As you think about AI and what's ahead, what is the relevance of Datadog's core technology in an AI world? Yeah, and sort of stepping back, what we saw in history is as the world got containerized, Kubernetes, serverless, as it became impossible to monitor these applications using legacy observability platforms, that enhanced Datadog. We see that happening again when it comes to GPUs, LLMs. In terms of AI, there are a number of things we're doing. I'll start with our platform itself, our product. One is our goal through our integrations is to monitor wherever the workloads and data goes. It's Datadog. We're essentially developing integrations into the AI tools. We have 4,500 of our customers now sending us data from AI tools. We want to be able to monitor, and we can, GPUs and CPUs, and we want to refine that GPU. We also want to be able to monitor the LLMs in the application, and we announced our LLM monitoring product. What we want to do is monitor all the content. That's one part of it. We want to AI-enable our platform. We believe that to continue to be the leading observability platform, we have to inject AI into our platform, and we've been calling it blank bits. An example of that is our service management, where we've always had machine learning, we've always had Watchdog, we've always had analysis of correlations and what might be happening. We're using our LLMs and outside LLMs and training them to get quicker in the diagnosis of problems and therefore be able to become more reliable and speed up the work of our clients. That's a good example of a platform feature. We have the customer base, which we'll talk about. We've always been a company, because of the innovation, the R&D, that has been the choice of platform for what we used to call cloud natives, but now we've created a new segment called AI natives. They're essentially cloud natives. If you look at some of the disclosure we've made, and we could talk about this further, we've been gaining quite a bit of traction in that market. That seems to be where a lot of investment is going. We want to monetize that in our customer base. It's an accelerant, as you just heard from Mongo, as you've heard from a lot of companies. Lastly, what about internal to Datadog? What are we doing? We're trying to dog food our own uses, and we're trying to use AI, whether it be coding tools or the service management, more proactively in order to accelerate our product development, as well as eventually we maybe become more efficient in the spending. Got it. Got it. Yeah, it was a long answer, but it's a big topic. It's a good one. That's pretty technical for somebody with a finance background. I think I've mentioned this a couple of times before. Are you sure you're the CFO or you're the Chief Product Officer? I'm not. I look at all the great work being done by our engineers and product, and try to understand it. Thank you. You've done really well observing that. Yeah. Now, to dig into consumption trends. Yeah. We talked about SMB improving at the margin, enterprise established coming out of Q2 results. Yeah. Can you just give us an update on the broader spending trends across these cohorts, enterprise versus SMB? Yeah. I think in SMB, when the bubble burst, as you all know, we had funding pullback. We had a change from growth at all costs to the combination of growth and efficiency. That hit the SMB. Now for us, because you have to have a cloud deployment, we're not talking about what some of you might think of as SMB. It's not your corner dry cleaners. Essentially, many of these companies have significant revenues and 500 to 1,000 employees. They had to change what they were doing and funding got constrained. We went through an adjustment there. What we saw in the last two or three quarters is stabilization. This is excluding the AI. If you add the AI in, you would see, because most of them are SMBs, as we define it, less than 1,000 employees, you'd see a material increase. In the non-AI, what we've seen is things return a little more back to normal. In the last quarter, we saw a pickup of our net retention there, meaning what they're doing is getting back to, maybe they've calibrated, they've optimized, and they're getting back to the appropriate balance between growth and cost. We've seen that. In enterprise, this is where we have a very, very long opportunity, meaning if you look at the percentage of workloads that are in the cloud and then modernized, not lift and shift, but modernized, you see we got a long way to go. That might be in the 20% or 30%. There are many enterprises that are right at the beginning of this. What we've seen is a return to, I would say, the priority projects. Some of them are AI related. We've seen steady growth and consolidation, meaning we've seen similar growth rates as we've had in previous quarters. We still have a careful environment, a balanced environment. What we're trying to do there is expand our enterprise sales team. I think we got a little behind in that. Maybe we risk managed a little too much. There are a lot of geographies we can talk about. We're trying through the combination of product innovation and go-to-market to accelerate that. We're in a good place. We're not in an ebullient place. Got it. Yeah. That's good to know. You've seen good growth from, actually tremendous growth from AI native to your point, which is not included in the SMB consumption. Mm-hmm. Over the last 12 months or so, how do you think about the potential for