Everybody, Tyler Radke here. I Co-Head the software sector at Citi. Welcome to day one of the Citi Tech Conference. We're towards the end of the first day. Happy to have MongoDB to help close out the software track. We got CEO Dev Ittycheria, and new CFO, Mike Berry. Gentlemen, thanks for making the appearance, and Dev, I think the first time you came to the Citi Tech Conference was maybe back in 2018, in the early days of Mongo, kind of in the SaaS mobile era. Can you give investors just a quick overview of Mongo, how you see the growth opportunity with AI, maybe in relation to you know, what you saw years ago?
Sure. So MongoDB is a next-generation database. We essentially organize data instead of in rows and columns or in a tabular format, we organize data in a document format, which is, we believe is a much more easy way and a more natural way to express data in the modern world. We also support structured, semi-structured, and unstructured data, which also more mimics the real world. And third, the range of use cases go from what I'd call strongly consistent or transactionally intensive use cases, where like system of record applications, to eventually consistent, where they're like things like logging or time series data that you want to track over time. So the breadth of use cases we support are very wide.
I believe it's not controversial to say that we are the, you know, most popular, you know, modern database in the world today. We have customers that range from the smallest companies, two people in a garage, to the largest companies in the world, and you know, we were founded here in New York.
Yeah. Great. And, Mike, welcome. I know you've been to the Citi Conference and in many prior other roles, but, nice to have you here. Maybe give us a quick background on yourself and what led you to join MongoDB?
Sure. So thank you for having us, Tyler. Happy to be here. So, and I know many of you folks already, so I've had the privilege of being CFO at several great companies, both public and private, mostly in technology, and the last one was, it's a great company, NetApp, and after that ended, the whole goal was, "Hey, we're gonna go enjoy time with the grandkids, maybe play a little bit of golf." Semi-retired, and then this opportunity came up, and it was just way too good to pass up. And I saw a company that was in a huge market, with a great product, with great growth opportunities, and then I looked and when I looked, and it's a little bit better now, it's not great, but it's a little better.
I looked at how many of you folks valued it, and I thought, "Wow, what a great opportunity," and I still, if I felt that way in May, I feel that way even more so in September.
Great. Well, I thought a good place to start would be the most recent quarter. Clearly a very strong quarter, you know, across the board, but particularly in Atlas growth, which re-accelerated to 29%. How would you just sort of unpack the drivers of the quarter? Obviously, there's a lot of excitement around AI, but there's also a lot of your own execution improvements you've been driving, and maybe there's some broader strength in the database category. Obviously, the Snowflake, Microsoft Azure continue to do well. So help us understand, you know, across those dimensions, kind of what drove the strength in the quarter.
Yeah. So, while we believe that we're well-optimized for this, you know, future world of AI-enabled applications, and that was one of your first question, which I failed to answer. AI actually did not play a huge role in the performance for our quarter. It's really our core business that we performed well. I would really assign most of the value of how we got our results more to internal execution. We had made some changes last year, where we decided that we want to move more go-to-market resources upmarket, because that's where we saw the best returns and the best sales productivity, and then to do that, we also had to better serve the small and medium-sized market through our PLG motion or our self-serve business.
And I think both, let's call it, engines are working well. The workloads we started acquiring, you know, in the middle of last year through this year, have frankly, you know, done quite well. While we don't want to declare a victory, we think that that a big part of that was that move upmarket. They seem to be growing faster and for longer than what we had seen previously. And then, if you look at our customer acquisition count, the customer acquisition numbers are growing very healthily, and that's a function of us being much more efficient in how we acquire and service the small and medium-sized market.
While those early customers don't really make a material impact on revenue, they're good precursors of future growth, and the fact that many of those companies are AI-native companies is also a signal of future growth, and you know, we've seen this movie before. In the cloud era, we had a lot of, you know, customers come to us through the self-serve channel who end up being very large customers, like Coinbase and a bunch of other accounts that are, you know, large customers of ours today.
