Confluent, Inc. (CFLT) Earnings Call Transcript & Summary

Hey, good morning. Thanks, everyone, for joining Day 3 of the Wells Fargo TMT Summit. And a good turnout as we're kind of rounding third base and heading towards the end of the event today. Very pleased to have Jay Kreps, Founder, CEO of Confluent. Just a little project you're working on a previous company that turns into something much bigger. But thanks for making time. I appreciate being up here and in Vegas.

Maybe, I mean, we can start with just your view of the event streaming market. Like I know one of the points of conversation that we spent a lot of time with investors is just helping draw prior predecessor anecdotes or helping them understand where the technology sits. And so maybe you can just start with your perspective on sort of the distillation of event streaming and where Confluent fits and then plenty to ask from there.

Yes, happy to do that. So the idea around event streaming is really all about data in motion. And so the big idea here is there's been a huge investment in systems for data, but mostly around data at rest. So the databases and storage systems. It's all kind of a place where you would put your pilot data for one application to look it up as it needs it. But in a modern organization, there's not just the one application, there's lots of software and it all has to interconnect and kind of act together as one thing in real time. And that really is the domain of data streaming and this kind of idea of data in motion. So it's taking a lot of the ideas of databases, but translating it to something that's about data as it occurs. So you could imagine as a simple example, if you were a retailer, maybe a decade ago, the way you would work would be at the end of the day. Each store might report out its sales and you might kind of roll that up at the end of the week and have some idea of what happened and then place some reorders and shipments and so on. But like a modern retailer is much more dynamic and interconnected than that. There's a whole web and online set of interfaces that have to reflect what's in the store. There's a set of mixed experiences where you need to be able to buy, pick things up in the store. So that this orchestration of things is now a real-time part of the operation of the business. And that means the systems in the store, the kind of core IT and inventory systems, all of that now has to be continuously interconnected. And that's just one example. I would say this is something happening across virtually every industry, whether it's financial services, the tech companies, insurance, all the way out to the kind of digital, next-gen streaming services and video games. They're all kind of building around this idea of data in motion. And so Confluent was really started around an open source project called Kafka that became really the kind of key foundational layer in this stack. Is it really the thing that holds all these streams that allows the company to tap into them and process them as they occur. And has really become kind of a mainstay of next-generation data architectures, something that's just kind of there, all the architecture diagrams, you would see and we offer a software product and cloud service around that, that helps companies to adopt this and give them a fully managed cloud native version of it.

You've done that before. Nice job too on the -- one of the things I was going to double click on was just the diversification across customers. We were out at the event in Austin. It was clear, there was a pretty strong financial services presence. And yes. I don't know if I fully appreciate it when we were talking to some of the customers and they're saying everything from fraud detection to payment processing, the trade reconciliation. So can you speak more to just the breadth of use cases that you're enabling? And then there's always this question of like what's production level versus experimental. I know you have a strong view here.

Yes. Yes. It's easy to miss if you don't follow the world of data technologies closely. But yes, Kafka is one of the most popular open-source projects in the world, and it's broadly adopted across virtually every industry, and we sell into all of those. If you think about the progress, it's been a little interesting. It's less industry-specific then I would look at the progress of adoption as going from the more technical companies to the less technical companies over time. And weirdly, that kind of started with very small tech companies and then also some of these very large technically savvy companies. So some of the big banks were early customers. And then it kind of progresses out into the middle, which is plenty of adoption happening there. And so the use cases are quite broad. There's a set of common things that you would see in every industry that is just a bridging across these different clouds and on-premise environments and having some kind of common fabric for data flow across powering a lot of these next-generation micro services layers. And then industry-specific things. So I mentioned the example in retail, [indiscernible], in telecom, in financial services, everything from the kind of next-generation customer experience and really getting to a richer digital interface, all the way to the back-end stuff, the trading systems and risk systems and a lot of the really core technology. There's a set of things moving either from batch into real time or moving to this more modern architecture for something that was already real time?