sustained growth for Datadog in this cohort? Why is Datadog so well positioned for this AI cohort? Yeah, it's a great question. We follow the workloads and we follow where revenues are being gotten. You can see a lot of these companies are publicly announcing their progress. They're giving you revenues. They're giving you funding rounds. We have a business here that has hundreds of customers, indicative of the demand environment. We have eight of the 10 largest companies by valuation in the cohort. We have, we said, over a dozen $1 million customers and maybe even more importantly, long term, over 80 $100,000 customers. The signs are there, like other companies are discussing that we're attaching ourselves. Now, why? Datadog, when you, when these are, we can call them AI native, but what are they? They're modern software companies whose whole business was invented in the last five years or so. They have a modernized tech stack and their whole business is delivering through APIs and others to their clients. That makes the delivery of the product mission critical. Because Datadog has invested most of its dollars in servicing the modern side of this, the cloud side, the reliability side, the breadth side, and the speed side, it's a perfect solution. We've always been the leader in the choice, I'd say, in cloud natives. If you want to call these cloud AI natives, it's an extension of that. Mm-hmm. Yeah. As you look at that cohort ahead, this question came to me midstream. Yeah. What are the lessons learned from servicing the cloud native cohorts during the big cycle that we had? Yeah. How do you apply those learnings to monitoring the AI native cohorts? Yeah. What are some of the telltale signs you're looking for as a CFO to make sure that you run a balanced business, not over-index too much? Definitely. No question. We learned a lot of lessons. We have pretty good transparency because we have the meter on it, its consumption. We can see the level of usage and the type of usage. I think what we learned in the cloud native. At one level, you could say there is an AI bubble happening in venture, like you had a cloud bubble happening in 2021. Yeah. Right? I think yes. We should be smarter now in this cycle. Definitely. I think you can see it's a much smaller part of our customer base. Essentially, the impact of whatever may happen, for better or worse, positive or negative, is going to be smaller. You see a workload growth. What probably will happen will be, there'll be some winners and losers. You're going to have some consolidation. You're going to have some companies that are really mission critical and their workloads are going to continue to grow. You're going to see more AI activity in all applications. What we learned was we're here for the long-time relationship with customers, meaning our application for the good part is frictionless. That doesn't mean we can let there be no friction. Sometimes we have to be the friction. We see what's coming in. We are proactive in helping the client use it. They may be sending us too many logs. They may be sending us the wrong logs. What we've always done and learned, in the cloud native, it's really important to have long-term relationships. We're focusing on the length of contract. We're focusing on initiating the increase of commitment where they can get a better price and what it means in the trade-off of commitment and size. We're focusing on our own platform. I think we talked last night that when you think about logs, it's not just logs. It's what are you doing with the logs, which is why we've created flex logs, frozen logs, a number of different things to try to match up the use cases with the SKU. That doesn't mean we're cutting price. That means the intensity of the cloud use of that application is less than real time. Therefore, what we're going to try to do is instead of pricing, and we've already done it, sort of unilaterally, we're going to try to match up on a gross margin basis the costs and the SKU price. That is creating, I would say, more stickiness. It's also creating a broader market use cases and logs that are beyond real-time observability. Those are some of the things I think we learned in the bursting of the bubble that we're applying in this generation. There may be more volatility, but we're going to try to, in my seat as a CFO, try to manage that volatility in a more proactive way. Yeah, I know people try to sort of get at, oh, why did this large AI native customer not grow the business? Mm-hmm. There are all kinds of conspiracies. Yeah. I want to flip that the other way and say. Yeah. The biggest AI native customer, Datadog, what are you doing right for them? What can we learn from that success? Why could that not be a template for other AI natives that may be on the fence? You know, should we do it on our own? Yeah. Look at this big case study, the shining example. Yeah, I think that's a very important lesson when you look at how the. The glass half full versus. The glass half full. Think about it. It's a very good thing that all of these companies are choosing Datadog. They're choosing Datadog because for their use case, it's the best product. Their DevOps teams love it. If you try to take away their Datadog, they protest. It makes their job easier. The time and cost to remediation is dramatically reduced. We've been able to prove over the years and with this cohort that economically, it makes more sense to use the platform than to build it yourself. You have