Okay, hi everybody. Let's get started. Let's have a fun day three today. I'm actually excited with at least the lineup that, that I've got. I was only half-joking with CJ when I met him this morning that having hosted him at previous conferences, in addition to him being a very, very capable tech sector executive, he's also one of the more fashionable executives that I know. So when I was pulling my wardrobe together this morning, I knew I had to totally raise my game. So this is the best I could do. I think CJ wins out.
Oh, yeah. Thank you. Thank you, Carl.
CJ, Mike, and Ben, thanks so much for coming to our conference. Your presence always makes it much better. You're also coming off of a wonderful print, which evidently most investors appreciated, so Mike will talk a little bit about the financials in a moment, but CJ, welcome to MongoDB. Welcome back to Scottsdale.
Thank you.
And maybe where I wanted to start is a little bit about, you know, your early vision for Mongo. I know you've only been here probably 30 days or something.
Yeah. That's right.
But, I guess I was struck by your comment on the call that you envisioned MongoDB as being a modern data platform. When I hear that, it sounds like you want Mongo to be, you know, more than an amazing document store database. And so maybe you could lay out that vision even though it's quite early.
Absolutely. So, you know, at the highest levels, Mongo participates in a large market, which is always a good thing that it participates in a large market. And this market, you know, my career started at Oracle Corporation.
Mm-hmm.
And.
You know a few things about databases.
I do know a few things about databases. And I stayed with Oracle for 7+ years, when Oracle was a high organic growth company in late 1990s. And so database as a technology has been around for a long time. And I would say 1970s, 1980s, and 1990s, 1970s is when Oracle started, was very structured data, rows and columns, and that's what we obsessed about in terms of performance, and it was always scale-up architecture. And, you know, looking at MongoDB, and I've been a customer of MongoDB, at ServiceNow, we did a lot of technical integrations with MongoDB, discovering MongoDB, in cloud or on-prem and so on. And so I have a lot of respect, first of all, on what the founders created in the first decade of this century.
Mm-hmm.
Enterprises, Carl, you know, when I speak to customers, enterprises everywhere, they have so much semi-structured and unstructured data. There are so many workloads nowadays with, as enterprises are looking at AI, and everybody's looking at it. I would say they're piloting. I was with a large financial services company last week, and they said we have 45-50 agents that we are piloting, which is an important term, whether it's about maybe three or four and in production, employee-facing, not really in a meaningful way customer-facing. When I see all that, the opportunity for MongoDB to go from a great database, document database with native JSON support and others, to become a truly data platform, that opportunity excited me.
Mm-hmm.
And that's why I joined. And one last thing I would say is that product-to-platform transition that I've done at previous companies, MongoDB has all those ingredients on when customers (and I've already spoken to 30+ customers in 30 days) when they look at, when I speak to them, three things they tell me. One, the cloud migration to a GCP or any hyperscaler you pick, that's still going on, and it will still go on for four to five years. As part of that, there is some, of course, lift and shift, but there is also, hey, we are gonna modernize these workloads. And given how much unstructured data they have, MongoDB has the right position to it.
Mm-hmm.
Second thing is AI workloads, how do we clean up the data? Of course, everybody's going to these customers saying, "We have agentic platform built on us. We have this. We have that." But customers are like, "If I'm going to fundamentally change my business or make it truly efficient business, I want something at scale." And to have real-time data store with the right platform is the high ground where MongoDB is positioned. And the third last thing that I would say is that the data estate still continues to grow. So you have confluence of this cloud factor, AI factor, and just core.
Yeah.
Data, and that's what excited me, and that's why I joined.
Yeah.
MongoDB.
And you've also got an amazing opportunity to leverage the developer mindshare that Mongo has built over the years. I followed your story from way before, when Dave became CEO, and.
Yeah.
One of the tenets of Mongo has always been that developers love the product. So that's an amazing base to work off of.
Yeah.
