MongoDB, Inc. ($MDB)

My name is Koji Ikeda. I'm 1 of the software analysts here at Bank of America. Thanks for coming to day 2 of our tech conference. I am absolutely thrilled to have MongoDB here with us today. We have CFO, Mike Berry; and Ben Cefalo, Chief Product Officer. Thanks so much for being here. Appreciate it.

Thanks for having us.

Thanks for having us.

So you guys reported results. Let's just kick it off, right off. You guys reported results last week, last Thursday. Maybe a question for you, Mike. Can you give us the key highlights. We're in -- I guess, most importantly, where are you getting the most questions, what topics around? And how are you answering those questions post results? .

Sure. So thanks for the question. So as we talked about in a lot of the different calls, hey, we had a strong Q1, fourth straight quarter for Atlas growth, 29% plus. It was a fifth straight quarter where we added incrementally more Atlas revenue year-over-year, which is a key thing for us. We rolled the beat for Q1 into the full year, and we raised the back half. Importantly, you guys -- you folks all have your own models for us. Yes, we bumped up EA in Q2 because of what we see in the pipeline, but everything in the second half for us was [ arrays ] in Atlas. We are starting to see some movement in enterprises as it relates to AI. It's still early days, but a lot of the feedback we're getting from the sales team in terms of, hey, the work that's going on there, enterprises are all focused on it. There's still a lot of has to be done for them to actually roll it out in mass, but we're starting to see some movement, which is great. And we continue to feel good about driving durable growth as well as driving revenue and profitability growth. And a small thing, second straight quarter where we were GAAP profitable, which is great. Biggest questions we're getting is durability of Atlas growth. We do get a lot of questions just in terms of, hey, EA and Atlas, and we just want to be clear, folks for us, this is an and not an or. We do not expect the EA growth to have an impact on Atlas growth. There are different use cases. It's largely a different [indiscernible] motion. And then, of course, there's all the questions around AI.

Got it. Got it. Fair definitely. Ben, since we have you, you're the product guy, I want to kind of go back to the core value proposition for MongoDB, especially in the fact that you guys are operating in such a dynamic and involving world. And so for those investors in the room and those on the webcast that are just trying to get back to first principles on MongoDB. Can you talk a little bit about what the core values that MongoDB delivers today. And maybe more importantly, as we go forward, why are customers increasingly standardizing on MongoDB, what is that reason?

Thank you. It's a great question. So I think first [indiscernible] wise from the very beginning of the birth of MongoDB, it was always about the document model in JSON, right, the JSON format. And that is either the founders, and I wasn't -- I've been here 9 years, but I wasn't part of the founding team either they were really smart or really lucky that all of the AI world has really done -- has standardized on the JSON document model format. It provides great flexibility. It provides great performance. It allows you to really look at your data model in this unstructured world. And majority of data that exists today and all the new data that's being generated, the majority of it is completely unstructured. So from the day 1 built from the ground up, we natively support this. It's the only type of format we really do support, and we're continually investing in that. As you saw when we released MongoDB 8.0, fastest release of the database that we've ever had. So we're still as much as we're investing in all of these other areas, we're still very much focused on making the database better and more scalable and more performant. . The second thing from just a value side is -- and to answer your question more on the standardization, customers really agree with our strategy of if MongoDB is the system of record keeping? If you think about like a circle and your core data is in the middle, why do I want to have to copy that data for my search use case or a copy of that data for vector search use, copy that data out for in [indiscernible] analytics use case, whatever it might be. And our strategy, especially when it comes to the Atlas platform, has always been, we have your system of record data and that's continually add more use cases, but not be bolting on plug-ins by actually building out new value-added services, but do all the wiring for you behind the scenes as well as keeping the query and the developer experience exactly the same. So we've done that with Atlas Search. We did that with Vector Search, we acquired Voyage last year. We've done the same thing with that. And so it's really about keeping the developer user experience in mind at all times. So the developer doesn't have to query multiple different systems for a singular use case. So I would say those 2 things are really why people are standardizing on us and why they find a lot of value in Mongo.

