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

Good morning, everyone. How is everybody doing? Yes. pretty good. All right. Good to see. Day 3 of the conference. So congratulations for making it this far. Hopefully, you have a few more sessions in you. Very happy to introduce the CFO of Confluent, Rohan Sivaram. And Rohan is going to go through some slides. But before I do that -- before he does that, I'm required to inform you that a complete list of research disclosures or potential conflicts of interest is available on our website at williamblair.com. We are going to do a breakout after this upstairs in [ Adler ]. But we'll have some time for Q&A here once Rohan goes through the slides. And Rohan, thanks for being here.

Thank you for having me. Good morning, everyone. As Jason just said, Rohan Sivaram, Chief Financial Officer at Confluent. And I'll be spending a few minutes just doing a quick overview of the problem we are solving and how are we solving the problem and just cover maybe some high-level financials. Before I jump in, the legal disclaimer. I just want to put it out there for a second. Okay. I wanted to take a step back and really understand the data landscape today that organizations are seeing. Today, every company is a software company, and every company is a data company. And how they harness their data actually defines if they're going to succeed or not. So with that as a backdrop, in today's organizational landscape from a data perspective, you see 2 estates. There is the operational estate, which is technically used to run the business, and then there is the analytical estate, which is used to analyze the business. The operational estate essentially comprises of the large applications. You can think CRM, you can think HCM, you can think ERP systems. And they have their own databases. And what typically happens is you have these application integration tools that make it all work together. On the analytical side of the world, you have data warehouses and of late, you have the data lakes, lake houses that essentially what happens is through some kind of an ETL, you suck in all the data from the operational estate, move it to the analytical estate and try to piece together what the world actually looks like and which gets extremely complex. With the advent of data products, this picture looks even more complex with more and more integrations happening between different systems, between different applications. And where we end up is what we call the giant architectural data mess. And candidly, when you look at every organization, they are dealing with this data mess. And just the operational estate and analytical estate in silos is not a workable long-term solution. So at Confluent, we're actually thinking about this problem slightly differently. And we are thinking about this problem with our data streaming platform, and we're a category-creating company, and our mission is to set data in motion. And we want to do that by helping our customers do it efficiently, helping our customers do it in an economic manner and helping our customers do it in a safe manner. So that's our mission statements, and that's the problem that we are solving. How are we solving the problem? Essentially, we're this central layer of data where all the events are little things that are happening in the organization are captured in the central layer and different applications can actually react to the real-time data that's in the central layer. We're essentially bridging the gap, the two silos between the operational and the analytical estate. The problem statement that I just spoke about. And how are we doing it? We're essentially doing it with our data streaming platform. We have this -- we have this platform, which includes streaming of data. We have a whole set of connectors that connect data from different parts of the organization and bring it back to that central layer. We have stream processing that analyzes those streams of data and then we have governance, we make the data safe, the right people have access to the right data. And that's how we're actually solving the problem. So a little bit of history on Confluent. We started off as an on-prem company. Actually, even before that, our founders were the original authors of Apache Kafka and then we basically started off the company with an on-prem product. A few years back, we had our first truly cloud native product. And fast forward, in the last 12 to 15 months, we've been making rapid progress from moving to a single product streaming company to a truly multiproduct platform. And the proof is always in the pudding. And when you look at the different industries, be it financial services, technology, healthcare, insurance, we're there in all industries. In fact, 10 of the top 10 banks are using Confluent for their data streaming needs. And 9 of top 10 tech companies are actually using Confluent. And recently, we also spoke about OpenAI being a customer of ours. So the traction we've been getting with our customers actually is driven by the hybrid nature of our business. I've always said this that Confluent needs to be wherever our customers' data and infrastructure reside. If it's on-prem, we need to be on-prem. If it's in the cloud, we need to be in the cloud. If it is a hybrid architecture, which is a mix of on-prem, cloud, on the edge, multi-cloud, we need to be there. And how are we doing it? We're doing it with our product portfolio. First, I'll talk about Confluent Cloud, which is our fully managed cloud native product. As of the last reported quarter, which is Q1 of 2024, this business was 49% of our total revenue. From a revenue recognition perspective, we recognized revenue for Confluent Cloud based on consumption, that is if a customer consumes, we recognize revenue. Second, we have Confluent platform, which is our self-managed software offering. And as of the last reported quarter, 46% of our revenue came from Confluent platform. For Confluent platform, it is -- we sell licenses as well as maintenance, so roughly 20% of total contract value is recognized upfront as license revenue and the remaining substantial majority is recognized ratably over the period of the contract. And we also have a small component of our business, which is services. When you look at the trajectory of our top line, since fiscal '18 through fiscal '23, we've grown our subscription revenues at north of 60% compounded annual growth rate with the last fiscal year, which is fiscal year '23, our subscription revenue was $729 million. Our cloud business has grown north of 160% in compounded annual growth rate during the same period and in fiscal year '23, our cloud revenues were $349 million. From a margin perspective, since we've gone public, we've improved our margins by 34 percentage points. And in the last fiscal year, our margin was negative 7%. With Q4 from a quarter perspective was our first quarterly positive operating margin quarter. I just wanted to finish off with a quick overview of our Q1 results. We reported our results a month back, and subscription revenue we reported was $207 million and it grew 29% year-over-year. Our cloud revenue was $107 million, which grew 45% year-over-year. And our margins were negative 1.5%, which was up 22 percentage points year-over-year. And our dollar-based net retention was in the range of 120% to 125%. I'll just finish off with this slide. We're solving a real problem and this problem is going to get more and more complex with the years to come. So we're actually in the early innings of an incredibly large opportunity. How you solve the problem matters. We are doing that with differentiated and disruptive technology. And we're actually moving from a single product streaming company to a true platform. And finally, we have a team that has a track record of driving both growth and profitability at scale. With that, Jason, we can jump in.