Yeah. CJ, let's talk a little bit about some of the broad themes out there. I think, I'm certainly not the only one noticing that across the software space, there are pockets where we're not really seeing any growth acceleration. Frankly, it's flat to down. But one area where, there's consistency and it feels like the whole category is inflecting is, interestingly in the data space where Mongo's accelerating, Snowflake. We had Ali at Databricks on stage yesterday. Palantir, so the whole data's category in software is accelerating right now. Can you just pinpoint the one or two things that are causing that? Certainly, when I talk to enterprises, I consistently hear this theme about needing to better utilize their corporate data to make it AI-ready. But something broad is clearly happening. And could you perhaps help the audience understand exactly what that is?
So I would say your observations are correct that infrastructure software related to data definitely is benefiting right now. And there are still questions in enterprises' mind, "Hey, what do we do with SaaS?" and, you know, "What is the long-term vision for SaaS?" But you contrast that with data, though. Right now, you, you mentioned some data warehouse companies. Data warehouse companies always were very good at, "Hey, I can ask these kind of analytical questions. My top 10 trends in the holiday weekend, just give me those answers." And then, for data scientists and AI folks in those enterprises, data warehouse technologies are helping them out with those questions. From my perspective, so that's one piece that trying to get more value out of the data you have to make the business decisions.
Now companies are focused on it rather than in 2022, 2023, you saw, Carl, there was a lot of efforts on optimization. Do I really want to do this right now? I have other supply chain and issues. If you're a manufacturing company or if you're a financial services company, you had other concerns related to inflation and so on, you contrast that on the OLTP side. On the OLTP, now people are saying, "Okay, we have, we have time to really modernize the application stack," as you saw in our results. They speak for themselves. What we discussed, Mike and I, two days ago is that where we are seeing is, of course, the high end of the enterprise where the consumption, that we reported for 3Q, but also broad brush strength in Europe as well.
Mm-hmm.
And then our self-service motion, coming back to your developer point, we are seeing a nice customer growth to say, "Oh, I finally have time to create a new application or modernize a new application.
Yeah.
And that's both, working in our favor. So you have data warehouse, the analytics phenomenon.
Mm-hmm.
You have the OLTP with us, a phenomenon that's helping us.
I'm not asking for guidance. That'll come in three months. But maybe I'll phrase it this way. These drivers you're talking about, CJ, how durable are they? Is there anything that you can share with us? Maybe it's anecdotally, backlog color to make the point that these drivers can persist next year, year after. Ultimately, what I'm getting at is what's the duration of this data investment cycle you appear to be benefiting from today?
I would say we are always, you know, Mike and I even spoke, at this, about this thing that how do we think about the future and so on on consumption. What we saw is that we don't want to be at the highest level, say, "This will continue forever," right? I mean, because we just don't know what we don't know.
Yeah.
You saw what happened in 2022 and 2023. From my perspective, when I see speaking to customers as they think about the IT budgets in next fiscal year, there is nothing that they are giving me indication that the budgets are gonna shrink like what you saw in.
I'm not hearing that either.
Yeah. In 2022, 2023. They are telling me that AI is still gonna be a priority, right? And they still have to show the ROI. I'm talking about CIOs, CTOs, and so on. And I think from my perspective, we just have to, of course, execute on our side in serving the customers at the highest level and continue to provide, right guidance that Mike is now on top of.
Yeah.
As we move forward. Mike, would you add anything?
So thank you, CJ. I would just do one thing on that, Carl, is that's the part of the industry, and we get a lot of questions. Keep in mind the changes we've made internally to address those industry trends. We feel really good about the, call it the durability of that. We've moved resources upmarket. We still have a huge amount to go get in that, call it in that Fortune 500 or 1,000. So we feel like there's still a ton of room to run there.
Yeah.
As we continue to shift resources. So the industry trends are great. We also feel good about our change in resources to go attack that market.
Okay.
