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

Jay, welcome to Communacopia and Technology Conference. This might be your first Communacopia and Technology Conference, I think.

That's right.

The conference itself was going to be 2 years old because we had the first version of it last year.

I haven't been here with this name.

Yes, exactly. Welcome. Rohan, congratulations on becoming CFO. How does it feel?

It feels great. And I've been with Confluent for 3 years, so a lot of continuity, but excited about what lies ahead.

That's great. That's good. Thank you, Gili, for joining me. As a fellow panelist. Gili and I worked together closely on Confluent right from the days of the IPO. So Jay, I think we first met many, many years back. You laid out the vision for real-time streaming, and now you're a public company. It's now close to $1 billion in revenue, et cetera. So just as you rightly visualized this from 2017, '16 time frame, what is your prognostication as to what you want the company to be like in the next 4 to 5 years? What are your goals for the company in the next 4 to 5 years?

The vision we hope we talked about back in the day, I think there's a lot of continuity, right? The role we're growing into is to be the central nervous system for data, so that everything that's happening in the company, there's a real-time stream of that, that's available. Any other part of the company, any software system can plug into that, can act on that. We think that's one of the most strategic platforms for data in an organization, and that's something we can build around over many, many years, right? Right now, we're just getting the full infrastructure stack to harness these streams and work with them. Over time, I think we can grow into a lot of the use cases that are emerging around that. And so yes, I think we're just in this kind of first phase of what's possible, but very excited about what lies ahead.

And talk to us a little bit about the preparedness of the company to execute on your vision 4 to 5 years. What are the things that you need from a go-to-market standpoint, product development standpoint, channels, et cetera to help achieve this goal.

Yes, there's a couple of immediate priorities. I mean, right now, the big focus for us is continue to invest in Quora, our engine which provides our Kafka service, make that something that's global scale, that's just a utility that can handle any stream anywhere in the world where it can flow anywhere across on-premise, different cloud providers, make that available. That's like the foundational layer in what's emerging around streaming. And then the next layer of infrastructure on top of that is around how you connect into all the different systems, how you govern these streams of data and then how you do real-time stream processing, how you can act on these build applications that work with real-time data without having to do a lot of difficult, cumbersome software engineering. We want to make this as easy as relational databases have become. And so those are the investments that are currently in flight, each have products that are in kind of different stages of adoption. That's that next wave that we're taking out to the market. And the last priority on our side is really leaning into what's possible with a best-in-class cloud go-to-market motion. As the proportion of our revenue that comes from our cloud product has grown, there's a lot more we can do to lean into consumption, to lean into product-led growth, to be able to accelerate how we're taking this out to market, help our customers use it for more things more quickly, do that more efficiently. Those are kind of the big priorities at Confluent.

And I apologize, I may have asked you this question 6, 7 years back, but what was the aha moment there? I know you worked at LinkedIn. You had this project. What did you see as the thing that stimulated you to develop this real-time streaming platform?

The big observation that was so weird about the world was -- I was working at LinkedIn. LinkedIn is a very digital business, right? It runs 24/7. It has lots of different software systems. They're all generating data continuously. And yet, our way of acting on that data or doing our most sophisticated data processing was to kind of store up all the data in each part of the company, extract it at the end of the day, put it in some central data lake, process it and ship the results back at the end of the day. And the reality was that was just very difficult to, a, get the data that you wanted; b, to actually act on it in the time that made sense and get it back in front of customers while it still mattered. It just didn't make sense. This kind of batch processing, it seemed like some weird hangover out of like mainframe computing where like the tape drive would spin. And so, we felt like surely there's a better way. And we went and looked at all the different technology that was more about the flow of data, the -- how you would act on it in real time. And indeed, there wasn't that much know. There was these kind of weird niche technologies that would do different types of data movement, but the vast majority of investment had gone into data storage databases, all the data at rest. And so it seemed like there was a huge opportunity around data in motion. And so that was what prompted the thinking. And then it was just a question of, well, how do you do it? How can you make something that can kind of plug all of these different systems together? How can you build something that's a basis for real-time processing? And that was more a continuous learning of like looking at what was out there in the computer science literature, looking at what had been done in previous systems and trying to put together a model that could be successful. And that was what led to the genesis of Kafka, which is an open source technology that actually predates Confluent and is the basis for our offering. And that kind of sparked a whole revolution in this idea of thinking about data not as something that's just static and stored, but as something that's a real-time stream, that kind of every activity in a business can be thought of as that kind of real-time stream generating data all the time and that businesses can react to that and use it as it occurs.