Perfect. Awesome. Thanks, Rohan. I think for the audience here and those on the webcast might be helpful to talk about some example use cases of the technology. For example, customers, however you want to go. But to just sort of make it less abstract for folks. Can you give us some examples?

Yes. Absolutely. What's interesting is the blessing and curse of Confluent is our use cases and the number of use cases we have is absolutely all over the place. So it's innumerable. So I'll maybe pick a couple of industries and share some of our use cases. Let's start with, say, retail. Point-of-sale inventory is something that Confluent in Kafka supports, providing real-time marketing information to customers, when you're say purchasing something on a retail site and you get, okay, you're buying this product. This is another product that you might like. That's powered by real-time data. New product introductions. Analyzing data and new product introductions based on data, I think CPG. That's also a use case. Let's move to Financial Services. Fraud Detection is a very important use case. I think most of you might remember, say, 5 years -- a few years back, when you swipe your credit card and you had a fraudulent transaction, you would actually get a paper mail saying that these are the 5 transactions that are fraudulent and just you can call a particular number and dispute the charges. Fast forward today, you literally get a text in a second. That's real-time data architecture at work. High-frequency trading is another area. I can keep going, but every industry, we have a lot of use cases. And from what I said, you probably got one theme that is most of our use cases are mission-critical. They are truly revenue-generating. They are truly what impacts the customers of the companies that we support. So...

Yes. Great. I wanted to ask next about generative AI and how generative AI affects Confluent in the data streaming industry because the way I think about it is it just seems like you have an AI model, which is pretrained somewhere on Google or AWS, and then that model is sort of has the risk of getting stale. So is it right to think about you guys as being almost like a real-time data spigot that can feed model and sort of keep it fresh? How do you think about your role in generative AI right now?

Yes. That's a nice way to put it. I'll share a couple of perspectives. In general, when I look at generative AI, Confluent, we expect to be a very important player in this ecosystem. And we've said this before. This is definitely a future vector of growth for us as we look ahead. So why? Well, when you look at -- as you mentioned, when you look at today's architecture with respect to generative AI, you have the large language models. You have the vector databases. And you have these RAG pipelines that are created. But you absolutely need real-time context, real-time organizational context. And the only way you can have that is if you have a data infrastructure that supports it and we play at the heart of it. So you can think about Confluent as providing that real-time context into these generative AI models that support multiple use cases.

Great. Are you starting to see actual business yet from this use case, from this generative AI use case?

Yes. I mean when you look at from an overall monetization perspective, there are a couple of areas this can be monetized. One is where you can sell to the large language model providers. A couple of quarters back, we spoke about OpenAI being a customer. And these -- they can have large infrastructure presence. So that's one area. The second area is where I think there's a bigger opportunity is as you see real adoption, mainstream adoption of generative AI. And we are seeing customers use it currently. But I'd put it like the bigger opportunity lies ahead. I mean, we're still in the early innings with respect to adoption of generative AI. But we have customers like OpenAI, Notion and a bunch of other customers who are actually in the space, but we see the opportunity more ahead than behind.