Right. And so I'd love to double-click on the workload strength that you called out, 'cause I think for a lot of investors, right, it's not a KPI that we can easily see. But maybe just give us some examples of, like, what are the types of workloads that you have acquired in the last year that are doing so much better than a year ago? Is it just size, or are these maybe more, you know, complex, high-growth apps? Just give us a sense for that change and how to kind of think more about those workloads?
Yeah, and, the reason workloads are important is in our business, the unit of competition is the workload, not the customer. And what I mean by that is in every account that we service and support, we're also our competitors on those accounts. So a large enterprise could literally have umpteen competitors, and we're one of the, you know, vendors that they're using. And in many of those accounts, even though they may be large, seven- or eight-figure accounts, we still have small wallet share, so we have lots of upside growth in those accounts. In terms of what types of workloads, they really range. One of the strengths of MongoDB is that we support a wide range of use cases. Unlike, you know, we've, we're...
I don't like the term NoSQL, because I feel like it kind of confines us to a small bucket, small niche bucket. But a lot of other NoSQL vendors are kind of like a one-trick pony. They're good at one thing, where MongoDB is very broad in the kinds of use cases it services, which is why it's so popular, because you can use us for a system of record, a use case, like a trading application, a billing application, some sort of system of record application. You could use us for, say, a gaming application. Obviously, we have lots of gaming companies have built their platforms on MongoDB.
And then you can use us for things like, as I said, you know, some other, some of these newer use cases, like time series logging, use cases, where people are doing interesting things to try to understand what's happening in the trends in their business. So it really varies depending on who the customer is and what use case it is. So I wouldn't say there was one thematic use case that we saw. It was more the fact that we were getting more strategic use cases coming to MongoDB with our move-up market, which was driving the growth in the quarter.
Got it. And one of the areas you called out in the quarter was strength in large U.S. customers, too, which I think was notable. Could you elaborate a little bit on that? Was it kind of traditional enterprise customers? What were these types of apps or use cases that you saw?
Yeah. So, I would say it's both the U.S. and Europe. We saw it was large, traditional enterprises. As we moved up market, those customers were basically the customers where we saw those workloads growing the fastest. And, I think it just speaks to the fact, you know, you're right, it's very hard to truly understand the users' dynamics of a workload when we first acquire them, because a lot of it is tied to the underlying fundamentals of the end customer's business and the use case that they're using us for. But in general, what we're seeing is that these more strategic workloads are growing faster.
It was obviously when we first acquired them. It's hard to tell because workloads do grow fast in the first couple of quarters, but they've continued to grow reasonably fast, you know, you know, as they're entering their kind of one-year anniversary. And that makes us feel good that these are high-quality workloads that i f you can keep repeating that pattern, it also suggests that the workloads we're acquiring more recently have good potential for growth longer term.
Got it. But to be clear, you're not embedding that strength kind of continues on those large U.S. customers in that count?
You know, again, consumption is always hard to forecast. That's one of the challenges of the consumption business. I would say, you know, workloads at some point do plateau.
And so it's our responsibility to continue to acquire high-quality workloads because that's how we drive the growth. And on a bigger and bigger base, it requires us to make sure we acquire enough workloads to drive sufficient growth. We think we have all the mechanisms in place to do that, but it's definitely a work in progress.
Right. And I guess if you just back up and you think about the go-to-market changes over the last few years and, you know, to use a baseball analogy, like, do you think we're, you know, in the later innings of all those changes? Are you satisfied where it is, or is there still kind of more room to improve, you know, either on the workload, quality, acquisition, or other parts?