Got it. Rohan, shifting to you. I didn't say shifting gears. We have -- I've sworn that I will not use cliched expression. Shifting gears, segueing, double drilling -- I'm sorry, double click. Drilling? We're not going to use. So Rohan, turning our attention to you, tell our investor base a little bit about your background. How did you get to be CFO of Confluent?

Yes. So well, I started my career in -- more on the -- on your side, on the financial services side and made a pivot into operating finance roles about 15 years back. And in the last 15 years, this is stop #3 and probably the most consequential and meaningful stop for me. And in the last 3 years at Confluent, I've joined the company before the IPO and built our [ FPA ], Investor Relations and Treasury teams ground up. And I was doing a lot of our financial strategy in close partnership with Steffan and Jay over the last couple of years. So this feels like a fairly natural transition to me. And as I think ahead...

What a natural transition from making money and the pressure of making money is the nature. What was it like? I mean, so was it like -- you said it's a natural transition. I find it hard -- I mean, it must have been something unnerving along the way, right? Oh, I don't have to make money every single day. And so, now I'll just do the long-term thing. I mean, what was that switch like?

From being on the buy side or rather to -- yes, I mean, like to tell you the truth, it was 2008 when I was graduating from business school and probably not the best time to get into a job in Wall Street. So it kind of made sense to take an operating role. I like finance. I liked business and took an operating role. And yes, that's how I ended up here.

That's great. And what are your objectives -- or what are Jay's objectives for the CFO of the company? Just as he has this audacious goal for the next 5 years, as a CFO, how are you supporting the company?

Yes, I think when I take a step back, what's really important, when I think probably 3 things that matter. The first is driving durable growth. The second one is driving durable growth in an efficient manner. And you essentially do the second one if you have a real sharp focus on unit economics and how you approach your day to day. So those are probably the 3 pillars that how we think about resource allocation, which is ultimately, thinking about not just the next 12 months, but how do you sustain growth and put in money in the right places over the next 3 years to Jay's points earlier.

So outside of Steffan, who is your role model CFO?

Well, I am very fortunate in my 15-year career to work under, I'd say 3 very high-quality CFOs, started at Symantec Corporation is a person named James Beer who ended up being the CFO at Atlassian, is kind of a mentor to me. My prior boss, Kathy Bonanno, she's currently the CFO at Google Cloud and Steffan. So 3 very proudly accomplished people that you get to work with and learn from.

That's great. Good to hear that. Jay, real-time architectures have been typically hard to develop at scale at an economic price point. So help us understand the unlock that you bring to this market.

Yes, that's been exactly the dilemma. On one hand, it makes total sense. Nobody ever said they wanted their data like slower or, at the end of the day, everybody wants it fast and that's the natural pace of business. That's the way application software works. But how can you make that kind of easy enough to make it the default? And if you were to go back 10 years ago, people would have told you like, oh yeah, there's all these fundamental difficulties, right? You can't get kind of transactional correctness. You can't get the kind of throughput and cost efficiency. It's too fragile and difficult. I think all those fundamental difficulties have been checked off now. I think a lot of it now is that last mile of really turning this into a cloud service where you can just kind of consume it on demand, not something you have to operate and hire people to manage. And then really building the tools to build these kind of real-time applications and make that as easy as possible. So like the investments by us on the stream processing side are really bringing the same kind of SQL-based tools that people have had with databases for decades, bringing that into the world of continuous real-time data. That's the kind of thing that allows organizations to take the skills they already have and apply it to the new thing that's possible. And indeed, as that becomes true, this whole area of batch data processing, of bulk data, all that goes away and it kind of moves into the real-time world. And that's the exciting thing that's happening in our space.