Ben, you mentioned something about 8.0 being the fastest release ever. Can you talk about that a little bit? What drove that to be the fastest release ever something you're doing on the R&D side, maybe customers pushing you hard for new features. I mean, why was the fastest release ever? And could we expect 9.0 to be even faster than that?

We always can go faster, and we're always going to be investing in the core technology. So that has not stopped. We have never taken our put off the gas there. I think what would cause it to be a big jump really was one around the same time we started development on 8.0, CTO came in, Jim looked at different engineering principles and what -- how we wanted to focus on I really got back to the first principles of how he came from a massive database background, and he just simply said, we can do better. We can make the bars better, and we just put high focus and investment in that area. And obviously, the results paid off.

When I think about Atlas specifically around the core offering, Vector, Voyage, all the other features that are around it. How important is this all-in-one platform for customers today? Is it 1 of their key topics or key focuses? And how does that keep being a key focus for customers in the future?

Yes. I think it shows up a couple of different ways. First of all, customers do like less vendors that they have to have to deal with. I think secondarily, what makes us really unique and something we don't talk about publicly a lot is that we're not locked into any one deployment model whether you want to be in a hyperscaler whether you want to run it inside your own 4 walls, whether you want to run it in a neo cloud, what do you want to run it in a sovereign cloud in Germany, we have different form factors that meet the customers' demand, right? And so being able to give them that true portability is something that no other database vendor, I'd say, modern database vendor can do and certainly not something you're going to get from the hyperscalers. I think all of those things added up is one reason why our customers really love us and are choosing to standardize on us.

And if I could on that, Koji, we talked about it last week. Now about 45% of our customers over $100,000 in ARR use more than 2 products within Atlas. And for us, the way we monetize that is it's more Atlas consumption. It's revenue because you get it included in Atlas. So for us, that's a good thing, and we certainly see the customers adopting multiple features.

Mike, remind me on the [ 2 plus ] front, what are the most common add-on products that are driving that 2 plus for the Atlas?

Vector and Vector Search. Search and Vector Search, sorry.

Last week on the call, you guys called out a couple of big -- some AI customers that you won. And I remember when we were discussing on the pullback and even on the call, One of the reasons why these customers were looking at you as one just Atlas period, but also the interoperability between hyperscalers. How good a differentiator is that for you guys? And how should we think about the capacity constraints at hyperscalers being a good driver for you guys for the next several years?

Do you want to do the multi-model capacity? Cool. So what we -- when we say we run anywhere, it goes back to what I was just saying a minute ago, customers really have their choice. And what we don't -- I think do the best job at marketing this feature when I'm working on fixing that is we actually, what we mean by run anywhere in multi-cloud is that you get to pick what cloud you want to run on, you can actually extend a singular cluster across multiple clouds at the exact same time. You just can't do that with [indiscernible] DB Cosmos, Dynamo because they're just native hyperscaler services. And that is then back into whatever else is going on in a particular region, capacity could be something. This could be that this app is slated for this cloud provider, this app is slated for that cloud provider. It gives the customer a lot of flexibility to say, oh, cool, I'm just going to add a couple of nodes of this cluster to GCP now, even though it was running on AWS. So it provides a customer a lot of flexibility, but -- and then also depending on what's happening in that particular region, it gives them an out -- to move into other cloud providers a bad too.

And capacity so far has not been an issue. There's obviously a lot going on with memory and everything else and all the hardware piece of the world. we have long-term agreements with all of them. So we feel comfortable there. And hey, it's a multifaceted relationship. Certainly, we bought them. There's co-sell that goes on there. So it is much more than the and their -- and there are certainly competitors as well. So it's -- there's a lot that goes into that. We watch it every day. We have conversations with them. It has not been an issue yet, but we're certainly keeping our eye on it.