Yeah. What I would say, the word one employee used with me is vindication. We had made these changes last year. Obviously, the stock had seen some volatility, and we had faith that these changes would work, but it took some time for it to show up in the numbers. And I would say we feel good about the numbers so far, but we're constantly examining and fine-tuning the model. You know, I use the term pruning the bush. It's like you're growing this bush, but sometimes you have to clean, you know, cut some deadwood, cut some dead leaves, and make sure the bush continues to grow and flourish. And there's always some fine-tuning involved, but we don't see any need to make any radical changes.
Okay. Okay, great. So hopefully after this last quarter, you're getting a lot less questions on Postgres, your favorite topic. But I did want to ask about that. Just, you know, from the context, I think when you know, investors and others look at many of these AI startups, some of which you have, but many of them are using Postgres. You know, the Stack Overflow survey, which just came out, showed Postgres kind of continuing to be, you know, one of the top databases. So, and of course, it's a big enough market, but why do you think it's become so popular? Like, how would you attribute, what would you attribute that to?
Yeah. Great question. To answer the last question first, the reason Postgres is becoming really popular is because the relational market is consolidating. People are moving off relational means they're moving off Oracle, moving off SQL Server, and moving off MySQL, and consolidating on Postgres. Postgres is an open-source standard. Actually, I don't know if people know this, but the name is derived from Post-Ingres. It's actually built on a 30-year tech stack. Obviously, it's evolved, and there's lots of contributors to Postgres, but the whole relational market is consolidating on Postgres. That's the biggest factor driving its popularity, because the relational database has been around for over 50 years. That's point number one.
Point number two is that, for many people, that's, you know, they learn about relational databases first in school, in computer science classes, so that's the first thing they know. This whole newfangled modern databases and document databases is something that's, you know, relatively new, so that's something they learn later. So invariably, people start building on relational databases. But there are certain limitations with relational databases. One, they're designed to be a tabular system, means that you have to organize data in rows and columns. Think of it as a glorified Excel spreadsheet. But as we know, the modern world has different kinds of data. It has structured, semi-structured, and unstructured data, and to manage that in a tabular, you know, or superimpose that into a tabular format becomes challenging.
Postgres has tried to evolve and basically support something called JSON, which is a new standard. We are a JSON document database, but the fact that Postgres is supporting JSON, I would argue, is tacit admission that their tabular architecture has limitations, and they need to make modifications to that architecture to support the way modern workloads are trending. The challenge with doing that is, like, much like software, the architecture you start with is pretty much defines how your product works, and so they have to, you know, they've kind of add on what I'd call a little bit of a hack to support JSON, and so has limitations in terms of the document sizes it can handle and the performance it can deliver.
So when we talk to startups who start on Postgres, they start running into scaling challenges, especially recent AI-native startups that we've been talking to in the Valley, and now they're talking to us about migrating off Postgres because of some of these challenges. They're not a native JSON database. And Postgres and all relational databases are designed to be single-node systems, so they're not. We were designed to be a distributed system from day one, so there's natural scaling challenges that they have. The third advantage that we have is that our data model is inherently flexible. So as the real world changes and evolves, your data model can reflect that. In a relational model, the model is quite brittle, so as you add changes, or what people call schema changes, they're quite complex and time-consuming, so they tend to not happen that often.
So invariably, relational applications tend to incur a lot of technical debt. And so one of the advantages and one of the reasons people choose MongoDB is, as the data model, you know, requires lots of changes, it's so much easier to make those changes in our database than, say, a relational database. So for those reasons, people do gravitate to MongoDB. But again, I cannot stress enough, Postgres does not have to die for us to win.
We have a very small share today of the market, and the blessing and curse of this market is that you can have a small % share and be a $2 billion company. The curse is that it takes time to educate the whole market about your inherent advantages and to keep up with all the things that we have. To that point, what I would say is the analogy of us against Postgres is actually wrong, because we also embed a search engine, a vector engine, and an embedding model. If you want to really compare MongoDB, you have to compare it to Postgres plus, like, a search engine, like Elastic, plus a vector database like Pinecone, plus, say, embedding models from a third party like Cohere or OpenAI.