And then, Jay, I was just going off of that a little bit and kind of entering the conversation here. I'm curious to get your thoughts on Kafka and that conversion to Confluent, right? So how do you first engage those customers that have elaborated Kafka use cases internally or on-premise and how did -- what's the technical process in actually getting them to be a Confluent customer?

Yes, it's a great question. So when we thought about the company, the open source played a really important role because we didn't want to be kind of going around and starting at the top of the organization and trying to pitch some big vision around data streaming as a new paradigm and then trying to connect hopefully eventually to some use case. You want something that's like a much more immediate fix to a problem that people have. And the open source enables that, right? We basically have something that is cloud native and a complete stack around data in motion and it's available in all these environments and we offer this as a service anybody can consume. And so we can come into an organization that is already using open source Kafka, and there's hundreds of thousands of those, or one that's thinking about a new project that will use Kafka. And we have a proposition where we have something that really has a more complete feature set, that has a better [ TCO ] proposition where instead of hiring a team to run it, instead of buying a bunch of upfront cloud infrastructure to run it on, you can get a service that's more cost effective and does more. And that's a really immediate value. And so for that first project, we're already better. And then as this spreads across the organization, it gets better still, right? This is something where the streams of data that are brought in for the first use case are often usable by the next use case that's trying to get that same data. And it's very likely that this is the only place you can get that real-time data in the organization. And that next use case will bring its own streams, which will attract more use cases, which in turn bring their own streams. And that's kind of the process of spinning up this central nervous system around data streaming in an organization. It's not a big bang. It's something that happens step by step as kind of real use cases make it out to production.

Thank you. And then Rohan, for you, a little bit on that is, as you just mentioned, thinking through these applications and it does have very real-time value and quick time to value. One of the things I'm curious to get your thoughts on, what you're seeing today and kind of the macro environment. I know that Confluent specifically has been talking about seeing smaller commitments over the last 12 months and customers getting are increasingly comfortable an exceeding those commitments. And how does that -- how is that trending today and how does that also marry with like the longer term vision of the company and the overall vision you guys have?

Yes, talking about the macro environment. I mean, over the last 12 months, we are seeing a couple of dynamics that were playing out. The first is we've been seeing this elongation of deal cycles, which was primarily driven by, I'd say more scrutiny and the CFOs getting more involved in deal cycle. And how that's panned out is when you look at the last couple of quarters, we've shared some commentary in our earnings calls around customers are committing to lower duration deals. The upfront commitment specifically for cloud deals is lower, which is not necessarily a bad thing for us, but it shows up in our [ RPO ] numbers. But purely from a cloud consumption standpoint, what we are seeing is our customers are consuming more than they are committing. And like I said, it's not a bad thing for us because we're doing the right thing for the customer. And when you couple that with our net retention rates of the cloud business of 140% plus good retention of customers, we feel pretty good with respect to the broad setup as we sit here today.

And then I mean, I think you just mentioned the cloud and the rate of growth that you've been seeing. And we've started seeing a little bit of a pace of deceleration there. And what do you think or what are you even evaluating internally and externally to give you conviction that that might re-accelerate or even pockets of unlock that you might see in calendar '24?

Well, our last reported quarter, we grew our crowd business 78%. We were very pleased with the results. And...

We're saying why not triple digit?

Why not triple digit? That's true.

Really excited about the cloud opportunity.

Well, 78% growth and candidly, at a run rate of $330 million plus. So broad brush, the business did very well. As we look ahead, of course, I'm not going to guide for the rest of the year or '24, but if you look at the cloud business and Jay touched on it a little bit, our data streaming platform and how we think about monetizing the entire data streaming platform will be very critical. Of course, it starts with the Kafka side of it. But then there is connect, the governance, the stream processing and our ability to share data across. And I think these are probably -- I'd put it in the category of the long-term drivers of the cloud business. And so those are probably the areas...