Let's talk a little bit about more specifically the effect of AI on your results. My impression from listening to you the other night is that you put up that stronger-than-expected result, but not because of some direct AI lift. I think the way you were describing it was that it was a core strength, actually, which makes Mongo interesting to a lot of people in the audience because if you've put up that strong performance in the core even before that direct AI lift comes, well, that's interesting 'cause then you've got a potential growth catalyst sitting in front of you. So maybe this is for you, Mike. Can you draw that distinction in terms of where the strength came from core versus AI most recently?
Sure. So to that point, and we've said it for the last couple quarters, we continue to see really good traction in AI natives.
Mm-hmm.
We see the work going on in the large corporations related to AI, but it has not been a material driver to our results.
Okay.
It was not in Q3 either. The growth and consumption that we saw in the drivers, call it in the core, it's been the large customers who are doing mission-critical workloads for their business. So if it's an insurance company, claims processing, if it's a bank, it may be check depositing. It's core workloads that we've seen. Do we see a lot of activity in customers in terms of testing? Sure. But that's not driving a lot of revenue. So we do think that that is a future driver. Again, we believe it is not, it is not if, it is when, but it is not driving the results today, nor is it a big part of the guidance that we did for Q4.
Is the catalyst, Mike, that we're waiting for, as simple as enterprises like UBS start writing more robust enterprise-grade AI applications that create a, a pull-through for database that maybe today a lot of the AI applications are somewhat light, lightweight, and they're not super database-heavy, but sort of phase two might be that catalyst? Is that conceptually the right way to think about it?
I think that's the right way, and they may be doing inferences. They're not creating a lot of data internally.
Yeah.
But once they actually do it with their internal data, that will create a lot more data, and we feel that will be the pull-through.
Got it.
Yeah. And can I add just something to what Mike said? Agreed. Here is why I'm personally paying attention, being based both half and half in the San Francisco area as well as New York, is we are learning a lot from AI-native companies because AI-native companies today, right? There are so many. And venture capitalists, who are on the sidelines, as you remember in 2022 and 2023, started funding in 2024, whether it's a vertical-specific AI company or a foundational model company and so on. And my observation, even when I was doing diligence on MongoDB and after joining MongoDB and speaking to them, Carl, is that they are saying, "Okay, to truly create a killer AI app," which AI-native companies, that's their aspiration, right? "To do so, what does MongoDB really have?
Can we build on top of MongoDB the next-generation killer AI company?" And so I asked them questions, "Hey, are you using our vector search functionality? If yes, tell me why. If not, why not?" As you know that, Dave and the team bought Voyage, and have you thought about our embedding models and re-ranking models that improves your search accuracy and so on? And they're like, "Oh, I didn't know that. Let me try that out. That's an easy few lines of code change that I can do." The third most important thing is scaling. If I become successful, because everything that becomes big starts small in AI-native companies.
Mm-hmm.
And some of them are getting traction, as you know. And what I see with them is, yes, we are actually growing a lot, CJ, and we use the example of Mercor, in the earnings script. And because we are growing and we are scaling, MongoDB is the right OLTP database for us. That learning from the AI-native companies, we want to take that to the enterprise to say.
Yeah.
Hey, because the AI team, Carl, when I talk to these large enterprises, what I find is that the AI team, as in the agentic team, is a separate team from the core database team.
Mm-hmm.
And the core workloads team.
Okay.
They are dependent on those workloads team to get the data via APIs and so on.
Yeah.
And they wanna move fast because they are like, 'Hey, I have a lot of pressure to experiment on this agent X or agent Y.' And then when I tell them, 'Hey, do you know that this foundational model company is using MongoDB in this way or this? Mercor is using it, and this is why they are using it?' They're like, 'Aha, thank you for telling us because we, we just wanted to move fast.' So we didn't really make an explicit decision.
Got it. CJ, let me ask you about a subject that's, somewhat, relevant in the markets today. The markets are a bit heavy today, I think, on additional media stories about the pace of enterprise agentic adoption. I think Dave and you more recently have been on the more reasonable, I'd say accurate, actually, end of the spectrum on that debate, hyping it versus being quite reasonable. Do you wanna weigh in with your view? You talk to a lot of customers, not just since you've joined Mongo, but prior. Where do you think enterprises are in taking truly agentic apps out of pilot?