So then the value is that you get all that on one platform, one unified, elegant developer experience. All the data is one place, and you can do some interesting things, like, for example, you can auto-embed data as it's entered into MongoDB using the Voyage embedding models. So it just really simplifies a developer's life as they think about building either, you know, regular applications or now these newfangled AI applications.
Yeah. No, super interesting, and I, I guess as you think about that dynamic where some of these startups in the Valley, or maybe startups is not the right word, 'cause in the scale of their revenue, where they are running into Postgres scaling issues, like, how far through that, you know, opportunity are we? I mean, I know it's early days in AI, but, like, as we think about the new customers you've added, are those, like, Postgres replacements, or is this something that just hasn't really played out, but you think maybe, you know, in the quarters and years ahead?
So a lot of the new customers we have added self-identify as AI, you know, companies, so that's a good sign for us because we are adding a lot of those companies who are kind of thinking about using us as their platform, which gives us, you know, makes us feel confident about the future. Some of those companies are not really hitting product-market fit, so they're still quite small. Some of those are thin wrappers on top of the LLM, so we'll see how durable their business is. But the point is that we are definitely attracting the next-gen of developers to our platform, so that makes us feel good. We do have some large AI customers.
They were not material enough to be material for our growth, but they're, you know, important customers and are, you know, well-known customers in the AI community. We feel like it's still very early days. I mean, if you think about it, there's probably sub-twenty ISVs who are getting any meaningful AI what I call AI-centric or AI-native ISVs, who are getting any meaningful traction in the marketplace. This is not like hundreds of them.
And I think you're gonna start seeing more, and so still think we're in the very early innings of this whole AI journey, and especially in the enterprise. I think the enterprise is still very, very early innings in terms of the AI use cases, most of them are focused on end-user productivity, whether it's code gen, business users using kind of tools to extract data, to summarize data, to distill data, to maybe, you know, auto-generate, you know, presentations and spreadsheets. But when I talk to financial executives, for example, you know, within a ten-mile radius of this hotel, and ask them, "Is AI transforming your business?" They absolutely say, "No." We're still in very, very early days.
Right. Right. And when you think about your use cases within these AI customers, do you know, do you sort of envision that the growth of the Mongo usage will scale with the revenue of those businesses? Or does it kind of depend across AI customers?
I mean, it's hard to give you, like, a standard answer. It really depends on the customer. But in general, as their business grows, our business should grow because we see that with almost every customer. Like, our usage is tied to the end customer's business. The types of usage, like, we sometimes are the memory layer or the library for the application. We maintain state. You know, inference is the high ground for AI, and to do inference, you need to have an OLTP platform. You cannot do it on an OLAP platform because you need to leverage the real-time data to make a decision on what to do based on what state something's in and what action you need to take.
We are well-positioned for as inference workloads become more and more popular in the enterprise. We feel like we're well-positioned to win many of those workloads.
Got it. Got it. Okay, I did want to touch on migration, the migration opportunity, you know, just given there's been a lot of product innovation with the Relational Migrator tool, and I know you've kind of been ramping up almost like a forward-deployed engineer-style model to help accelerate these migrations. But just talk to us, you talked about the relational market consolidating. How is that? How are you taking advantage of that in terms of that migration business?
Yeah. So when we think about it from a first principles approach, why would customers want to migrate? Well, if you look at, there's thousands of relational applications that an average enterprise has, and the challenges are numerous. One, the tax of running on those legacy platforms is very, very high. There's a ton of technical debt that those applications incurred because it's hard to innovate on those platforms. Third, some of those platforms are end-of-lifing, like Sybase. Fourth, if you're in a regulated industry, like financial services or healthcare, regulators are saying there's a systemic risk in your business, that you're running your business on these old platforms. And fifth, if you want to AI-enable these applications, where not only you just need the data, but you need the metadata, it's hard to do that on those legacy platforms.