One of the things that we've gotten questions on just really since the start of the macro conversation was just open source versus proprietary. And I think there were initial concerns hey, is there a potential for trade down or for customers to stay on open source. And it doesn't seem like that's been the case, at least not from spending time and talking to some of your customers and some other things. So what are some of the reasons that even in an environment like this, customers that are on Confluent would make that journey now?

Yes, yes. That was one of the areas where people had concern around us or that we heard we felt like we were in a pretty good position heading into this. And then I have been pleased to see that yes, indeed, the NRR has remained very healthy and above 130%. And then I think we said in the last earnings call that if you kind of break that out, gross retention has remained very strong, above 90%, 90-plus percent. So we didn't -- empirically, that has not occurred. So then why has it not occurred? I think the biggest change -- I think the reason people had this concern is there is and has been that dynamic in a lot of the open source companies where it's like, there's some kind of -- are we using the paid version, are we using the free version. It's kind of a premium/freemium dynamic. What's changed is actually quite structural which is what we're selling in large part now to customers is a managed cloud service. And if you think about the TCO of a cloud service, it's not cheaper to bring that in-house and hire a team of Kafka experts to run it. You will end up using and spending more on the underlying cloud infrastructure. We have huge economies of scale doing that across thousands of customers. You will end up spending quite a lot on the people, and it will be much slower to get anything done. And so that move to the cloud, the move to managed cloud services, that is something that's quite different between like a Cloudera and Confluent, is actually just a very different delivery model. And so both have an element of open source, but it's just something quite distinct between the two. I would say that's the -- one of the biggest reasons. The other reason is just one of the things we believed was it's very important to have a deeply differentiated product. And so another thing that can wash out when there's some elements of open sources, okay, that's great. If there's something that has a lot of traction, traction is something that's hard to get in the world, but what is the product? Is there a real moat of software that kind of defends that business in troubled times under competition, whatever it is, you can't have the economics of a software company without having the software, right? And so that's an area where, from very early on, we built around kind of sort of pillars of differentiation trying to make our service deeply cloud native, like building something that was meant to be a truly Elastic Cloud service, something that's really a complete stack for this kind of data in motion, like beyond just and then making it available across all these environments in different clouds and on-premise and connecting all of that in a way that's kind of unique to us. And so those capabilities are also, I think, what helps defend the business model when times are tighter, people are looking at what must go and what cannot. I think that's another aspect that's quite important.

I appreciate that. Another question we'll get is just around, like is this an evangelistic sale? Is this something customers are aware that they need? And I'm wondering from your perspective, just how much having the large population of existing Kafka users helps with -- because a lot of the question is just what is the replacement market for this. I think with some markets, there's a much more common parallel and because this is newly defined in some way.

Yes, yes. Obviously, the exciting and challenging thing about Kafka is a new paradigm, like it's something new and big. And so you could imagine, well, that must be very hard. And indeed, if we were kind of going door to door saying, "Hey, we have a new paradigm for acting on data, let us educate you about it from first principles". That would be like a slow and difficult proposition. And that is where the open source comes in. You want to have something that is effectively part of the next-generation stack where engineers are reaching for that, for the next project. When they go to a new company, it's part of the toolkit they take with them. If you didn't have that, I do think it would be a very evangelistic sale. I think if you were starting Confluent with none of that and hadn't put the effort into the organic adoption it would be hard. And indeed, there was -- this is -- some of the ideas in this space had a history in computer science and there were all kinds of companies that started this without the open source and just kind of didn't get it to take off velocity even though there's clearly an obvious need. And I think that's why. And yes, so yes, because of the open source adoption, you can kind of just prove the evangelistic thing really in a straightforward way, like open source, we're not selling it person by person. There are hundreds of thousands of companies that have gone and adopted this and put it into production. And then with [ OpenSearch ], we can't track each of them exactly. We don't know what everybody is doing. But there's no question that this is one of the most successful open source things that was broadly adopted. And that does mean there is an organic demand there.

