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

All right. Thanks, everybody, for joining us. I'm really excited for this one. It's one I've been looking forward to for a bit here. It's Confluent, a company that I've known for a long time and we at RBC use it internally here. And so we're excited to have Jay with us today and joining Jay on stage, Anurag Sehgal. You can introduce yourself in a second, but he runs kind of all things data for us here at RBC. And so I thought it'd be a great opportunity to not only talk to Jay about some questions, but also get sort of a kind of an end user's perspective to the relevancy of Confluent and Kafka and Flink and all things data. And so that was sort of the agenda here today.

So maybe, Jay, I'll start with you. Let's just start at a very high level. There's a lot of things that we can talk about following the quarter. But I think your data -- your view on data streaming and now data processing, it just resonates with like a lot of folks. But the question I still get is like, isn't this just like TIBCO 2.0? Like how is this so different than the data piping that I think a lot of us who don't have a lot of hair remember?

Yes. Yes, I think it's a great question. So the -- a little bit of the history of the data space. There was a bunch of little products in the storage area that really kind of converged around relational databases. And we ended up with something that was very powerful in general and fully featured, could guarantee correctness, could support complex workloads, really a general purpose tool in that space. But if you think about the areas of data movement or flow, that didn't really happen there. And so kind of data at rest, converged, powerful, broad platform, a ton of computer science that went into it. Kind of data in motion, it's more like a hodgepodge of 12 different categories. So you've got these kind of message buses, which are fast but very simplistic, maybe not really scalable or able to support the kind of complex processing. And then you've got ETL products, other kind of application integration, really just a hodgepodge of different categories. If you kind of squint, they're all sort of doing the same thing, which is handling data as it moves, but they all have kind of pros and cons. Some are fast, some are able to do rich processing. Some are able to kind of scale out, some support applications, some support kind of custom software, but none do it all. And the result is you have this kind of weird set of small $1 billion companies that each serve a portion of the use cases. And then from a customer point of view, it's really annoying. Like each piece of data has to be kind of sent through 12 different pipes. Each of those pipes has different limitations and problems. What happened in the streaming space is actually, I think, starting with a revolution in some of the underlying computer science of how do you think about this stuff. How can we come up with something that is more general, more scalable, is able to provide strong guarantees around the ordering of data, is able to scale to the size of the company? That suddenly means you can have the same framework handle the integration between applications as well as the core database changes that are flowing through, and it can handle delivery into analytics systems. It can handle delivery to other things. And it can be a basis for very rich, sophisticated real-time processing that can start to have some of the same guarantees that databases would have. So what does that mean? It means basically a convergence, right? You're seeing kind of a very similar thing to what happened in relational databases, but now happening not around data at rest, but for data in motion. And that opens up a whole new set of use cases as well as just making a set of things that companies have always been doing to connect applications, to connect different data systems, making that a lot easier and able to be done kind of at scale in organizations. And that's a really important part of the overall evolution in companies. I think if you go back, whatever, 10, 20 years, a lot of the focus was how do we build in application. But increasingly now, a lot of the problems that companies are dealing within software is how does it all come together. How do the things happening in this part of the business feed the customer experience over here? And how does that flow into the AI tools that we're trying to build out over here? And how does that come back into our analytics over here? It's now very much kind of 1 system that's all interconnected. And how you do that is the next set of challenges both in the data space, but also elsewhere that I think companies are kind of grappling with. In the data space, this has been a big driver of the rise of streaming and of Kafka.

Maybe I've got a couple of questions -- it's more than a couple on sort of the results from a week or so ago. But maybe before Anurag, for those who don't know you, introduce yourself quickly and then maybe give your perspective on what Jay just said and maybe how we think about Kafka and Confluent inside of RBC.