huge investments you have to do yourself. That's why with so many of these cloud natives, we've been able to grow the business and why our gross retentions are so high. The vast, vast majority of customers choose to stay with Datadog and grow their use. I think we have a whole team, business value team, that does nothing other than relentlessly prove this to customers. You can look at it both on the cost side, but you can also look at it on the revenue side. Having, and a lot of these, as you build companies, you have certain accidents or things happen. If that happens, you lose a lot of revenue. We've been able to prove that it's a good decision, net-net, to use the Datadog platform. This is like value engineering, engineers that go in and say. Yep. Okay. This is what. There's value, there's prioritization, there's cost, all of those things. The vast majority of our customers have chosen that way. I don't know if we can get into the large customer if you'd like to and talk about that. I don't know if that's where you want to turn next. The largest, the most customers are not building a Datadog internally. We can't tell what happens. We certainly don't retain every customer. We have a very long track record of keeping upper 90% of customers. We think it makes sense for them to use the platform. David, did you know that I can vibe code my way into a Datadog competitor? I mean, I did not know that. Wow. I can't, the problem is it doesn't scale that well. Yeah. I mean, it does not integrate. It does not have governance. Yeah. It does not have security. It does not have authentication. Definitely. It is not a foldable. I think you just heard that. When I walked in, Dave was talking about that. Yeah. When you think about the difference between a consumer going into a model or chat and all the things that happen in the enterprise where these are your mission-critical applications, you have to balance uptime, putting new functionality in, security, privacy, speed, the platform being used by everybody. It's not at all trivial, which is what's made Datadog what it is. I want to get into some of the growth businesses. Mm-hmm. It's been amazing since you started disclosing. Yeah. Logs, APM, those businesses have grown pretty massively. Yeah. They're approaching $1 billion in revenue. Yeah. Can you talk about what's going on in the APM market and logs? Yeah. I want to get into security just a little bit. Yeah, definitely. You have the observability where we repeat this a lot. We call it all these products, but our clients call it problem solving in the platform. What they are speaking loud and clear is they don't want to go to different point solutions given, you know, the real-time nature of it. I think, as you just mentioned, we've done a really good job of creating a platform that covers metrics, traces, and logs well. We've been able to extend it into a number of the things that affect the application, network, database. Now, we've announced that these products are growing very fast. Synthetics and RUM, what does this mean? You're taking it from the back of the infrastructure all the way out to the device, product analytics, things like that. In the platform itself, service management, we've been able to create additional SKUs that have become significant. On top of that, you have some growth vectors that are tangential, somewhat related, and security is one of them. We have a lot of the data. We have a pretty good real estate of customers. We didn't come from security. What we've been doing is investing in the three areas of security, which would be cloud SIEM, cloud security, which is posture management and vulnerability, and application security. I would say, in the DevSecOps world where they abut very closely, we can attach, and that happens a lot in cloud nativity. What we're doing is moving to the next level, which is essentially, how can we use our infrastructure and our, for instance, logs and create a cloud SIEM product that is able to address the nature of compliance and other use cases besides observability, where we've been very successful. We're starting to see success there. I think we announced that security had gotten over $100 million, which is, you know, an achievement. We have some game plans in product, in marketing, in creating channel relationships, and expertise in sales teams to try to push that. I think we have a great opportunity in cloud SIEM, given where we already are in logs and some of the other things that have been happening with some of our competitors. Also, we have the AI, which we mentioned, the LLM and the GPU, and then service management. I want to address this as a kind of a combo of an observability platform, but extending it, because we generally have been a company that analyzes data, produces clues of where things might be wrong. We haven't been a workflow company. What we're doing, I think we think AI accelerates our opportunity to reinvigorate, reinvest in this, and essentially go from what's wrong to who's going to fix it, and way out there may be auto-remediation. These are some of the areas we're most excited about in sort of growing on top of the observability. That prompts the obvious question, service management. Uh-huh. Is that who you're trying to, I'm not suggesting that you go up against them. Have you uncovered a niche in the market that they're not addressing so well that your product is naturally suited to address because of the adjacency, right? Yeah. What do you see in the