I wanted to talk a little bit about Kafka. You alluded to Kafka as sort of the foundation of the company, really, right, in terms of taking open source project, which your founder, Jay, created at LinkedIn and then built a company by creating a commercial version of the open source Kafka project. Can you just talk about, number one, sort of adoption of Kafka and how popular it is. I don't know if you have any metrics on that? But then also I think the concern from some investors on open source companies and open source models is most customers have the option of using free, right? So how do you guys think about that in terms of products, road map, go-to-market, et cetera?

Yes. From an overall Kafka perspective, I -- beginning of my presentation, I spoke about the problem. And since Kafka was -- basically the Kafka was written by Jay, Jun and Neha. It literally had virality. Fast forward to today, we have over 150,000 organizations using Kafka in the open source. And it has become a de facto standard from an overall data architecture to have real-time data in different parts of the organization. So that's great. With respect to us differentiating and actually monetizing the open source, I'd say two big areas. The first is product differentiation. We -- like I mentioned, our overall evolution, we started off as an on-prem product. We had a truly cloud native version of the product a few years back. And in that short period of time, a little over 5 years, we've -- we basically, I would say, grown our cloud business to almost 49% of total revenue as of our last reported quarter. So a cloud native product is a true differentiator. So that's number one. The second differentiator from a product perspective is as we move to a true data streaming platform, which is integrated, that's going to be the second differentiator from the open source. The second big point is around the economics of it. There is this misnomer that open source is free. It's actually not. When you're running open source Kafka shop, there are different areas of spend that you have. Number one, you need to hire these really expensive Kafka engineers. And depending on the size of how big your Kafka shop is, the number of engineers could vary, but that's one area of spend. Second area of spend is infrastructure costs. Infrastructure costs are not free. And when you don't have a self-managed product, you as a company, you need to pay your infrastructure costs. And the third area of spend is around making sure you're getting everything to work from an overall operational standpoint, be it the bolt-on security products, be it upgrading to a newer version of open source. So when you combine all of these, there are real costs involved. With Confluent, we're actually combining all these three together as a managed service. So just to summarize what I said, differentiation from a product perspective with our platform and meaningful ROI and TCO yearly because we are providing this managed service.

Got you. All right. And then -- what about competition? I mean, what you guys offer? I know you compete against open source to some extent, and you have to make the ROI pitch to customers to go with a commercial version of Kafka and your overall platform. But who do you compete against today? Are you seeing the competitive landscape change at all?

Yes. That's a great question. When I think about the competitive landscape, essentially, think about 3 pillars. The first pillar is not necessarily competitive, but the biggest opportunity for us that's open source Kafka. And I just spoke about how we are differentiating ourselves from open source Kafka. So that's number one. Number two is, again, a little nuanced. I put it in the category of co-optation is with these public cloud providers, AWS, Azure and GCP. All three of these public cloud providers have their own version of Kafka product. But we have a lot of strong areas of partnership with them as well. So fundamentally, when you think about the public cloud providers, what's top of mind for them is the amount of data that's flowing into their respective ecosystems? And with Confluent, we're actually sending moving hundreds of terabytes of data into their ecosystems. In addition to that, in this category, we have strong go-to-market partnerships with them. That is their reps can retire quota selling Confluent. Their customers can use their credits if they are using Confluent. And on the technology partnership side, we're actually selling Confluent in their marketplaces. So it's a -- this is interesting because it's more partnership than competition, but that's category two. And category three, I put it in and say, catch all where you have some of the -- for lack of a better term, some of the legacy providers, the ETLs, the point-to-point solutions. And then you have some of the venture-funded startups. So we obviously keep a very close eye on category three, but clearly, one and two is where we spend most of our focus.

Yes, I think one of the common questions about Confluent too is, how many applications or use cases really have the need for the real-time data. I'm sure you hear that one all the time. I continue to hear it. And obviously, you guys continue to grow the business. So it seems like the use cases are expanding. But how do you guys respond to that question of -- well, how many apps really need this type of technology?