You need a more modern platform to do that, and so, for a whole host of reasons, kicking the can down the road is no longer an option, so we see a lot of pent-up demand of customers wanting to migrate off these legacy applications. To migrate, you have to really do three things. You have to move the data well, first, you have to remap the schema, then you have to move the data to the new schema, but the third thing you have to do is you have to rewrite the application code. That has always been historically the most painful and time-consuming and costly endeavor, well, what's interesting is now with AI, you have all these code gen tools that can automate the code migration or code refactoring process, and so that's what's really given us a chance to really go after this opportunity.
So, we are building an automation tool to do this migration. We're gonna talk about this in a lot more detail at our Investor Day, which I know you're coming to, in two weeks. Literally, it's here, two weeks here in New York, where our CTO and others will go into a lot more detail about what we're doing there. But it's basically, to use a product-centric approach to automate the migration process of moving off relational to MongoDB in a cost-effective and safe and reliable way.
Got it. Okay, great. Mike, I did want to bring you into the conversation here. You know, obviously, Dev mentioned the Investor Day, but before we get into, you know, what you have planned for all of us, I'd love for you just to kind of give your perspective on, you know, how you think about efficiency at Mongo.
You know, I think you have two quarters under your belt, or at least two earnings calls, and you've raised margins twice which is good. So what were kind of your observations around the efficiency, and what are sort of your goals as it relates to that?
Yeah, thanks for the question. The great part about MongoDB with the efficiency is this is not all about... Well, there's small changes we can make, and we did a small restructuring in Q2. This is all about how we invest all that great gross profit from the growth of the company. It's got a wonderful business model. This isn't about going and taking a bunch of costs out, it is, how do we make sure that we are incrementally investing to drive growth? That's the one thing I want to make sure is, yes, we are very confident we can drive margins up, but folks, we're not gonna stop investing in growth because that growth then feeds all the profitability. For us, it's about making sure we optimize our go-to-market. We have the right people in the right spot.
We have the right spans and layers. We have the right mix between direct and partner. In R&D, making sure that, hey, we're really focused on leveraging those engineers across all of the code that we do, that we're using AI to drive efficiencies, and then also in the back office. So there is a great opportunity for us to continue to drive margins up. But again, I will underline, we will continue to invest in growth. And if you look at our guidance for the second half, operating expenses are still growing pretty significantly, and there's still gonna be an investment there. So I don't want you to think that we're pulling back on a lot of stuff. We're just gonna get better about making sure that all the incremental spending is gonna drive growth.
Yeah, that's great. And I'm sure there'll be a lot more you'll share at the the Investor Day coming up. But, I mean, do you philosophically, do you kinda think about, you know, some balance or Rule of 40 in the growth of Mongo? Just anything high level without, you know, you committing to specific targets, which I know maybe we'll get in a couple weeks.
Yeah. So can we talk about a couple of weeks, and then I'll hit that? So we have our Investor Day in two weeks. We're really excited about it because we want folks to walk out of there as excited about the business as we are, and really focused around two key areas. One is the durable growth of Atlas and why we feel so good about that, and the other one is being able to continue to drive margins. So we're gonna hit that a lot. You'll hear from Dev, that's gonna talk about the market, the market size, how we go to market, a little bit about the go-to-market changes, but we won't focus a ton on that.
You're gonna hear a lot about the product, the power of the platform that we have, as well as all the new products that we've rolled out and that we're looking at doing. You're gonna hear from customers and partners as to why they chose MongoDB, and we're gonna give you a new view, which is, hey, the self-serve model is super efficient for us. And May Petry, our new CMO, is gonna talk to you about the power of our self-serve model, and then I will give a much... Not a business update, but I'm gonna talk much more about the financials, and it's possible we might talk about a long-term model at that point.