And so what is it -- where are you now in just the process of bringing some of the largest Kafka and organizations, the more sophisticated customers over? I know New Relic has become a recent example of this. But how would you characterize where you are with the go-to-market and with the cloud platform on that side?

Yes. There's a -- there's still a lot to do there. We have about a little over 4,000 customers. And as I said, there's hundreds of thousands of open source users. So there's a lot to go after there. And you can look at that as either good news or bad news, right? For us, we're always like, well, why?

I think that's going.

Why is it only 4,000. But on the other hand, like, look, this is one of the fastest-growing open source things. So there's new people coming up all the time. The good news is, especially with the cloud product, we feel like there's a very compelling value proposition for every user of the technology. Like nobody really wants to be doing this stuff in-house anymore. And so our ability to go out and get those customers, I think is very strong. And now even up to the very largest users of the technology, where this is something where even internally doing it themselves, they would spend tens of millions of dollars on cloud infrastructure and people and so on to run this platform internally. And so being able to come in with something that is just a vastly better product experience is very meaningful to them. And what we offer has a very strong TCO case for it, as I said before, and that kind of that offering, which is better and not more expensive is actually very compelling.

And I think one of the stats, one of the campaigns that was out at the customer conference was 10x more performance.

Yes, yes.

Is that the right step first of all? And then kind of what goes into informing the perspective that Confluent Cloud is 10x more performance than Kafka?

Yes, yes. So one of the things -- I mean, obviously, any marketing is kind of boiling things down to a few data points, and you can do 1 million benchmarks and I'll be a little different. But one of the things we wanted to get across is just the level of investment in the kind of core cloud service, whether that's latency, scalability, the kind of elasticity, how quickly it can scale up and scale down, the ability to store data over time and how much that can handle. On all of those dimensions, there's really a very significant meaningful advantage and that comes all the way from the smallest customers, where we have huge advantages from multi-tenancy and the ability to effectively have shared infrastructure that serves many customers more efficiently. All the way up to the very largest customers where there's structural advantages and how data flows across cloud networks and the underlying processing that we do and the integration that makes it both faster and much cheaper for us to run than it would be for them.

There's been also just a lot of talk around consumption models, and there's some element of that within Confluent's cloud model. There's a lot of kind of talk down the hall around optimization and some of the exercises that are going on with the broader cloud infrastructure layer. So maybe you can provide your perspective on -- I mean maybe you can just start by level setting what the business model is, what the pricing mechanisms are for customers and then your view on consumption and where Confluent currently sits.

Yes. Yes. I think this is a really interesting one. So the model for us as for any of the cloud providers as for Snowflake or MongoDB or whatever it is, an ability to come in and spend on just what you use as you use it and optionally to layer a kind of committed spend on top that, hey, we commit to spending $100,000 or $1 million with Confluent, which will get you additional discounts. So that model is kind of common across a number of these technologies. And it's actually a very foundational change in how the go-to-market works, what it is you're trying to drive, how you recognize value, everything. And there has been a lot of interesting discussion I've had with investors about, okay, what does this mean? I would say, maybe six months ago, it was wild optimism about consumption models and maybe recently more pessimism as some of these things that come under pressure and maybe some customers consume less. I would say, I actually think the consumption model is purely a good thing. And I think there's very little downside. You're taking a lot of risk off of customers in their adoption. I think there is the potential for cost optimization, right? And that depends on how much have you oversold. How much of what you've sold is creating value for the customer and how much is not. And I think the reality is that is true in every product. If you're selling seats you're selling whatever customers are going to cost optimize. It shows up quicker in consumption, but it doesn't really matter. I mean if you're optimizing seats and it shows up at renewal time, at the end of the year, you're still going to see that compression effect. The fact you delayed it a quarter or two, I don't know that it really helps you that much long term. The only cure for that is actually creating real value and making sure that whatever they bought is actually being used effectively. One of the things -- and each company does that differently. And no company is perfect, right? Any company with the sales team, there's some customer who bought more than they needed and didn't use it all maybe because of themselves or maybe because of some mistake on our side. We've done very well through this. We have not seen a compression effect in the overall numbers. Our Q3 results were sequentially up in what we were adding in cloud revenue. So like really strong growth over 100%, so year-over-year. So basically, we've done very well through that. I would attribute that to serving probably more, call it, production or mission-critical use cases, which tend to be relatively well optimized and pretty high value. And so they tend to get less trimming. It doesn't mean that we've never had a customer trying to optimize their spend with us. Of course, they do. It just means that overall, the growth and expansion kind of outweighs that. And there's not a big accumulated set of wasted spend, right? And so -- which there is more of a tendency towards that with some products, right? I know maybe we go and look at our reports every once in while I realize only half of them actually get looked at internally, then we delete all the old ones and of course, the consumption on the reporting and the goes down a little bit when we do that. And so yes, there's less of that with this kind of production software, where obviously, had software engineers build it. It's running some part of the business. It's less likely to get just kind of turned off one quarter.