Because you don't have those capabilities in the Confluent cloud that you think the future that might lead to more driver of growth.

You said those will be the drivers because you don't have those features available robustly?

Yes. Each is in a different state of maturity. I would say for this kind of cloud infrastructure, there's a bit of an arc of kind of getting it to full maturity, getting it across all the cloud providers, getting it to work with all the networking types. And as you see that, you see unlock. And so we saw the same kind of S curve with our Kafka offering. We've seen very good adoption of some of these same features and functionality, and we're now kind of starting to see that in the cloud. So if you think about those 3 things that Rohan just mentioned. Connect is kind of just hitting that up into the right period in the cloud. And we think that, that's going to be a nice driver through next year. Next up is our governance offering that just -- we just released the paid version of that, which has grown faster than any product offering we've ever had. And we think that's on a really great trajectory with a lot of exciting features coming. And then last but far from least, we are launching a Flink offer and this is an early access -- it's a technology called Flink. This allows real-time processing of data streams in a very powerful way. And this is in early access with customers. And we think that, that can be as big a portion of our business as the core Kafka offering that we have today. So we're very excited about where that's going. So I would think of each of those as a kind of S curve where they're just coming out to the customer base and we think we'll drive growth both next year, but into the years ahead and be a very substantial driver on their own right as kind of line items that are consumption revenue drivers as well as drivers of the core business, like each of these helps bring new streams into the platform, helps you generate new streams off of that. And so they will generate more Kafka usage as well.

Yes. So I mean, I think that, that really leads to the question I had around Flink and your TAM opportunity. And you guys talked about this at your Analyst Day and just thinking through the other components of your TAM and what catalysts or unlocks do you see to really get there? I know that you just mentioned a lot of those Connect and governance and Flink and how would you see those expanding over time and also maybe even attracting customers?

Yes. I think there's an amazing opportunity in this space. I talked about the fundamental setup that all the investment had gone into data storage, databases. This area of kind of data in motion is kind of largely white space. There had been a connection -- like a collection of kind of older point technologies, the message queues and [ ETL ] products and application integration layers. And they each kind of solved a little bit of the problem, but in a very limited way. And when you look at what's happening now, all of these niche categories are coalescing into something that's much more broad and much more powerful. And it draws not just from data movement, but some of the underlying databases that would be used with that movement. The common pattern people would have is copying a bunch of data into a database, running some batch processing on it, shipping the results somewhere else. And all of that's kind of moving into this continuous kind of data in motion, data streaming world. And if you put all that together, there's a $60 billion TAM. I think one of the most exciting new data platforms in decades that is emerging and the basis for a whole ecosystem that's emerging around it of new technologies that plug into these data streams and use it. And I think that's why there's so much excitement about this area among customers, among technologists, just kind of broadly in the industry.

Yes, that really leads me to your partner ecosystem and something that you guys have definitely been expanding over the last year or so, even longer. And so maybe touch on what's been really beneficial from those partnerships and also maybe, Rohan, for you, how they've impacted your economics both on the deployment side and the go-to-market side?

Yes, I can start. I mean, we have -- I would say the early days of strong partnerships across SIs, some of the technology ecosystem around us and then maybe most importantly the cloud providers. Our cloud offering is not something in its own environment. It's something that's in AWS or in Google or in Azure. So working in a really tight way with those layers, all of their technologies top to bottom, and then selling through their marketplace, cooperating well with their go-to-market teams, that's all essential to doing a good job by our customers. And I think all of those have started to contribute in a meaningful way. You may speak to more of the potential there.

Yes. On the leverage side and how we think about the overall economics, I think in general, if you look at software companies and the partner ecosystem, it helps drive non-linear leverage as well as top line growth if you really get it right. And as Jay mentioned, our partner ecosystem opportunity is probably in the early innings where we are, we're pretty early days. And looking ahead as we think about leverage in the broader business, sales and marketing and go to market will be an opportunity and within that partner ecosystem will contribute a meaningful part of it over the long term. So yes, that's how we think about it.