ITSM market? Yeah. We had Marc Benioff from Salesforce also talk about we're getting into the ITSM market, right? Yeah. What is Datadog? Yeah. You have to then go below that and figure out who is it. IT or, you know, when you call your corporate IT group, that's not our customer base. What we're doing is the whole thing's platform is basically, we have a real-time use case that emphasizes speed. What we're trying to do, I think you might have looked at Opsgenie and things like PagerDuty. Mm-hmm. We're trying to do it for DevOps and security reliability engineers. I think there are, in this case, you have to look at who the end market is. I think ServiceNow obviously has done a fantastic job in a number of markets, but we're not trying to boil the ocean. We're trying to have this be tightly aligned with our platform to create more value to our customer base. I think in the end, if we're successful, we'll sit alone. Those customers will have ServiceNow for what they do, and they'll have Datadog for what we do. Got it. Got it. Got it. Mm-hmm. I want to talk about some of the newer products. Mm-hmm. We did touch upon this a little bit. Yeah. AI observability, LLM monitoring. Yeah. Database monitoring. What do you see in the opportunity set, and what is your investment philosophy to nurture growth in these nascent markets? Yeah, definitely. When it comes to looking at sort of prioritization, since we put everything on a common platform and about 50% of our investment in R&D is platform, that is a huge birthright, meaning we're more efficient than others in putting out new products because that's sharing in a very large investment in platform. Some of the things that we've been able to do is, as in database, for instance, as the database, you just heard, you just heard from Dave and Mongo, as the database world and the data world has innovated, there's more and more connectivity into the applications and more variety. In that case, it's been really, you cover another database, your revenues increase. Mm-hmm. Because we need to see, our customers need to see everything that affects. I think database has been a really good opportunity for us. Take RAM. Particularly this MongoDB database, right? I mean, what's that MongoDB thing? I don't know what that is, but people are using it. I just heard that, you know, it can't compare, that Postgres, I don't know. I don't know the database world. I know about monitoring it, but, you know, when you think about how this is evolving and this is the same thing as our other integrations, we need to be comprehensive. As it gets more complex, as customers have more choices, that's when you need Datadog. I think at dinner last night, you made this point. Mm-hmm. I just thought of it. Yeah. 50% of the research and development dollars are for the platform. Yeah. Everything is an extension of the platform. It's so underappreciated because you can think of building an APM company and then another product that then you got to extend the breadth of the platform. Yeah. When I first met Ali, it just blew my mind how he got the idea for this company in San Diego decades ago. Yeah. I mean, the view back then when you founded the company, had this idea, was a wide spanning view. Right. I think the pieces are falling in place under that view. Yeah, it turned out that infrastructure was the ideal place. Yeah. To start because everybody needed ubiquity, you got a large canvas. Then, when you think of what others maybe didn't do first, Ali thought of the underlying architecture and the sort of coding of data. If you sit in meetings with him, with our product meetings, you see that he is obsessed with UIs and customer activity. A lot of companies have a great product, but it's so complex to use, and you can't see how to use it. Every time Ali sees that it's not very intuitive and native, that somebody can't pick it up, he challenges. I think he also created a very, very customer-friendly UI with workflows in the platform that could attach really quickly. Those are some of the things when you get the platform that made Datadog. I don't know if it was for me, when I first saw Datadog, it was 2011 or 2012, AWS re:Invent. Uh-huh. I went to a demo booth. I don't know if it was Datadog. Uh-huh. I had a massive monitor with flashing signals all over the place. Yeah. What is this? Yeah. It's so animated, so expressive, and so full of details. Yeah. It's Datadog. Datadog. That was my first product, Datadog. Yeah, that's right. Almost 12 years ago. Subsequently, my other wake-up moment was 2023 Dash in San Francisco. Mm-hmm. We'd love to have Dash in San Francisco. Yeah. You launched a refreshed version of the logs product. Yeah. It just blew my mind. Definitely. The live demo, the LLM monitoring is just, of course, Selena and my team went to the Dash conference in New York City. Yeah. She said, "Dash, look at all the stuff that I picked up. I spoke to all these partners and these customers. Yeah. Just a company that's got a lot of buzz. I think when you think about it, you want to look just methodically at what's going to affect the application. Of course, LLMs will be in applications. You can't cover everything else and not cover the LLM. The goal is to be comprehensive in a single unified platform. Yeah. Dave, what's happening? By the way, at the five-minute mark, if anybody has questions, just raise your hand. We'll get to you. Doing a pulse check. Okay. What's happening on the GTM side? I know that. Yeah. Last year, when you