We haven't spent much time on it. You talked early on just on sort of the macro and just you're not seeing much of a change in decision making with regard to open source relative to Confluent. You mentioned the cloud revenue growth sequentially has continued uptick and you've communicated both confidence in that continuing into Q4 and then provide a preliminary growth target for next year already. Is there anything that you would mention that you're seeing in sales cycles or the macro or some customers that we should mindful of within that?

Yes, I wouldn't want to come across as overly whatever pollyannaish. Yes, I mean it's a harder environment like customers are putting a lot more attention into where they're spending money, and that affects us as every other company. We try to boil down the level of impact over the last few quarters, which was, I think, a small percentage of what we would have done. We kind of gave a quantification of that for both the quarters, that's an estimate. But what is that playing out as, I would say it's increased scrutiny on deals, which tends to extend the cycle and take a little bit more time in analysis. So hey, what is the real ROI on this expenditure. And I think we've done very well through that, like where we've seen delays, we've seen it come through a little bit after a little bit of time with the finance team and the customer. But those delays are not a good thing. That adds to the burden of sales. And so yes, I would say we have seen that -- we've seen it sporadically across geographies, tends to be concentrated on the larger deals. And the net impact of that was relatively small, which is why we did well throughout this. But it's certainly present. I think as it is practically speaking, with virtually everybody in our space at this point.

Confirmed, that we've seen several of these sessions throughout the week. So given that and given the consumption model, your confidence, your ability, your visibility into customers' profiles that allows you to put a growth target out for next year at this point in time?

Yes, yes, yes. So we did give sketch of next year, and I think the intention was just provide a little bit of clarity on where things are heading. Obviously, it's an uncertain environment, right? But we felt like, hey, we have -- obviously, we have committed spend from a bunch of customers. We have some view into their kind of trajectory of consumption. We definitely have a reasonable picture of how that's likely to trend. And so it made sense to give out some ranges in the last earnings call, which, of course, will continue to refine over time.

And then on the proverbial growth margin question. Confluent has been focused on growth and a strong balance sheet. Market opportunity is large. We've talked a bit about that. But do the trade-offs change at all if the macro environment changes? And just walk us through how you and the team evaluate the growth margin framework and trade-offs.