Absolutely. Hi, everyone. I'm Anurag Sehgal. I lead our client banking, data analytics and digital tech solutions at RBC across Capital Markets. I've been with RBC for about 18 months. Prior to that, I was at Credit Suisse where I also worked with Confluent. We had a really good partnership with Confluent at CS as well. Yes, look, I mean, maybe the context of -- and I'm sure everyone is aware, but there's just an exponential demand for data analytics. We talk about GenAI and a lot of that has to do with data. As the demand for analytics has grown, the demand for more real-time analytics has grown as well, right? I mean there's so much real-time data that you only can get in edge if you're tapping into that data, whether it's real-time client analytics or whether it's alternative data-driven analytics or our research teams or our clients or for -- if you look at hedge funds or long-only investment managers in financial services, a lot of them are looking at real-time insights and companies or macroeconomic equations. And so -- and by the way, they're doing that because there is so much real-time data available now, either through mobile apps or through e-commerce platforms or through satellite imagery or through everything that's going on around us with IoT devices, connected cars. All these devices are generating billions and billions of events that are going somewhere, and they need like that sort of data and motion to get to what you really need to see as an outcome from that data. And so as that demand has grown, really Kafka as a real-time streaming service has just continued to grow in many forms. And so we at RBC in Capital Markets leverage almost 5 different services, as you mentioned, for real-time capabilities. And we use -- even within the Kafka domain, we have a lot of teams that are using open source Kafka. There are teams that historically leveraged Horton was Kafka. And a lot of teams are now moving to more and more resilient needs that they have, are moving more and more towards Confluent. And so I think because of the resiliency and the administrative capability and many other wrappers that Confluent provides around Kafka, and that's been the mover for people to move towards it. There's still a long way for us to go because we also have Solace and many other technologies over the last 20 years that have sort of grown exponentially. And so I think there is a path ahead for us in how we think about this space. And as we look at moving more towards the cloud as well, I think Confluent will play an even bigger role for us in the coming years.

Super helpful. A couple of comments from the quarter a week or so ago. Jay, on the call, you noticed there was a particular customer that is moving back on prem. And I think that surprised a lot of folks, and Anurag and I talked about this yesterday. I think, Anurag, you said, I mean, sometimes it's a negotiation tactic that customers will use against vendors. But maybe on that point of that customer, can you talk about who has the ability to actually do -- I mean we use Kafka internally, but who has the ability to run completely Kafka sort of stand-alone without some sort of paid version? And ultimately, do you think that customer, is that a potential win-back customer when you look at an on-prem platform?

Yes. Yes. I mean there's still a good chunk of open source usage in the kind of large tech companies and so they have the capability. It may not make sense. Certainly in the cloud, the TCO of these managed offerings, I think, is just much better, building out a team of highly paid engineers to do it all yourself is not that compelling. The movement back into data centers, I don't think that's a broad trend. I think it was specific to this company. We haven't seen that across our customers. Yes, I do think it's a potential win back, but this isn't about -- was it a decision that was based on Kafka or Confluent. This was a decision about their overall environment. And it was borne out of a pretty significant set of outages they had with another vendor as well as a pretty dramatic change in Kafka profile they're trying to orchestrate. And I don't think moving out of the cloud is necessarily a good way to save cost, and there's probably a whole other set of companies that are moving into the cloud as part of the cost optimization. But at least in this case, that was the cause of it.

Interesting. The other question that I get from a lot of folks too is, are there things that Confluent is thinking about doing to make it even more advantageous to move to the paid version? In other words, I know you want to try to drive as much functionality in Kafka as you can, because it's a great funnel into Confluent. But are there things that perhaps in the future, might make it even more challenging for customers to sort of trade down to Kafka?

Yes. Yes. I mean the areas where we've built a lot of advantage and continue to invest is like, first of all, just build a true cloud-native offering. This is particularly true as in the managed service, but that ability to have something that it just comes, scales up and down elastically where you kind of just pay for what you use, where you can really treat it as a utility. It may sound like a small thing, but the difference between a Snowflake and a Teradata is really that kind of cloud-native operations. That's the first pillar, which I think is a huge thing and it's just impossible to do with a kind of do-it-yourself open source or really any other software, self-managed software. The second is in this space, there is really a convergence of capabilities that is, if you think of Kafka as the pure stream of data, there are -- increasingly companies need all the connectors that plug that in. The kind of real-time processing capabilities where we're bringing Flink into our platform, so you can do kind of rich SQL query. SQL being the language of traditional databases, but also programmatic access to really build rich applications around it. This has become an essential capability for companies as they're working with real-time streaming and want to make that really easy for their engineers as well as the kind of governance of streaming data across. So that kind of a complete data streaming platform, I think that's the other area of difference. And then making this something that's available across these different environments, so that companies that are hybrid that have some presence in their data centers, have something in the public cloud, maybe even across multiple clouds that, that can all link together into 1 fabric for real-time streaming. Those are all the areas of kind of differentiation. I think it's quite substantial. So like when I think about, hey, overall, you can have -- with any of these open source offerings, you could have somebody who kind of stays with the open source or you could compete with different managed services. For us at Confluent, I think the focus is really very much on that managed service area. I think that over time, the idea of kind of doing these big data services and systems in-house, I think it becomes less and less appealing to all but the very largest tech companies that really want to kind of do that from scratch with very heavy investments. And even for them, the question is, are they really winning out? And I think you're seeing that trend more broadly. I mean even within the digital native cohort that had a very significant open source Kafka usage, I mean, we are seeing a move of that class of companies into our managed service, and it's because it ultimately isn't that cost-effective to try and run everything yourself. Notwithstanding this one company that's actually moving out of the cloud. I mean in the cloud, why would you want to try and replicate one of these massive multi-tenant services? Like there's huge economies of scale of doing this across customers, both in how it operates and your ability to build it. And so I think that's the ultimate driver. So I think that's a positive force for any of these offerings that are built around open source. Traditionally, on-premise, you might look at the overall conversion from open source to paid as maybe being around 5%. I mean, it sounds like even at RBC, right, there's some Confluent staff and there's some other stuff, but it's really a mishmash. I think in the cloud, that changes quite significantly. I think it is TCO negative effectively at any scale in the cloud to try and do it yourself. And over time, that proposition becomes better and better, like the investment in R&D for a managed service that goes up, whereas the open source is kind of still just the open source. And for each company to try and replicate that internally, it gets harder and harder to match, right? Maybe that was something that was doable when Confluent had 50 engineers working on the cloud service, it gets much harder when it's 700 engineers working on the cloud service to kind of match that level of investments and the quality of the product and set of capabilities. And so, I think overall, like we're in a world where the commercialization of open source is going from something that maybe traditionally was kind of like 5% of users to something that's probably going to be closer to 90 plus, maybe accepting some of these very large companies that, for whatever reason, want to kind of stick it on their own, but even those I think we're seeing changes.

