MongoDB, Inc. (MDB) September 2026 Conference Transcript

Well, pleased to welcome our next, I can't really call you a presenting company. I think anyone here remembers the days of the presentations? Used to give a little canned presentation at these.

Oh, that was a long time ago.

A long time ago with the breakout after. But our next fireside chat, MongoDB, with me to my immediate left, longtime friend Mike Berry, who's the CFO. And then Ben is the CPO in charge of products. So we'll get into that a little bit as well. Didn't get this after my lab, excuse me.

And thanks for having us, Rob. Absolutely.

So I think the obvious question, mix of business and how we should think about it as investors. It seems to be a hyper-focus around Atlas and what's happening with Atlas, but the customers are making a choice. And the EA business has been far better, I think, than you would have thought. So maybe just walk us through some of those different dynamics between EA and Atlas at this point and what's driving each and how investors should look at the business.

So, again, thank you for having us. And you hit on it. This is all us meeting customers where they are. And what we've seen over the last, call it 12 to 18 months, is just a renewed interest in the self-managed part of the world. Do we invest and focus and obsess completely over Atlas? Absolutely. It's super important. In addition, what we've seen is really a pickup in growth in EA. It's always been a good, stable business, specifically around regulated industries, governments. And I think what you've seen is, I'll say, two areas, and then one reason why we like it so much is, is, hey, the run anywhere strategy is real. And for us to be able to run not only across the three public clouds, but also anywhere they want to deploy is super important for them. And then now you bring all the additional regulations. You bring AI into it. Can they even deploy in a cloud given the capacity constraints? Economics matter. So we've seen not only, I'll say, the core workloads where certain industries are going to run on-prem anyway, but now you've seen all the new things that they have to deal with, and that's really driven the growth in EA. And then for us, again, is Atlas important? Absolutely, and we obsess over it every day. But we also love EA because they're our largest customers. They will typically, the larger customers, run both self-managed and Atlas. And in that cohort, we talked about this at a conference last week, When you look at the large customers that have self-managed and also Atlas, Atlas growth is actually higher than the total company average. So there's this perception, I think, Rob, which is, hey, it cannibalizes it. Our view is that simply hasn't been the case. And from our perspective, it also drives a boatload of profitability as well. The other thing is keep in mind we have disclosed ARR for EA because of the revenue recognition volatility. Keep in mind that that's EA and other. Other is almost always just going to be OEM transactions. And this isn't the large Asia-Pac deal we've done before, but you'll see a little bit of that variability driven by other. If you look over the last three quarters, EA ARR has accelerated each quarter, and to the extent now that we actually guided double digits. I'm sure we'll talk about this in two weeks at our analyst day as well. but it's certainly now turning into a double-digit growth driver, and we feel really good about it.

Mike, is there any common characteristics of those going with EA? You mentioned large customers, but if you look internationally, are you seeing a push back to EA for nationalism or kind of running an on-prem versus being in the cloud and having risk of data? Is there any kind of common denominator you can speak to at this point? You want to take that?

Yeah, there's not really a common denominator. I just want to correct one thing that you said about going back to EA. Customers aren't looking at it that way, although that is a security blanket for them. A lot of these laws and regulations and new certifications and everything else, they're still trying to figure these out. Some of them are talked about but not even cemented into law yet. So customers like the flexibility and portability that we're able to give them that, quite frankly, not a lot of other technologies can give them. And that's the fact that you build an application on MongoDB. You can start out in one place, move it to another place, bring it on-prem if you absolutely need to and self-manage. Or quite frankly, we have some customers that operate in some countries or some regions that don't have a cloud provider. So to give them a truly portable experience is really valuable to them. So that message of run anywhere and the portability of the application is really resonating.

But investors shouldn't have the view that it's a legacy type of customer or a legacy type of implementation.

Certainly not. And that's why, what was it, five months ago now, four months ago now, when we announced the GA-ing of search and vector on to the self-managed. We pioneered that in Atlas, and we had enough demand. This was all driven by customer demand. It wasn't a bet. Let's bring this to the self-managed world because they wanted to be able to have the flexibility of going both places, but also demand for the self-managed environment workloads that just will never go to the cloud for whatever reason. They want the same functionality.

And if I could, Rob, is Atlas important? Absolutely. And we focus on it a lot. I think there's a perception, which is we will only be an AI winner if Atlas accelerates. And we struggle with that greatly because that's not what our customers are saying. They're saying we want to deploy, run anywhere. We might want to self-manage that. And I think going forward, it's going to be really interesting. Hey, economics matter. If they need to absorb the incremental, call it AI spend, how are they going to deploy that within the envelopes they all have? And that's why I think that part of the self-manage becomes a bigger piece of that. And we'll see how that plays out.