Yeah.
Actually into production? And Mike said, "When, not if." Can you give your best guess as to when the when is?
I don't like to do that, Carl, but I would, I would say in speaking to customers, this is not days and weeks, right? This is not days and weeks. It's.
Yeah.
Few quarters, depending on, because enterprises, if they want to transform their business.
Yeah.
Which is a very important point, using an agent, whether it's single-purpose agent or multi-purpose agent, you have scale, durability, availability, real-time, learning, context switching for the agent versus human. There are a lot of factors to be considered. And yes, I created a pilot, and it does something cute internally for productivity that I can write email fast or I can do this Copilot or CodeGen. But to truly create an agent that transforms your business, right, or creates a new business for you, my viewpoint is still few quarters away. Like, I've been speaking to customers, as you know, for a long time, nobody has come and said, "CJ, this completely transformed my business, today." And we are piloting, tinkering. I heard this financial services firm I talked to, they said we have twenties of agents.
Yeah.
But they're still doing this one task.
Okay.
But nothing at scale.
Got it. Let's switch the subject maybe to competition in your space. There's always been intense competition in the database space for as long as I can remember. Still.
Yeah.
The case.
Yeah.
But in your defense, what I would say is that, 10 years ago, you were one of probably five or 10, at the time, NoSQL database vendors, and you've actually been the one to emerge from the pack while a lot of them have still stuck at 100, 200 million or sold themselves. So you've won that race. But today, you're still looking at some competition. So let's go through three categories, and then maybe you wanna weigh in as well. One is, a lot of us have been hearing about the popularity of Postgres relational databases where Postgres running on Azure and AWS have partly, maybe not entirely, solved some of the scaling issues that used to be a blocker years ago. Ben, do you wanna take this one? When you contrast Mongo against Postgres, where do you live in this debate?
Is it sort of room for two because the overall demand is so strong, or do you feel like you've got an opportunity to take some share from Postgres?
I definitely think we have a lot of opportunity, but let's just go back to what Postgres is because it's not just one. It's 30, 40.
Yeah.
Different flavors of this, right? And I talk to customers all the time who started out with Postgres and AWS or GC, name your hyperscaler, and now for whatever reason, they wanna go to a different hyperscaler, they wanna bring it back on-prem. You can't just move that application. It's actually just 'cause it's Postgres doesn't mean it's truly portable.
Yeah.
So one thing that really especially resonates with enterprises is that MongoDB actually has a truly portable database, right, without any application change. So we can move your data from one cloud to another, bring it internally, go back to cloud, no application change. And so especially enterprises, Europe is big with sovereign cloud. We're seeing a lot of this portability be a massive driver of a requirement of, like, what data platform they're gonna choose. So I would say that's number one. And then number two is, and you touched on this earlier about the developer experience that we started out with that, you know, we really still to this day focus on is that developer experience, that developer velocity, that efficiency.
And so if you look at what we've been doing over the last few years building up to this, you know, AI era, is we haven't been bolting on other workarounds with different API layers to add new use cases. It's all part of our, you know, idiomatic API that we've always invested in, which is, you know, MQL. So we added search, part of the same query language. We added vector, same query language. We're keeping that developer velocity and efficiency in mind at all times, and that's just something that the competitors are not doing.
Got it. The second category, which we were thinking about yesterday because we had Ali at Databricks on stage, is some of the OLAP vendors are stepping into the OLTP database space, and Ali was quite proud of their new Lakebase offering, so when you think about those guys, the likes of Databricks and Snow, taking early gentle steps into the operational database space, what do you think of that?