I know you guys get the Mongo versus competitor, ABCD, what the architecture -- why Mongo versus all these competitors? And maybe a question for Ben. If we could dig down into the technology of Mongo, what are maybe the top 3 reasons from a technological aspect of why customers like to choose Mongo over the competition out there?

Yes. So first of all, it's what I was saying earlier with the JSON document model, we're built from the ground up our storage engine, how we actually write the bits to disc or to memory is all native JSON from the very first line of MongoDB, right? So 17 years old, 18 years old now, again, first principle from the very beginning. Number two, the way we scale is vastly different from a performance and price perspective than how you would have to scale our relational database, and that goes for all of them. And then third is really around the portability and will allow the customer to have all that different levers about how they need to deploy it. They want to start out. We still have an open source database, right? We still invest in that as well. We still get customers that started out there, and they scale to a point of now they need to have the different performance that they want us to manage a form inside Atlas. And so that portability gives a lot of people comfort in choosing us as a technology.

When you talk about scale, the performance and price -- what's the one-liner on how you guys are able to do that, offer such good performance in price?

So we scale different. We don't just need to scale vertically, which is either more machines or bigger machines. We can scale different things independently. So you can have different storages you might need. Your workload might be perfectly fine on a subset of, I would say, like this 1 tier of compute, but you need more storage because your data size is good, bigger. We allow you to either add more storage or you can scale out horizontally, where it's the same type of box we call it starting. And so we do our own data routing inside of MongoDB's distributed system. So it's just a different way of doing it, but we're able to do that at a lower cost because smaller boxes are cheaper than the bigger boxes or bare metal from the cloud providers. So you can have more -- most of the time scale of economy is, you can have a lot more smaller boxes then you can have a couple of larger boxes.

Was that something that Mongo was founded on originally way back when, like this type of architecture? Or was that something that had to be built and later as you guys got bigger?

I've been here 9 years. It was the charting technology that we've built existed when I got here, but I -- if I'm rewinding the whole clock, I don't think it existed in like the first release of Mongo, but we developed it over time.

Okay. Something that you had to do because of the success that you -- I got you. Okay. Okay. So Enterprise Advance, something that has become much more topical for you guys. I guess first question for Mike. On the nice guide for the second quarter. Can you just talk about that a little bit? What is the visibility that you have? Is it a big contract? What's going on with the second quarter guide in EA?

Sure. So let's back up for a second on EA. So as we all know, this is a standard license support model. And we said this last September. If you look at the $500-plus million, about 70% of that folks is support coming off the balance sheet. That's most of what's sitting in deferred revenue, and we're going to get it whether we come into work or not. So a lot of this is -- there's a big piece of that that's ratable, then the license piece is either a 1-year deal or a multiyear deal, where the hard part for us to forecast are the multiyear deals because we don't know customers really don't know because it's based on their budget. Are they going to do a 1-year deal, they're going to do a 3 or even a 5-year deal. So that's the part where we try to be candidly conservative/pragmatic on. We're not going to lean over our skis on those. So entering Q2, we know what's coming off the balance sheet. We expect to see what's going to renew. And then there's already been deals that have been booked as multiyear that we know we're going to recognize. So that's what's in the Q2 number. Keep in mind, too, that in the second half, it's a very tough compare versus Q4. Remember, EA went up quite a bit. So what we said is, hey, for the full year, we expect somewhere around 5% growth, understanding that, hey, if more multiyear deals come in at the back half of the year, great, we're not going to include them in guidance. And then again, the activity we've seen and the momentum there has not been at the expense of Atlas. These are customers saying, hey, I was going to do something else. Now I'm going to deploy. And then I know, hopefully, your next question that Ben's going to be, what are you putting in here as it relates to AI. This is a part where I scratch my head a little bit, which is, hey, today, yes, you can see the growth in AI through Atlas. The future is going to be -- you're going to see it in Atlas [ MBA ] because companies are going to deploy AI internally in their own data sets, and they're going to use EA to do that.