So I think this brings about two questions, and I'm sure I'm leading you into your analyst day. So your answers will be somewhat gauged and guarded, Mike, if I know you. But relative to the durability then of EA and your perspective of how that has changed. My sense is this is going to be a more durable growth driver possibly than it was before.

We talked about that on the last earnings call. We haven't guided past this year, but we said we feel like we now have two growth engines, Atlas and Self-Managed.

And number two, just in terms of the customer economics between the two and monetization, the big deltas, obviously it's more profitable for you to sell a license to someone and they run it themselves and have that cost. But maybe walk us through the monetization opportunity on each and what the experience has been.

Sure. So, and I think this goes a little bit to the lifetime value of that. Up front, yes, you get the profitability. Over time, that license model will typically be a little bit more profitable, call it gross margin. But when you go all the way down, keep in mind too, you still have to spend R&D, other dollars. So for us, we are excited about each dollar. The important part for us is that if somebody is doing self-manage and then they're running Atlas, that for us is great because over time, will stuff get deployed in Atlas and migrate a cloud? Maybe. Are they going to deploy in that new workload? Hopefully they do that in Atlas. So for us, there's also that synergy that comes with it. And keep in mind, and Ben can talk more about this, it's the same product. From an R&D perspective, there's very little difference. So it's not like we have to go spool up a ton of R&D. There's a little bit for the features, but it's also very efficient and it serves both products.

So in terms of metrics that we should be looking at then, how relevant is CRPO to a leading indicator? It obviously spiked last quarter, but I know you give us an ARR metric and usually we're going to have to triangulate between three, but given your hybrid model, is CRPO a relevant metric for us to look at? Or too much noise?

I don't think it's too much noise. I think it more goes to the go-to-market motion than anything else. In terms of us, as we move up market into the enterprise, they like to have no longer commitments. There will be requests that they have. There will be requests that we have. And what you saw in Q4 of last year was just a much more growth in commitment. It was both across self-managed and Atlas. And when you move up into the enterprise, that will be a part of the go-to-market motion. And keep in mind, people say, well, I want to correlate CRPO growth with revenue growth. If RPO is, call it $1.4 billion, and current is about half of that, $700 million, we got it somewhere around $3 billion. So it's a part of the revenue stream. I would also ask investors, don't assume that's all incremental because it's not. You'll have existing customers that will renew a commitment. They'll do a long-term commitment, and then we'll put that in RPO as we need to. Also keep in mind, our RPO is everything above a year. We don't estimate below a year, so it's going to be a part of it. Is it important? Yes. Is it a leading indicator of growth? It's a part of it. It's not the biggest stride.

Ben, where do you think the most opportunity is within the core product set now? And when we think about the next incremental dollar of R&D, what are you putting it towards?

Yeah, good question. I think there's a couple of things. I mean, one, we announced some new capabilities about four weeks ago around, you know, MCP server, which is essentially call it the SDK or the interface that allows agents to talk a singular language to a bunch of different things behind it. So think Terraform or CDK that's been out there for a really long time. Now we have a better pathway for agents to communicate with our services. We're going to continue investing in there because that ultimately helps drive what we call unaided AI awareness, which means that the human's not potentially in the loop of those decisions and those recommendation engines from TerraLabs. So I think that's one, obviously, big area. I think number two is as we're talking about self-managed, we've pioneered a lot of different services in Atlas first because of how fast we can deploy into that environment that we fully control. We're going to constantly, as now Search and Vector are now part of the self-managed ecosystem, what other things in the Atlas experience that we can continue to make that self-managed experience just as capable. We'll always be looking at that, too. So there's a lot of work going into that. So more information to come later, but that's my focus area right now.

And CJ has spoken a lot about his vision to evolve Mongo into a data platform company. So from your perspectives, what does that entail from a core product?

Yeah, I think a couple of things. One, the way I think of it, and I've been at Mongo for nine years, the original thesis behind MongoDB and Atlas was we already have, MongoDB is a good OLTP database. I think it's the best, but I'm biased, of course. And we have the system of record for all of the customers' revenue-generating applications. We want to provide the best developer experience to that real-time data as much as possible. So if you look at what we've done with Atlas, we're not, unlike some competitors we have out there, we want to not make the customer have to duplicate their data just for another use case or just onboard another service. everything that we've ever done inside of Atlas is, okay, we've added search. You just turn that on. It's against that same data set. You want vectors. You just turn it on. We have Atlas Data Federation. We have Online Archive. We have all of these different robust services that now exist inside of the Atlas platform. And why we think it's now Intelligent Data Platform is the add-on of our Voyage acquisition to turn on auto embeddings. all of this is encapsulated into that singular view of what the customer is already operating in. They're not copying their data outside of that boundary. They're not duplicating that data for another use case. So as we continue investing into that platform, we're going to continue that same thesis.