I think it's cute. Like, look, like, they tried for many years to build an organic OLTP, right, 'cause they saw very early on that OLAP wasn't gonna be fit for purpose for these real-time production applications that are gonna be business mission-critical, as CJ was just kinda pointing out, and why all these applications are in pilot. So they realized that they couldn't do it organically, so now they bought someone, but they didn't buy like the A players, and so we're, from a look at the OLTP landscape, I think we're 15 years ahead of where we need to be from an OLTP landscape for these AI applications.
And I'll add, Carl, one thing because it's a question that comes up a lot, and you know, we have great investors here. Like, make no mistake, even when I was at Oracle, Oracle was known for OLTP, and we experimented a lot with data warehouses. I'm talking late 1990s, early 2000, and then you know what happened with Business Objects and all those firms back then too. And never OLTP and OLAP were always two sides.
Yeah.
Within an enterprise, different buyers, and based on the use case, what you're solving for, and from my perspective, when somebody that is just focused on data warehouses starts buying OLTP-based companies, it's a validation of that our space is great and.
Yeah.
It has durable advantage, and they are trying to expand their team. And then, you know, I'll just add to what Ben said, that even Postgres adding support for JSON, and when you actually really look at that, why they did that is JSON is popular with developers. Why? Because of MongoDB, and that's a bolt-on versus we just naturally work with that. And for AI high ground, I would argue that us having native JSON support and just how MongoDB works for AI even if you look at any of your popular ChatGPT today, they talk about images, video, this, and that. There's so much unstructured data.
Yeah.
We are correctly positioned for that.
The third and last one is the hyperscalers. They're native databases. None of us can leave a meeting with Microsoft without Satya and company bragging about Cosmos DB and how customers are coming to them for their native databases. Has anything changed in terms of competitiveness versus the hyperscalers, Ben, CJ?
From my personal view of what I'm talking to customers about, it usually starts out with, "It's really easy. I click it. It's in the console. It's available." So they just start there, right? And even, you know, all the hyperscalers claim compatibility with Mongo. You know, when we run our own compatibility test, we come out a lot less than that. And the way we do that, we actually publish the benchmarks publicly.
Yeah.
So it's available on GitHub that everyone can download the same test and do them themselves. We have thousands of database schematics that you can, you know, query against. When they do their benchmarking, we think it's sub-100 of the actual features of the MongoDB API to give them a better percentage of compatibility.
Yeah.
When we take a conservative view of, let's say, 270 of the most critical, mostly used, pipelines, we are into the 40%-50%. So even when we give them a little bit extra.
Yeah.
So I just think that people start there 'cause it's easy, similar to why people start with Postgres. It's just like easy.
Yeah.
And so we eventually get it. I wanna get it earlier, don't get me wrong, but we eventually get it.
On the high end of enterprise on that point, besides the features and functionality and scale-out and this and that, recently, I mean, if you look at 2025, you had GCP outage, you had AWS outage, and you had Azure outage.
Mm-hmm.
Speaking to customers, including large financial services as well as healthcare companies, just recently, in the last couple of weeks, as Ben said, that "Hey, on one hand, it's easier for me to turn on and use the first-party database." Resiliency perspective, now you're locking in yourself. So if one of the hyperscalers is out, your workload is out. And MongoDB, on Ben's first point, that I can, you know, I was talking to this large customer, and they said, "From resiliency perspective, CJ, we chose MongoDB rather than the first-party hyperscaler because I can now replicate MongoDB from AWS to Azure, without a problem." So not just in intra-region of AWS, US East, one end, so on, but I can do that across clouds.
Yeah.
That is a massive advantage compared to the first-party clouds.
Okay. Hey, Mike, let's turn the conversation to you for a moment. Congratulations on the October quarter results.
Thank you. From all the employees at Mongo.
I think one thing that stood out to me, you know, in my judgment, the 3Q results were very good, but what I thought was really outstanding was actually your fourth quarter outlook, where you raised it quite substantially.
Mm-hmm.