I think they will, right? Because at the end of the day, like the question to be asked is if you -- let's take the cost factor is one factor. But beyond the cost factor, is there anything differentiating or is this just like why do you need to apply your high-end engineering resource on doing something and engineering a solution around open source when you have something that is as robust and resilient, right? Like that's a question that we ask ourselves on every single aspect of this, even if you look at Airflow or Spark, like the same question that comes up, right? So that's one part. But then the second part is, okay, well, this is going to be more costly. Okay, well, when you add up the pieces of maintenance and vulnerability management and upgrades and like we don't always think about that as cost when you look at -- okay, well, we have people that do that. When you actually look at the total cost of ownership of these things, there's no way that going for something that provides all of that as a service and takes the headache out is absolutely the right thing to do. I guess my question for you, Jay, would be more around pricing and how you think about pricing strategy in cloud, on-prem, like how are you evolving that especially for large clients like us that are in this sort of diverse equation?

Yes. So especially for the cloud, one of the things we've started to do is really exactly what you described, which is just work with customers to come up with kind of a shared view of, hey, what's the cost for you to do this with the open source? And what's the cost to get the managed service? And if you fit in that envelope, it becomes very compelling for most customers that were like, hey, I get something that has a lot more capabilities than I get it for less money than I would spend doing it myself on cloud infrastructure and people and everything else. And then our pricing, we want to make it clear what the cost structure is and make sure that we kind of fit within that. And I think that's the thing that ultimately makes it an easy decision. If it's something where it's better product, more money or worse product less money, then you kind of have to weigh the pros and cons. I think if it's a better product for less money, then that becomes something that's a bit more of a no-brainer.

So it's a good segue into, I think, something that -- after your earnings call, I assume there was a lot of confusion about kind of your go-to-market evolution. Because I think a lot of people interpret it as revolutionary or massive disruption. Maybe can you level set us for what's changing? And maybe more importantly, what's not changing? Because I think the reality is, I think it's the right thing to do. And I just think it's my view, it's less disruptive. And I think the market [ struggled what it ] read.

Yes. Yes. I think that's probably true. So all of the cloud infrastructure companies have gone through a change to be more oriented around the consumption of customers. And so the first part of that change for us was really having a consumption model where customers pay us for what they use and they have the ability to lock in a commitment to some amount in exchange for discounts. So we made that kind of business model change maybe 3.5 years ago or something prior to going public. And that was a very big change, right, because just how you're taking money from customers changes everything, all your systems change, how your billing changes. The change we're making now is much smaller than that, but it's actually quite important, which is how does our go-to-market internally function? Are they motivated in goals around selling the kind of commit? Hey, you're committed to x dollars in spend. Or are they motivated in goals around that actual contracts and consumption coming online? And this is a change that I think all the infrastructure companies have made to positive effect. And it tends to be good for customers because the value for them comes when they realize a successful project and they're kind of getting value out of it. It's good for Confluent because that is actually when the revenue shows up for us. And so you want to have your go-to-market focused on that. It's definitely a change and that it encompasses both sales compensation as well as kind of the systems that manage and run pipeline and that you run the business by. But it's not revolutionary change in our business model or something like that. And it ultimately aligns to what it is that we want to do, which is help customers build a lot of successful projects around the technology and successfully take those to production. So I think that's the change we're making. Companies have done this in different ways, but I think all have seen positive impact. And we factored some disruption into the first half of next year for this, because any time you're making go-to-market changes that has impact. But it's not like significant thing where we need to make big personnel changes or we're somehow changing our business model or there's some multiyear shift. It's just kind of ultimately a change in what the end goal is that the go-to-market [ orients here ].