gave guidance for Calendar 2025, we had built in some expense buffer to wrap up the go-to-market engine. Yeah. What is happening there? What is, I know you said you were a little bit behind in hiring, but how much of this is actually proactive? Should we read this as a sign that you're actually, I always, I'm an old-school software guy. Yeah. When companies ramp up hiring and sales, that is a bullish sign. Yeah. When companies ramp up CapEx, that is a bullish sign. Yeah. In software, what am I to make of your signal that you intend to step up your sales and marketing? Definitely, I think your old-timeness is right on. We have a reading of significant uncovered white space and believe that there's a correlation between ramped quota capacity and top line. I think we, on the back end of COVID, got a little conservative. Some of it is we couldn't travel. We couldn't develop the international markets as well. We saw a lot of white space we weren't covering. We have a lot of proof points. This might surprise you, but we had no one on the ground in India. Mm-hmm. Brazil when, you know, COVID ended. We were covering them centrally. I think what we learned is in looking at the white space and looking at the competitive environment, there were huge opportunities in a large number. I would say the Middle East would be an example that we just had nobody. We came to understand that we need the combo of centralized sort of SMB and commercial sales and marketing and feet on the ground. We've been developing those markets, and it's paying off. We're seeing great growth. I think, yes, you're right. It is a bullish signal that we think we can get really good return from increased investment in sales and marketing. Is there any trends in the time to productivity? Mm-hmm. Changes in how quickly people get productive? Yeah. Because Datadog is now institutionalized today. Yeah. Versus two, three years ago, a rep joining the company today ought to be, it's got to be easy. Yeah. Not easy, but less complicated. Yeah, we definitely are focusing on enablement, but it still takes an enterprise, it still takes a year for a rep to get ramped. That is because they have to get educated, but then they have to make their champions, go and make their champions in companies. After that, since we are still somewhat land and expand, we have to get our landing spots and then grow them. I think, you know, potentially we have a longer ramp, but a ramp that you can monitor along the way. I still think you should think about a year to ramp reps. Just to finish up here, any last question? It's your last chance. Okay. Maybe, you know what, I'll do it slightly differently. Do you have a question for me? Yeah. What do you think of the opportunity or risk factor of AI on, in one case, application or SaaS software, and in the other case, infrastructure software? Yeah. Infrastructure obviously more insulated because at the end of the day, it's about compute, networking, storage, and bottlenecks. Yeah. These are things that are homogeneous across tech cycles. Yeah. As far as SaaS is concerned, I liken it to the web browser. In the late 1990s, the web browser became the front end for most things. Enterprise software lagged pretty badly. Uh-huh. Mark Benioff, who's going to be speaking later today, had this idea that we need to put our web browser front end to the boring, drab world of enterprise software. That not only just became the UI for software, but it replaced the front end application layer, right? Mm-hmm. One thing led to another. Mm-hmm. The back end logic of how business does business does not change. Mm-hmm. I think what we saw was a catalysis of the enterprise software industry, the web browser front end. Everybody said, Amazon, Netscape, all these companies are going to destroy enterprise software. Mm-hmm. No, actually they catalyzed. Mm-hmm. It was the birth of companies that changed up the user interaction model, the application code of the front end, and we had a 20 to 25-year run right now as a result. Yeah. When I look at AI, maybe I'm being completely wrongheaded about this. Yeah. AI is the new UI. Yeah. It is going to change the front end of the enterprise software industry, the application industry. Yeah. I see a graceful world where you interact with the software through AI, whatever your prompt engine is, whether it's a foundation model, X, Y, Z, du jour. People, I think, are always ultimately very curious. Mm-hmm. When you enter a prompt, you get an answer back, and you want to find out more. You want to dig in. I want to go to the source. Mm-hmm. That transition from UI. Which is AI. Into the back end of software, the back end of software will also change. Yeah. To accommodate the graceful transition from AI into the software, companies that make the transition graceful and are able to accommodate that business model aspect to them, I think one of the panelists on the VC panel yesterday said it best. Mm-hmm. I think Byron, he said some of the SaaS companies are trading at 5x multiple today, will go to 3x, and some will go to 10x. Yep. Yeah. That is what keeps me super excited. Yeah. There is going to be some massive transformation. Yeah. It's not going to be the same, but there's a lot of money to be made. I want to thank you once again for your partnership. Thank you. You've been tremendous. I really love these discussions. Yeah. We can. Thank you very much. Thanks for having us. Absolutely. Thank you. Bye. Bye. Thanks. Thanks, everybody. Have you had?