Mike, you got your question. I'm sorry, then you got your questions from Mike.

Yes. And so as this is an and, it's not an or. For various reasons, we're seeing customers continually invest their own deployments of EA for various reasons, could be data sovereignty, residency manufacturing use cases where they need to be inside the facility, hospital use cases where they just need to protect themselves against from a hurricane or anything like that. So there's all these reasons why EA is a great business. And again, we're still investing in the core technology. And then we've said that we're working on from a road map perspective, bringing not full parity from Atlas into EA, but where it makes sense and also based on what our customers' demands are, we're going to be adding more functionality that exists in Atlas only today into EA for those use cases because customers want to be able to have true hybrid workloads, and hybrid multicloud is something that -- a good chunk of enterprise that I've talked to over the last 3 months is really top of mind for them. They want to have that flexibility, and they have -- to have that flexibility, the same core functionality that they've built on inside of Atlas if they have pull that back or they're going into another geographic region where there is no cloud provider region or something so they're going to run enterprise events themselves that they want to make sure that they don't have to change their application.

And 2 big, if I could on this, 2 big pieces. Hey, it's over 20% of the business. It's high margins, it generates a bunch of profit and cash. I love that. That's a big piece of it. The other thing is, in the past, I think we've been a little bit of we've done this to ourselves. We call the business Atlas and non Atlas. We've changed that folks. It's Atlas, and it's EA and other. It's a meaningful piece, biggest customers in the world buy it. It's a big driver of the financials, and that's how we're going to talk about it going forward.

I'm going to ask you 1 more question, and then I want to open it up to the audience to see if you guys have any questions out there. And so I know you guys like to introduce new products around .local events. And so as much as you could without telling us everything that you -- what. Local should we be really paying attention to? I pay attention to all of them, but is there 1 or another that we should kind of be honing in on when we see it on our calendar.

So you should pay attention to all of them. We've reduced the number of. locals to a smaller number. Look, the big ones are -- San Francisco is always a big 1 just because of what's going on there with AI. Our biggest event is .local New York and CJ talked about it. We'll have our next Investor Day right around that as well. Those are probably the biggest ones. Hey, London was a good 1 as well. I think India, I think there was a lot of good stuff. So really New York and San Francisco are probably the biggest ones. But we've also trunked that population to make them all more meaningful.

Got it. Any questions from the audience? I got a lot Okay.

You're going to fill in.

I got it. So 1 thing that has been coming up more with Mongo is how to attach Mongo to kind of the code-gen vibe coading tools out there to become that either prefer default database of choice or 1 of the top 3 or whatever it may be. What exactly is going on there with that? And is there something that needs to happen from a technological aspect to get that initial default database? Or is it just kind of like a partnership agreement? Help me understand that a little bit.

Yes. It's [indiscernible] either, to be honest. It's not it's not a partnership. It's not technology. We're not missing technology. We actually have more functionality. What -- there's a big thing in the industry right now around AEO for agent optimization. So there's a lot of things that we're doing just from a content generation perspective, making our even little things making our website even just easier for agents to read, which is a whole new skill set on its own new profile of worker. But the reality is MongoDB came out in what 2010-ish, right, post-crisis has been around since '96, SQL came out a lot longer before that. There's 50 years of content related to relational and SQL Server, and we're playing catch-up there. And so the prompts, the LLM, regardless of the Vibe coding platform, they're all using something else behind the scenes as their model. They're trained on public Internet data. And so we're just in a content game right now with how much content is out there versus how much content is out there about MongoDB. So we're obviously working on that. . Number two, we are -- we have done things to make agents like us better. We have MCP servers. We're publishing all of our own skills. We've released an architectural Center last year on the documentation page. And so we're hyper-focused on all of these different avenues, but to drive that awareness. So it's a work in progress. And I think it's going to be work in progress for a while, but I think that's just the macro of how fast things are moving. It's not just -- it's not just a Mongo problem.