One of the -- a skeptic would say, is it a demand problem? Is it customers just don't want to make these big commits and therefore, it's sort of reactionary to what they're saying?

Yes. I think here's what we've seen over the course of this year. I think a lot of these vendors have made this change. And I think some customers did end up overcommitted to certain either cloud providers or other vendors. And they were unhappy with the result. We didn't have that problem. So like when we look at utilization against our commits, we've consistently seen consumption above that commit amount. But nonetheless, over the course of this year, I think there was a lot of pressure on IT budgets and a lot of scrutiny. And so if your motion is, hey, flesh out all the Kafka projects that may happen over the course of the year or migrations that may occur and then size some very large commit that we'll cover all that for the next 2 years, and then I bring that to your desk. The first question you're going to have is like, well, how sure are we about all of this? Like are these things really going to happen? Do we know exactly how much Confluent they're going to use? And it kind of gets picked apart and gets harder and harder to get done. An easier motion is prove out the TCO and help bring workloads on and take the commit up at a customer-driven pace. Some customers may want to commit ahead for a bigger discount. Some may want to be more conservative because there's risk in their own time lines and plans and letting them drive that as much more successful. How does that help us? Well, it basically orients our go-to-market around the useful thing, which is actually finding these workloads where we can add value, making sure that we're attached to those, making sure that they succeed and get out there. That's the thing that ultimately is going to be valuable to our customers. And it's the thing that ultimately like brings Confluent the revenue. If we're spending all our time negotiating some big upfront, then we're just slowing down the rest of the work.

And I think to a customer like RBC, we may not even notice a difference. But I guess when you hear Jay talk about that philosophy of incentivizing sales, how does that resonate with you, Anurag?

I mean totally resonates, right? Like if you think about any of the tech partners that we operate with, the early stages of that, there's no way we're going to commit to a multiyear deal. We have to get familiar with actually like leveraging the technology, understanding the partner, being able to influence the partner on their product road map, having a stronger relationship overall and be able to sort of crawl, walk, run, right, and be able to scale from one program or product to multiple products. And so you can't really think about multiyear commits upfront, like that just doesn't make any sense. So invariably, we take that approach crawl, walk, run. And at some point, as we are starting to get to jog a bit, we start thinking about, okay, how do we scale this relationship now because we think there's significant demand for us, for this capability and that we've actually penetrated training and awareness and knowledge sharing across this capability. And so I think that's absolutely the way to go. I mean you would never be able to commit something upfront for many of these things.

So it seems to me that it -- the way I interpret it is that it's a potential accelerant of new business. But that for some of these big customers that are already strategic, like they're using most of their commits anyway at this point.

Yes, I think that's exactly right. So we prior to doing this, we went through the first year of this year. We're basically accelerating the next 2 years kind of pulling that in. We talked to the peer companies that have been through it. So each of them saw this as like net positive for growth and then net positive in just their kind of interaction with customers. Like customers are happier, higher success rate, et cetera. So I think there's a little doubt about the outcome. When we talk to companies about like how disruptive was the change, that varies. So for some companies, it was very smooth. For some companies, there was a quarter or 2 where they're kind of getting their arms around the new system. Obviously, our intention is to make it very smooth. But when we plan, we want to be conservative [ with final arrangement ], some impacts from that. And so that's how we've thought about it.

The other -- just -- we only have 5 minutes left. We have a jam like 30 minutes into 5. Competition is one that comes up a ton in my conversation with investors. And we all know the hyperscalers and some of the cloud data stores, but even smaller companies like Redpanda comes up a lot from an investor perspective. Could you address maybe some of those questions around is the -- like let's take Redpanda, they're a smaller vendor. But like how does your technology stack up to some of these other competitive offerings that maybe would consider not down to Kafka, but maybe an alternative?