If, you know, I know you, you wouldn't encourage us to do this, but if we assume sort of a normal-ish beat on that guidance, you're getting a fairly marked acceleration in your total revenue growth rate in the January quarter. Can you unpack one or two things, Mike, that gave you that confidence for that acceleration in the January quarter? 'Cause to me, that was the highlight of the print.
Sure. So let's break it up into Atlas and then non-Atlas if we could. So for Atlas, as we looked at the second half, when we got at Q3, we were confident, we, call it around that mid-20% growth rate. What we saw coming out of Q3 was, and we said, "Hey, it was a, it was very much what we thought, and it was very consistent throughout the quarter." The nice part is when that consumption builds, then that's the starting point for Q4.
Right.
So when we forecasted Q4, then it's a higher base, and we felt better about raising that to the 27% with the caveat, which is, "Hey folks, it's the holiday season. Happy holidays to everybody." There are. We've always seen a little bit of unpredictability based on where days fall. So we just wanna take that into account. The big thing for us was around the EA business. And from that business, not only did we see a little bit better for multi-year deals, and I'll say this publicly, folks, this is not pilots. This is regular business that was coming due in the second half where it's tough to forecast, are they gonna do a one-year deal or are they gonna do three, possibly even five?
The great part about that is these are the largest customers at Mongo, and we love them all, committing to us long-term, and we saw much better commitment not only in the multi-year, but just in, call it the run rate business.
Okay.
So when we built that bottoms-up pipeline, we felt better about that. So those were, if you look at the Q4 raise, about a third of that was Atlas.
Mm-hmm.
About two-thirds of that was EA.
Mike, the other thing that stood out to me was that since you came aboard, I'd say on the margin, you've been giving a little bit of color about how to model Atlas, but more directional. You went one step further on this call where you gave a point guidance for Atlas, which I think everybody appreciates. It leaves out a little bit of the guessing game, how to apportion total revenues. But one interpretation of that is that you wouldn't be making that disclosure improvement, if I could put it that way, unless you felt fairly confident about the Atlas business. Was that the correct interpretation, Mike?
I never wanna give a number externally that I don't feel good about.
Okay.
And what we would say is we debate this all the time, and you can look at ranges. We felt really good about that number in terms of, "Hey, from a pragmatic point of view, that's where we think we are. Hopefully, things go better." Hopefully, they do, but that we felt good about giving that number.
The last thing, I wouldn't even call it a flaw because I think you can explain it, but one of the only metrics that stood out to a couple of investors that hit me, Mike, was that the direct customer count fell slightly for the second quarter in a row. But I think there's an explanation for that.
Yeah.
But it might be good to just give you a chance to explain that one.
Yeah. So thank you about that, Carl, and we'll probably change this going into next year. Keep in mind that when we talk about direct versus self-serve, this is how we internally categorize those customers. It's self-serve, not they started in self-serve, and then they stay there forever. At some point, they will shift to a direct customer. Sales will take over. So what you have is a little bit of the apples and oranges. As we have added more sales to go direct, we've pushed that team up.
Mm-hmm.
Focus a little bit higher. And then what happens is self-serve covers that gap. So it looks like direct is actually going down. Folks, look at the total customer count. That's what matters. The bifurcation between direct and self-serve is our own internal categorization of how we go to market. And just, hey, little bit of warning. In 2027, folks, we were probably not gonna give this number because of this exact issue.
Got it.
Look at the business as a whole. Look at consumption growth. Look at revenue growth and total customers. How we bifurcate internally is how we go to market. It's not a driver of the business.
8,000 customers added this year on year to date with 67% growth. Even adding 2,500 in the quarter specifically, which if you look at last Q3, that's a 40% growth.
Yeah.
On just new customer adds.
Got it.
Carl, I mean, you know how much obsessed I was about new logos and others in the past. These are very good leading indicators for the future.
CJ, Mike, and Ben, thanks so much for coming to the event. I think along with everybody here, looking forward to watching the Mongo story unfold over the next 12 months.
Thank you.
Thank you so much.
Thank you for having us.
Thank you.
Thank you.
Thanks, CJ.