Okay. That's a good preview for a couple of weeks from now. So going back to AI, Dev, you hit on Voyage a little bit, but I thought that was an interesting acquisition earlier this year. Certainly, embedding models, you know, are increasingly important as it relates to, you know, the emerging AI stack. And I think Anthropic actually kinda recommends Voyage as their default embedding model. So how do you sorta. What's the vision around Voyage? And, you know, are there other parts that you kinda expect to fill out around that, you know, kinda the emerging AI stack?
Yeah, sure. So, one, it's important to understand that when you do vector search, you're doing what's called semantic search. You know, you're doing searches that sound like, or look like, or feel like X, right? And so to do that, you need to embed your data so that the computers can basically compare a numerical representation of data to other data to determine if there's some similarity match. And so what's interesting is, obviously, LLMs use, you know, vector search engines to basically find the right information.
What's interesting is that people kind of took this embedding thing as kind of, you know, for granted and said, "Oh, you just embed your data, and then you move on." What people are starting to realize, and I think what we were prescient in understanding, was that the quality of your embedding model has a high correlation of the outputs of your LLMs, right? Because your embedding model is your bridge between your private data and the LLM. No one is gonna give their private data to OpenAI, or to Anthropic, or to anyone else. So you have to have a mechanism to be able for the LLMs to reason about your private data.
The way to do that is to embed the data using a high-quality embedding model, and then the better the embedding model, better the embeddings, which means the better the LLMs can reason about your data, so you can ask and get more. You know, have a more sophisticated understanding of your private data, and obviously, have more sophisticated answers to questions that you may be asking the LLM about, and so because of that, people have started to really value the quality of embeddings, and to your point, people are now recommending Voyage as an excellent embedding model. We just came out with Model 3 that has very sophisticated features around understanding very large documents, understanding both the big picture, as well as being able to chunk up the document to ask very precise information about particular pieces of information.
So these kinds of sophisticated techniques are not available with standard embedding models. What it's done for us is, one, it's obviously, you know, given us more exposure to the AI community because they value these embedding models. Two, you know, Voyage itself is acquiring customers, so we're bringing more and more customers into the MongoDB ecosystem, and we do see some future cross-sell opportunities. And three, we are selling the Voyage models, and the API access to the Voyage models stand-alone because we don't want to limit people who are using other platforms, and we think that the more people use Voyage, the more attractive they will be to MongoDB. So it's been a great acquisition for us. The team at Voyage is excellent.
We're very thrilled to have them, and we have some big plans with them for the future.
Got it. And Dev, I wanted to talk about a little bit about, you know, top of funnel and continuing to attract, you know, the next generation of developers. I think it was probably seven years ago when you made the move away from the quote, unquote, "open source licensing" to SSPL. Clearly, the company has grown by orders of magnitude, both in the customer base and revenue since then. But how do you, what are some of the things you track internally to measure that you still kind of have that, you know, developer, you know, network effects or popularity?
I mean, just given the rise of, you know, other technologies in open source, how do you kind of stay close to that without, you know, obviously, the risks of giving, you know, your code base to Amazon or something like that?
Yeah. I mean, you know, when I joined MongoDB, we only had one business model, which was to sell subscription software. And the fundamental challenge with these, the classic, let's call it open source model, is that you have to find this happy medium between what do you give away for free and what do you charge for. You give away too much for free, it's very hard to monetize, and companies like Redis and, you know, struggle with that. Or if you don't give away enough, it's very hard to drive adoption, which then basically makes it harder to monetize because then you don't have enough people to upsell to. So finding that happy medium becomes quite a challenge, and I frankly didn't love that model.
So to me, open source as a service is the best way to monetize open source because, you know, no matter what you use, you're paying for a little bit. It could be a little or a lot, depending on how much you're using open source. So that's why we found the Atlas business model really attractive, and I know there was a lot of skepticism when we first launched Atlas. People said, "Wait a minute, you're going to partner and compete with the hyperscalers? They're going to eat you for lunch." And I think people underestimated how customers would value best-of-breed, you know, technology.