For an application that was built on something else other than Mongo from an AI native startup, whatever it may be, how easy is it for them to transfer what they had over to Mongo for the future?

Sure. So there's 2 different flavors of this, right? Even with what I was just saying, the funny thing that the LLM are doing is that they're modeling the data properly, meaning that the data model is JSON or in the document model, they're just being some post [indiscernible] flavor with JSON to say the least. That's the most common one that we've seen. But when they do that, that migration is really simple. And I'm not even talking about the data side. We're really good at moving data. We've always had a migration practice inside of Mongo over since the beginning. We're very good at moving data from Postgres to MongoDB, we do that all the time. What makes it easy is the -- if the application is already thinking in the documental perspective, it's on the easier side to then tweak that application, use our driver and then start using Mongo. So it's both flavors of it. is if they're out with something else, but still on the JSON model that's on the easiest side. But even if they didn't model their application use JSON, it's more relational or they're using key value, whatever the case may be, that is something that we've done all the time too. Nothing about this AI generational dip has made it harder or easier to migrate into Mongo. But we have a really good migration practice that focuses on that.

Got it. Got a question for you on voyage. So 2 weeks ago, I spent a lot of time trying to understand what Voyage is and why maybe you guys bought it? And then I realized I had you on the stage in 2 weeks. I said, forget it. I'm just going to ask you. What is it specifically about Voyage that makes it so important for the Mongo strategy?

Yes. I'll -- I have an answer. I will admit, I am not the expert on Voyage, but I know enough to get myself into trouble. So a couple of reasons why this made a lot of sense. Number one, as we were talking about earlier about really focusing on the Bay, we've announced a lot of -- with .local San Francisco and the focus of what we've been calling reclaim the Bay. We -- somewhere between -- before COVID into now, I think we've lost a little of the coolness factor in the Bay Area, right? And so one, Voyage had an amazing brand, not just -- the technology is amazing, too. But we also were thinking of it from a brand perspective. And we wanted a way to make sure we had credibility and a coolness factor in the Bay, that had something to do with it. But the technology is stellar. We're investing heavily in there. When we bought them, they were the #1 embedding model on Hugging Face. They're still a #1 embedding model on Hugging Phase. The Frontier Labs actually suggest customers use voyage and [indiscernible] models. So that was obviously really important to us. But then third, if you look at what the value prop of Atlas has always been and what we've been building as a platform is we don't want customers to have to stitch together all of these different things. And so that -- even though we had Vector search for a really long time, the customer had to bring their own embedding models, and we still support that today, right? You can use Bedrock or whatever else you want to do to embed your data. But now having it inside of our ecosystem directly inside of our platform, it's something else a customer doesn't have to worry about. They can just start using it, right? And having the re-ranking models there at the same time is also super important because then we can continually keep those embeddings up to date. And then the last piece that customers are really gravitating towards is, they like that it's all what's inside of the MongoDB ecosystem. And as Mike said earlier about security and governance and reasons why the enterprises are typically slower to adopt any new technology, they already trust us. They don't need to go get another contract for -- or to use some other random embedding model that it's all part of our ecosystem. So all of those things combined were why Voyage was an important acquisition for us.

Maybe a follow-up question for you or Mike. Clearly, embedding and reranking models are important. That's why you guys bought it, good technology. I mean, when I pull it -- when I did my work, you guys always were top-top-notch with voyage. And so if it's so important, why doesn't the embedding and reranking side get more competitive? Is this something we need to watch out for from a competition front?