Yes. Yes. I mean probably the surprising thing in this space has been how uncompetitive it is actually. If you look at the database market, it's kind of sliced up into 101 vendors. And there's probably a new database company every 6 months. The streaming area hasn't been that way. Kafka has been extremely dominant since this kind of whole paradigm emerged. There's always been something new that comes around. So a few years ago, it was Pulsar and probably the same conversation and everybody would be like, this Pulsar thing is going to kill you and it'll -- and it didn't happen. And I think it didn't happen for a whole set of reasons, which is the world is kind of standardized around Kafka. The new vendors now have given up on any kind of real innovation outside of that. They're just implementing the Kafka protocol, which is kind of a weak position to come in from. And at the same time, the set of capabilities that people want is now Kafka plus other stuff. A successful streaming platform is now Kafka, the connectors, governance, and kind of Flink and stream processing capabilities, all working together in a holistic way. So it's a much bigger product. And I think it's harder for a start-up to go after one slice of that and do it really well because ultimately, people want all those things to really work like one thing and they want it across the full spectrum of environments. And so doing one of those things well is hard and takes time, doing the full package together is harder and takes investment at some scale. And doing that across the different cloud providers and on-prem well as a first-class cloud product and software offering is harder still. But that's kind of where the value is, and that's what customers need and want. And I think that's ultimately the kind of barrier we have. It's like a lot of hard software that we've done and taking that out across the environments where customers need it to make them successful.

The other thing that I think is really intriguing about the story right now is the opportunity to accelerate growth. You noted a number of things on the call, Flink, FedRAMP, obviously, anniversary-ing this slight go-to-market change. What are -- what would you tell folks out there in terms of like if you stack all these things -- GenAI, goes without saying. What are you most excited about for the future that we're sitting here a year from now, were like, wow, that was a real opportunity to look at this story a little bit differently with these new opportunities?

Yes. I mean, look, the exciting thing in this area is there's a major new paradigm around the use of data that wraps up a whole bunch of problems, people having a lot of spend with the $60 billion TAM that is really just coming into being now. It is getting broad and rapid adoption across virtually every type of company across a broad set of use cases. And that doesn't happen that often. When you have a new kind of data platforms, they're usually either the next-gen of an existing thing where it's a little better, but it's kind of the same thing you had before or it's relatively narrow on the edge and I'm sure it has a couple of use cases, but it's not broad. So this is one of the few examples where there's kind of deep value across a broad set of use cases. And I think that's a great position to be in. I think we're incredibly well positioned in it. And then as you said, yes, mechanically, there's a bunch of things coming online in the business that we're super excited about. The kind of rest of the data streaming platform around Kafka. This has seen a ton of early success, and we're very excited about what that brings to the business. Some of the security unlocks for highly regulated customers. FedRAMP in the cloud for kind of public sector, but a whole set of things across financial services and elsewhere as a lot of the banks are shifting spend into the public cloud. That's a huge deal for us. And yes, the kind of AI tailwinds is probably harder to put a firm number on, but like the reality is there's a ton of investment in data flow and making use of data and a whole new set of applications that are kind of springing up in our customer base around that. I think that's very exciting. So yes, there's a couple of different moving parts, but I think that set of tailwinds is really impressive.

So yes, the -- maybe just to wrap up here in last minute, obviously, the reaction to the print was pretty extreme. I think a lot of -- it doesn't -- when I talk to investors, there's still this very bullish view on you being an important conduit for the future of data and movement and everything that sits on top of that data streaming process. If we were to sort of like -- sort of wrap up with a final comment on like your bullishness towards the future and that this is not just this decelerating growth story from here. How would you leave it with investors around the future?

Yes. Look, I talked a little bit about the market. I think that's a really unique opportunity, and I think we're uniquely positioned to go get it. Like there's no other company that's as well positioned as Confluent to really capture that by a lot. And so the -- with any new market, of course, it's -- there's more uncertainty and there'll be a tendency to draw a line between the last 2 data points and project it out to infinity. I think that's natural because it's not an established thing yet. But I think at this point, it's becoming established enough. And so you can talk to the customers, you can talk to the users, you can talk to the technologists, you can hear about the use cases. You can look at the overall growth in the space and kind of convince yourself that there's something really remarkable happening here. And that that's a foundation for a major new data platform and a ton of innovation around that over time. And I think if we get that right, that's an incredible opportunity for Confluent to grow for decades to go.

30 seconds, Anurag, does that resonate with you as an IT profession?

Absolutely. Absolutely. I mean I think we're going to talk more with Jay and team on some of the consolidation opportunities we see, and I think there's a lot to be done here.

Excellent. Well, that was the best 30 minutes I've -- we needed 2x there. But Jay, really from all of us at RBC and Anurag and the IR team, which is somewhere around here, I don't know, maybe in the back. Thank you, guys, and best of luck, Jay, in the future.

Thank you.