Yes. I mean when you really look at company by company, industry by industry, what's clearly evident is for some of your key mission-critical use cases, which are customer facing, you absolutely need a real-time view of the world. So that -- I mean, even before getting into the numbers, if you want to just look at it from that lens, you'll see we can go industry by industry, and you'll see absolutely critical use cases are driven and needs mission-critical architecture. So a real-time architecture. So that's number one. Number two, like from a percentage, I mean, there's no real data behind it, but there is a large number of applications that need to be real-time. And we spoke about GenAI earlier. With GenAI, if you don't have a real-time architecture, it will be very difficult to provide the outputs and what you need from the benefits and the efficiencies of GenAI. So in general, I feel that a large number of applications within organization needs to be real time, and there's a lot of anecdotal evidence that shows that.

All right. I want to shift gears to this go-to-market side and some of the changes you guys have made to the sales compensation model. Can you just talk about that change and just the impact so far?

Yes. We made some go-to-market changes, and I'll kind of just talk about a couple of them. The -- starting Jan 1 of 2024, we've changed our incentive structure from a go-to-market perspective, whereby we are incentivizing our field based on incremental consumption and not commit. So what's important to understand here is the why? So when you really think first principles, what is the value driver for consumption and consumption business models? It is essentially the unit of workload, the next new workload, the next new use case. And it actually doesn't matter if you have a large multiyear commit and the customer does not consume, unlike other business models, say, SaaS or a traditional license maintenance, Confluent does not see any value if a customer is not consuming. So this change was -- I'll put it in the category of getting alignment from the value driver perspective. Our reps get paid when our customer consumes and which is beneficial to Confluent as well as beneficial to the customer. So there is a lot of alignment from an overall value driver perspective. So that's number one. Number two, in 2023, what we were seeing was this inherent go-to-market fiction, with respect to our reps trying to sign the largest commit where that was increasing the deal cycles, et cetera. So this was addressing that point as well with respect to you know, we can land a customer and focus on unlocking the next use case versus trying to sign the biggest commit. And finally, on the go-to-market perspective, in addition to the comp changes, we're also focused on some key critical pricing-related updates from product introductions that we had, whereby, like I said, the opportunity we have with respect to the open source communities, huge. And we're increasing our surface area so that more and more prospective customers can join the Confluent ecosystem. And we did that by a couple of new additions. One was we called it the freight cluster, where for latency, insensitive workloads, we've added a new SKU. Then we've also added a new SKU, which we call the enterprise SKU where if you don't -- if you are okay with a multi-tenant architecture, you can actually join us through the enterprise SKU. So we're making a series of pricing changes in addition to the comp plan changes, which is aligning a lot of value. So these are the broad changes we've made. At the end of Q1, what we've said that 1 quarter in, we're very pleased with where we are. And early, I would say, metrics indicators were our customer count numbers in Q1. Our customer count numbers in Q1 were the highest we've seen in 5 quarters. But again, it's early days, but we feel good with the progress we are making.

Great. And kind of last question here, just -- I guess just from an execution standpoint, is it right to think that -- you guys, like a lot of other software companies, kind of went into, call it, 2022, 2023, thinking that things were going to be like they were in 2020 and 2021, and therefore, maybe there was some tweaking and some tightening that needed to be done from a go-to-market perspective. Because in 2020 and 2021, it was like shooting fish in a barrel. And the reps didn't maybe need to have the kind of sales playbook that they have today where the environment is different. Is that kind of a fair way to think about the kind of evolution over the last 4, 5 years of the business?

I mean, what I'll say is from an overall philosophy perspective with respect to how we run the business, it's always about making sure you have the right balance between growth and profitability. And I mean different times are different. Sometimes you invest -- forward-looking you make investments and then you kind of reap the benefits of those investments. So 2021, 2022, like you mentioned, or clearly, there were opportunities for growth, and we were taking advantage of it. And over the last 3 years since we've been a public company, we've shown meaningful leverage, but while maintaining high growth rates. So that's how I'd categorize it. And looking ahead, our recent allocation philosophy will continue to be the same. It's going to be balancing growth and profitability. And more importantly, making sure we are investing to drive those future vectors of growth which we spoke about in our last earnings call with our new product introductions that we had.

Okay. Great. We'll wrap it up there. Thank you, Rohan. Thanks, everybody, for joining. And those who want to go up to the breakout. It's in [ Adler ] upstairs. Thank you.

Thank you.