So I think it's something that, hey, it always will, to the extent that. And we've seen that. We've seen other folks join. So -- and we welcome all that because it makes us better as well. So it is something that we watch. I think what Ben talked about is it makes. It's more imperative for us to make sure that we are continuing to evolve the product, and that's a big piece of it. We also talked on last earnings call, the number of Voyage customers doubled quarter-over-quarter. Last quarter -- not year-over-year, quarter-over-quarter. So it's still a small revenue contributio,n, but that's driver. And we also talked about in the 2,500 net new customers. We actually saw voyage become a bigger piece of that as well. So yes, it certainly will become more competitive. It's incumbent upon us to keep adding functionality, but we're seeing great traction.

I'm going to guys -- I'm going to ask you guys the when AI tailwind question here, but maybe in a different flavor, not looking specifically at Mongo financials as the indicator of the trigger. What as -- for those in the room and those on the webcast, what should everybody be looking for as signal broadly that the enterprise AI tailwind that should benefit Mongo is really inflecting higher? I mean, is there something to look at something, some sort of demand signal that we all should be focusing on?

So I guess the way I look at it is, hey, all of us are -- have the enterprises are our customer -- where their customers as well. I think you started to see them roll it out in things like chatbots, right? You can't talk to a human today if you try. There's -- you're going to see some of that incrementally move. I think for us, it is when you start to see actually things like your bank is telling you, here's the stocks you should buy. When you're an insurance company is -- now it's a virtual agent, not a physical agent. When you're starting to see AI actually impact what you do every day outside of doing, hey, it's a new recipe on ChatGPT, I think that's when you're going to start to see the inflection point. When enterprises are actually going to their customers using AI to drive their business.

Think about Atlas and EA, 2 good products, 2 different line items. How do we think about the durability of growth or maybe even the mix of growth of these 2 products as a composition of total revenue over the next 5 years. Is that something we could even think about right now? Or do we just have to wait to see how it plays out.

Yes. So great question. So we'll talk about this again in September. Last September, what we talked about was, hey, expectation for, again, 3 to 5 years, not next quarter. Atlas growth continuing to be north of 20%, and then at that point, we said total revenue growth in the upper teens. And that basically embeds you can do the math says that, hey, EA is, call it, a low single-digit grower. I think what we've seen is, we continue to be very excited about Atlas. But to the extent that all the investment and what we've seen from a customer perspective can bump EA up, maybe that moves it up, mean we'd love to talk, and we guided this year for 20% total company growth. But no matter what, Atlas growing where it is, 75% of the business will continue to be the biggest piece and the biggest driver of growth. I think it ever goes to 100%, there's -- at some point, it's going to asymptotically stop. And where is that 85-15, is it 90-10, we'll have to see. I think a lot of that depends on how do enterprises want to deploy AI between the public cloud and on-prem.

Coming up on time here, so I'm going to ask you the growth and profitability question. And I think it's become more interesting after the second quarter earnings cycle where we've seen a bunch of infrastructure software companies prove out the growth and prove out the strategy. And so my -- last question for you. How do you think about that balance between growth because clearly, the growth is there, but profitability matches. And so how do you think about it?

Yes, it's a great question. And this is 1 of the things when I started almost a year ago now, I didn't realize the efficiency of the model. And what we talked about last year was we took a little bit of a step back on OpEx. We said, hey, let's make sure that we see a return on all those dollars. We've started now to increase the investment in fiscal '27, and you see it in our guide. OpEx is still growing upper teens. The focus for us is where can we put investment in that's going to drive revenue growth, it's going to go to the bottom line. It's a great model, and that's how we're going to drive margin expansion by growing the revenue line. We're not in the game of cutting heads at this point. We're in the game of where can we get efficiencies, invest more in product and in quota-carrying reps, most importantly, that then we can touch and feel and say, there's a return there. There's incremental revenue and boom that's going to go down to the operating line. And it's a great model. It's very efficient, and that's the goal now. And we look very hard at incremental investment needs to drive incremental growth, but where we see that, we will absolutely make it.

Got it. We're out of time. Ben, Mike, thanks so much for doing this. Appreciate it.

Thanks for having us.

Thanks for having us.