One of the key things that we saw with the more traditional open source licenses was that other companies, in particular some of the cloud companies, would take the open source, put it on their platform, monetize the technology, but not have to give anything back to the original contributors of the code. And we had incurred a lot of you know, we raised a lot of capital from investors, and we had you know, we're hiring very expensive engineers. We felt like we needed to get a reasonable return on that investment, so we felt it was important to protect our IP. So we first were on AGPL, which is a standard open source license, and then we came out with our own version, which is called Server Side, a license called SSPL.
You know, we tried to work with the OSI, but at that point, they were not, you know, really open to the idea of a new license, and we felt that there had to be a new license in the world of cloud. So we came out with our own license. A lot of people was like, "Oh, my God, you're going to have, you know, people abandon MongoDB because you're not on a standard open source license." I thought about that problem. I said, "If you're a developer sitting in Shanghai or Mumbai or New York or Palo Alto, are you really going to care about, like, what license you're using? Or are you going to care about the fact that it's free to use, free to modify? You can still do everything you want to do.
The only limitation is that if you create a competing MongoDB as a service offering, that you have to open source all the extensions you built to MongoDB and the underlying management plane." I said, "The average developer is not going to worry about that," and that bet paid off, and our business only grew faster after SSPL. There are some people who are quite religious, you know, and dogmatic about licensing, and so some people still don't view us as a real open source solution. It hasn't hurt our adoption. We are still the most popular modern database in the world, and we're continuing to see, you know, millions of downloads, you know, every month of our software. You know, our job is to continue to educate the market about what we're doing and why we're doing it.
Our partnerships with the hyperscalers is very productive. We partner and compete, and we know how to do that well. And we have a strong community. We'll be making some announcements on how we're continuing to invest in the community product. We're bringing new capabilities to the community product, which we'll talk about on Investor Day.
And on the keynote before Investor Day. So there'll be some interesting announcements showing our commitment to the community, and that, I think that will be well received by, you know, all the users who either want to use MongoDB or use MongoDB today.
Sure, sure, and I guess you touched on the hyperscalers. You know, last year at your Investor Day, you had a gentleman from Microsoft, you know, obviously on the Azure Marketplace business. You know, Microsoft obviously has seen their growth accelerate quite a bit, since then. Last quarter they did call out strength in both Cosmos as well as Postgres, you know, database strength across the board. How do you kind of view yourselves within that context? I mean, is there a bit more competition as opposed to partnerships, or are you kind of benefiting from that Azure fair game as well?
The Azure relationship are. I should just say, roughly, our share of the Atlas workloads on the hyperscalers kind of mirrors their share in the marketplace. Just roughly, it should... That'd be a good framework for you to think about. And our Azure business is actually growing quite well.
Yeah.
So we are actually quite pleased with the growth and the, you know, the quality of the relationship and the partnership we have with Azure. Again, I cannot stress enough, this is not a zero-sum game market. This is not, you know, this is not a social, you know, social media market. This is not a search engine market. This is an enterprise infrastructure market, and there can be multiple winners. And we have a low percentage of share, and we just grow the share points a couple of points up and to the right, and we're, you know, going to be double or triple the size of the business today.
Yeah. Great. Well, in the last thirty seconds, I just wanted to hand it back over to you, if there was anything you wanted to get across to investors, any other previews you wanted to slip in for the Investor Day coming up, but appreciate you making the time here.
Yeah. Thank you very much for having us. Obviously, we're proud to be a New York headquartered company. Glad to be here in New York City today. We feel really good about the market. I just want to remind people, we're going after a massive market. There can be multiple winners. We have a really durable technical advantage. We think we're well positioned for the coming AI wave, and we serve some of the most demanding and sophisticated customers, which I think serves us well going to this next phase of this technology shift. Thank you.
Great. Thank you.
Great. Thank you.