Yes, yes. So we have been thoughtful, I think, about the trade-off between growth and profitability. We gave a framework that we had in our IPO of how we -- what we're going to hold ourselves accountable to, both in the midterm and in the long term. And we've said our target for positive non-GAAP operating margin is coming out of Q4 of '24. And we've made good progress towards that. So I think 13 points of year-over-year improvement in this last quarter. The -- and underlying that is, yes, strong gross margins. Those are at 71%. A lot of work has actually gone into that, even though our gross margin has been quite boring to look at. One of the things you have to understand is that we have a portion of the business that is software licensing and a portion of the business that is a managed cloud infrastructure service and the managed cloud infrastructure service has a very different internal cost to provide. It turns out it's very cheap to make the next copy of a piece of software, whereas running big at-scale infrastructure services in all the clouds around the world is a lot of work. So as the percentage of the business that is cloud has gone up, and that's gone up at a very rapid rate. Obviously, the gross margin for that cloud product has improved quarter-over-quarter quite significantly to be able to have this very boring roll-up gross margin. And so we intend to continue that investment. And we think that there's an opportunity to really make that a very attractive part of the business. And of course, the focus as well on the net retention, which I think is that kind of efficient growth, okay? If you're bringing on customers and they kind of continue to expand with additional use cases over time, that's -- those are the easier dollars to get.

Question I was going to ask you is just there's been, I think, a little bit of just dialogue among investors around the right metrics on which to evaluate Confluent. There's been a lot of focus on the cloud revenue number and the sequential cloud revenue number you give us, RPO and the retention stats and kind of split those out between hybrid or cloud, on-premise platform customers. So what do you view as the most important metrics, like what should we be looking at?

Yes. Well, I mean, obviously, the ones we gave. Yes. So we pay attention to largely the same things that people do externally. So we feel like for us, it's not about cloud versus on-premise, both are important for our customers. Like we have a significant presence in financial services. Banks have big sets of older systems on-premise, those will die off over time, but it's over a lot of time. And that bridge between environments is actually quite strategic in this space. The purpose of this technology is to kind of act as that central nervous system that connects everything. So you have to be able to support all the modern environments that customers have, and you have to be able to connect across them. And so both are important parts of the business, but it's very important for us to show that success in the cloud. And we think that is a better way of delivering this kind of infrastructure service. So we look at, yes, the growth rate and sequential add of revenue in cloud. We look at the NRR across every segment of the business, the on-premise customers, the cloud-only customers. And then particularly these hybrid customers, which interestingly are actually the customers with the strongest NRR, those who are actually bridging across environments. And that's quite important. Those are a few of the kind of key stats that we want to make sure we're showing strength as well as kind of obviously, just overall growth, et cetera.

Time for just one more. I want to zoom out a little bit and go bigger picture. Some of the things we talked about during the IPO process. But just in terms of where this technology can head and end up from a product road map perspective. We've talked a little bit at times about the potential to move up the stack or put some more vertical-specific applications on top of Confluent. Are there opportunities to move further into storage or other areas along the infrastructure side? Or what would you highlight is, hey, it's not just stopping with event streaming, but this continues and what do the permutations look like?

Yes. Well, the first thing I would say is like, look, the opportunity in the core event streaming area is massive. Right? So we believe there is an opportunity to build a data platform and a business that's comparable in size to what databases have been. And that draws on a number of existing areas of spend. We think that is $60 billion TAM in and of itself, right? And then if you have that platform. If you're the place in a company where all the real-time streams of what's happening are, then you have obviously an incredibly strategic opportunity to start to move into some of the use cases for the technology and build on top. And this is the flip side of a new category. Downside of the new category is it's new and it's more work to understand it. The other exciting thing is, look, there's a lot of white space around anything that's new. And that is actually very exciting. So we recently released product offering, which is just around building data pipelines, and that takes a lot of these kind of ETL use cases or connectivity use cases and makes it something that is as easy as any product out there, but now sits on top of the same shared central nervous system that can power everything else. And we think that's one of a number of very common use cases that we can grow into over time. And real-time analytics, in security use cases, in logistics, in IoT that are kind of the common things. And then we think that there's an opportunity well beyond that, that we can't pursue directly as a company, but there's many partners and other companies that are really taking event streaming kind of out to all the industry-specific use cases. We can do some marketing around that. We can't build specific products for each of those, but there is a real ecosystem opportunity there.

That's great. A lot still ahead of you. Jay, I appreciate you spending the time with us here today.

My pleasure. Thank you.

Thanks for joining.