Okay, great. And can you expand around where the average enterprise is in terms of deploying custom AI use cases into production right now and just maybe get a view of the last six months, how that's changing and where people are being held back to the extent they are right now?

Yeah, I think it's certainly changing from the fact that every enterprise is piloting, right? And I think the stakes are just different between what you do in OLAP, which is a lot of internal use cases, data warehousing use cases, reporting and analytical use cases versus the revenue generating applications that are just natively the OLTP database. The stakes are just different, you know, the difference of someone being mad versus the headline of a Wall Street Journal article. And so, you know, while all those pilots are still occurring, I think it's mainly security, governance, laws, regulation, privacy, GDPR, all of these things combined are making the enterprises, especially the regulated industry enterprises, just move slow. And it reminds me a lot of, you know, early on cloud adoption when the big three hyperscalers first started coming out where I was working for an automation company at the time. And one of our biggest drivers of growth was go discover where all of these people are using cloud or they're not supposed to. And there was an idea that we have to be insular and keep it all in before we figure out how to like truly control it. And every article or essay from a CEO that comes out over a weekend makes people scared and be more conservative about like, how do I want to control access to these things? And if I'm going to use these things, are we going to let them use it against our mission critical revenue generating data? So I think it's just a conservative approach and security and governance as that stuff gets figured out, then more enterprises will start using it in production.

That's great. And then there's been some early examples of Mongo being used as a memory layer for agents. Can you contextualize this opportunity a little bit more for us, and what's it mean for Mongo to be at the memory layer, per se?

So we're excited about this because it's a natural extension for MongoDB. So agents speak, from an input and an output perspective, they speak JSON. And what better place to store JSON inputs and outputs? Well, MongoDB, we're the only native JSON database ever built from the ground up. So this happened organically with some customers where they were like, hey, this is the best place to stick this. Let's go, we're going to use this. We already have MongoDB. We're going to just use this as our memory storage layer. And we see customers have. We've invested in the partnership with, like, LaneChain. Being in part of that framework, that all has to do with being part of that agentic ecosystem of where our customers are developing their applications.

Maybe address the top of the funnel issues that you guys have talked about previously, and just in terms of Postgres, how you become the default, right out of the gate versus, gee, this isn't a technology that scales Mongo is? How are you changing perception? How are you getting it at the top of the funnel?

Yeah, so the first thing is I talked a little bit ago about MCP. That is part of that awareness activity. What we announced four weeks ago is now that our MCP server is part of all of the different connector libraries that exist inside of the Frontier Labs and Vercel V0 announcement, which was last week. It's all about meeting the developer where they are, but also allowing their interface of what the developer is working on to seamlessly have access and provision resources both on MongoDB and inside of Atlas. That's certainly going to help from an awareness aspect of it. The second thing we're doing is we have to influence the training of rounds that happen when these new models get trained. And that's by making our documentation more AI-friendly, our website, having new example applications in open GitHubs. So all these models get trained on actual code and development against MongoDB. All of that is to help influence those models. But it's lagging, right? There's not a new model that comes out every day. And so when we see a new model, we play with it, we test it, make some tweaks. And then it's multiple days, sometimes weeks, until that next training round, and then that model gets published. So it's a slow game, but it's obviously an area that we're paying attention to and investing in.

And related to that, we still are focused on developer awareness. That's a big piece. We talked about Reclaim the Bay, which we've now deployed, call it worldwide. So it's both, hey, we want agents to love us. We also want to make sure that developers continue to love us. And so it's a two-pronged approach.

And in true Mike Berry fashion, since you joined, operating margins are up five points, free cash flow margin up 15 points. So not surprising to someone who's known you as long as I have. But what changes have you affected internally to get there? Obviously, scale helps. But and how do we think about this business moving forward in terms of your your typical enterprise company, if you will, Mike, in that journey to higher levels of margins without giving too much away from your upcoming analyst day?