Jay I wanted to ask you about adjacencies. I think something that Gili asked you about. Sometimes Oracle relational database becomes a pretty big data market. Otherwise you have to imagine adjacencies. And the idea is that what are the other customer problems that you're not solving today that you could be solving that are adjacent to the Confluent data streaming platform? What kind of opportunities do you see on the horizon to do things?

Yes, I would think about this in 3 phases, right? First of all, there's core hub of all the streaming data in a company, that kind of core central nervous system, that's a huge opportunity in its own right. And we're pretty early in just capturing that, even just going out to all the existing open source users and getting them on our cloud offering. And it's not like that open source user base is fixed in time. It's growing, right? So that's the first opportunity. The next opportunity is expanding that functionality, like bringing real-time processing to bear, making this something that it is a central part of your governance strategy for data across the organization, hooking into all the parts and unlocking it. That's that second wave of functionality that we talked about that's coming out now. If you look at what's next after that, it's really getting into the use cases that people are adopting this for. This is a key component in the stack for real time analytics, for security, for working with kind of IoT projects or things that bridge out into the edge of the real world, projects around logistics, projects around all the kind of intelligence about the operations of your business, projects around your customer and personalization and how you interact there. I think there's opportunities across all of that for us to add more value and grow into that. And of course there's also partners who are helping to complete that picture with our customers. So I think about our growth in kind of 3 layers, really completing each of those. And each of that is kind of in a different phase. The core of Kafka is broadly adopted and becoming a kind of de facto standard. We're well into the commercialization journey there. That next wave of infrastructure is just kind of reaching the market. And beyond that, I think there's a lot we can pull in on top.

As people build custom applications on the platform, does it become easier to upgrade to the cloud or is it a little tougher? How do you go through the process of upgrading?

I think everything we do adds to the argument for our cloud offering. As we add more functionality, each of those is more things that the next customer gets out of the box as they adopt. As we started, we had something that was just kind of a bare bones thing around Kafka, that was already pretty good for a lot of people because they just want something that's managed for them. As we've expanded that with more capabilities, it's become increasingly compelling. And there's now many reasons to move towards it. And so I think that journey continues as we have more functionality. Customers will of course build directly around the APIs in our platform. That's not a bad thing. It's actually a great thing. It's one of the things that makes us sticky and durable, and it's also one of the things that teaches us what to build next. If you look at your smartest customers, whenever they're building some kind of internal layer around you, there's a good chance that many other customers would need that same thing. And that's something you should turn into kind of a fully supported area of product.

Got it. Confluent Cloud, where do you see the -- I know that at this point in time we have a paywall. It's for a certain set of use cases. At what point does that become the default cloud or default Confluent product that customers naturally gravitate to? This is the first evaluation.

Yes, you're asking like, when is our cloud kind of the default starting point? I think increasingly that's happened, right? I think a lot of this is kind of an overall mindset shift in tech much more than anything Confluent orchestrated on our own. I think it used to be that customers went and looked at open source things they could kind of download and build a team around and run internally. I think increasingly that seems like a lot of work and customers are looking for a managed service that just kind of gives them some superpower around data that they can have instantly, that's going to be world class. It's going to be available around the world that will be operated perfectly and they don't want to wait for any of that. And so I think that's a shift that we've helped enable with our -- in our products by making our product meet that bar. But I think it's a mindset shift that's happening in every company that builds around technology that's actually made it more of a pull than a push.

Got it. And I had a quick question for you around Generative AI. You spent the last 30 minutes on a software session without talking about Generative AI. So I think that's a new record. And I know that it's not necessarily a direct benefit to Confluent in the way that you've spoken about it. But I'm curious to get another view from you guys in the sense of if that's changing the conversations or the nature of the conversations that you're having with customers as they're reevaluating the data that they need and the real-time nature of these LLMs and the outputs that they're looking for.