Yeah. So thanks for the question. So one of the things that when I joined Mongo that Dave and I talked about was I think there was a perception, which is if you're going to grow revenue, you can't grow profitability. And I just fundamentally disagreed with that. And I give kudos to all 5,700 employees at MongoDB because it takes a village to do this. So scale is the number one driver. What we did in 25 was we really took a step back and said, let's make sure we know exactly where we're spending our dollars. And it's easy to spend money. It's a lot harder to say, am I getting a return on it? We did a small restructuring in sales because we had added some capacity that wasn't driving productivity. And then that allowed us then to take a step back going into fiscal 26 to say, okay, I'm sorry, that was 26, into 27. Now we can start to spend money to drive innovation. And you see in the guidance, I mean, R&D dollars are up about 30%. We're investing a lot in that. Sales and marketing more like the mid-teens, and then lower single digits for G&A. That has allowed us to expand margin. and going forward, what we've talked about is you'll continue to see us invest in R&D, probably at a lower percentage growth rate, but dollars. We do need to invest a little bit more in sales and marketing. Make sure we have quota carrying sales reps. Ryan's looking hard at, hey, I want to make sure I have the right specialist team. We want to make sure that we're focused on the AI natives. Also, all the new Voyage folks, we want to make sure we get them converted, so we're going to likely pump up the dollars there. Here's the great part, too, Rob. We've not really benefited from AI internally, and we've only started to roll it out. Our goal is, hey, we want to reallocate those dollars to drive growth, and then around working capital, hey, it's just discipline and focus in terms of how you manage your working capital, and the whole team has done a great job. We're super proud of the margins. Keep in mind, though, we are investing to drive growth. Growth will always be the number one priority, but the business model allows us to invest a lot and still drive margins up.

At ultimate state, should a modern software company have better margins than we used to see, or do you think that that needs to get reinvested from a product velocity standpoint, that maybe it's a little bit more at parity? Because you're going to save on sales and marketing. Clearly, there's G&A efficiencies that you're going to find. But how do you think about shaping this thing for the future?

So it's a great question. I think one thing different now than before is AI and the ability to automate. I think that that's a big piece. Whereas before, yes, you could automate and get efficiencies, but it was still very people-centric. There are so many processes that I believe over the next couple years are going to be much better run through AI. As an example, everyone's doing customer support. Hey, things like BDRs and stuff like that, you can do that automated. I think AI will change that game, and the big question for lots of companies is, do I want to reinvest those dollars, or do I want to take it to the bottom line? What we've said is we want to reinvest those dollars to drive growth, and I think that creates the flywheel even more. That, I think, is the difference between now and before, which really says you should be able to drive better margins and still grow.

Great. Are there any questions? We've got about three minutes left. If not, I've got one or two left. Maybe we can speak to customer acquisition and the strength that you've seen over the last year, some of the changes that have been put in place and how you're affecting that.

Sure. So Ben talked a lot about us wanting to make sure that agents love Mongo as much as developers. A couple years ago, we took a step back and said, hey, we probably took our eye off the ball as it relates to developers and AI natives. We reinvested a lot of dollars around marketing and developer awareness, specifically in the Bay Area. We call it Reclaim the Bay. And we also wanted to make sure that we were also driving the marketing awareness, doing all of the events there. That drove really good top of funnel. We talk about our PLG, which is our product-led growth. That team does a great job driving that growth, converting from community, which, by the way, is a huge community. They use Mongo that touches us every day, moving them through the stages, and we've seen that in the net customer ads. The last two-quarter inflection has really been driven by Voyage. And net new customers in Voyage, we've acquired that business a little over a year ago. That's now starting to see the velocity in terms of customer ads. It has not shown up as a percentage in revenue, and it will, but Voyage has really been that driver. Now we need to convert them to Atlas customers and increase that revenue. That's been the biggest inflection. Questions?

I'll ask one high-level one to wrap us up. With CJ at the helm, just changes in terms of the business, how it's functioning, direction. You know, I think Dave did a fantastic job getting you guys here, but what's been the relative change, I guess, with CJ taking the helm?

Yeah, and to your point, hey, Dave, what an awesome job he did, right? You get the company where it is, so kudos there. All companies go through changes. I think CJ's brought, I would say, three things. One is a myopic focus on customers. Where Dave had it before, CJ spends the majority of his time talking to customers. It is truly a customer-led company. That's number one. Number two is focus around the enterprise. He's very focused on driving that enterprise growth. That's probably the area he's talked publicly of, hey, I thought you folks would be further along on that, the C-suite thing. That's number two. And number three, and this is kind of both of us, is the great part about Davis, he knew every part of the business. He built it. We're putting much more discipline into the rigor of running the business. What does that look like? monthly operating reviews, quarterly operating reviews, are we getting a return on investment? And that's been a change for the company.

So have your marching orders then changed as a result?

It's really that last piece, which is most companies do a great job, Rob, as you know, looking out the rearview mirror. What happened? Our focus is what are we seeing as leading indicators to allow us to change? Also being better from my perspective about giving the leadership team. Here's how the quarters look. Here's how the years look. What changes do we need to make inter-quarter that we didn't have before?

Great. Well, with that, I think we'll wrap. Thank you.

Thank you.