Yes, I think that's exactly the reason this is a driver for us is companies are looking at the opportunities with these large language models, and they're looking at all the data they have locked up in different systems spread across their organization. And they're thinking about how to put those two things together. And these AI use cases really start with that data unlock. And increasingly, the way that you would combine the two does involve data that's up to date about your business, about the context of the customer that you're interacting with. And that does require this kind of real-time streaming. So a good example of a customer that has done this that we shared in a past earnings was a major travel provider that was integrating all the information from their systems about where you were, what flights you booked, what was on time, what was late, and being able to combine that with a large language model to answer questions intelligently with our customers, this is a lot better than kind of waiting on hold for a long time to talk to a human. And -- but it's very important that, that information be accurate, not hallucinated, that it be up to date with your business. And you're seeing a whole set of technologies that we partner very closely with, the kind of vector databases and different data stores that would ingest these streams and try and serve it up for models. And that's a very tight partnership and integration we have that's showing up in all of these different architectures to try and enable this next generation of AI-enabled enterprise use cases.

And how important are vector databases and vector search capabilities for Confluent? Do you see yourself getting involved in that?

No. Right now we're just partnering. There's a broad set of different providers that are trying to add vector indexing some of the existing databases, like Elastic or MongoDB and then a bunch of specialized databases, the Pinecones and [indiscernible] of the world. Rather than try and beat them all, we want to integrate into all of those. If you think about what we're enabling for customers, it's first of all, unlocking data by connecting to all the systems you have, enabling that flow, being able to process it into the right form at the right time, and then being able to hook into all these technologies so that even for our customers, they don't have to pick the winner in a very confusing space. They can use what works for them today, but they can switch in the future if something new comes along.

Got it. Any question from you guys? We still have a minute. If you have a question, just raise your hand. Mike's coming down. Yes. There's a question here on the other side of the room here. Yes. Thank you.

Hello. Thank you very much for coming to the forum and spending the time. I have a question on the competitive landscape. So I think, like just recently, Mongo has announced their own streaming product and then I think there is a smaller player called Redpanda which is claiming to be cheaper than Confluent. So it seems like there are a bunch of people coming at a different angle into this market. So would love to see -- love to hear how you think about their entrant and how does that impact our business down the road?

Yes. I would say broadly it hasn't had much impact. It's natural as an area gets big and exciting, you're going to see more people try and get some part of it in different ways. There's been a series of kind of early start ups, including Redpanda, that are just kind of at the beginning of the monetization journey that will have products in the space. There's been a sequence of those that would include technologies like NSQ or Pulsar. Most of them haven't really gotten to the level of traction that Kafka has, and none has gotten anywhere close in terms of commercial scale to Confluent. In the stream processing space, this is an area where we partner very closely with Mongo. There's very commonly architectures where data is flowing from us into them. There's processing that can happen on both sides. The power of our platform and our model is twofold. To make this kind of streaming data easy, easy enough to make it the default, it's really important to combine the processing with the stream itself. That's what gives you the transactional processing guarantees. That's what makes it easy to kind of secure, end to end. That's what has made databases so successful. Database combines storage with data processing. In the streaming world, it's about combining the stream with the stream processing. So I think that's something that pulls that into our layer. I think the other thing that naturally encourages this is that data doesn't just go to one place, it goes to many places, right? Data goes to MongoDB, it may also go to a data warehouse or a data lake or to a Generative AI service or out into different SaaS services. And so when you think about the kind of clean-up and processing of data, it doesn't really make sense to try and replicate that in each of the destinations. It's much more likely that that pulls upstream so that these reusable streams can flow to all of those destinations. Those are the two biggest reasons that the stream processing world has maintained this very close attachment to Kafka and the reason I think that will be a very successful part of our business as well.

On that note, we're out of time, but thank you for the question. Thank you, Jay and Rohan, for attending the conference.

Thank you so much.

Thank you for the discussion. Much appreciate it. And have a wonderful rest of the evening.