GitLab Inc. (GTLB) Earnings Call Transcript & Summary

[indiscernible] great conference. This is that I was delayed because there were so many people, could not get up on time. So welcome to the Goldman Sachs Communacopia and Technology Conference. I think it's your first time as CEO of GitLab.

First time as CEO of GitHub, yes.

GitLab. Brian, you've been here before, right?

Yes.

Yes, you have. This conference has been -- it was just day 2 of a 4-day conference and it's been -- would you say it's been an amazing success, 3,000-plus people. For our industry, that's a really huge attendance. We're up like mid-single digits percentage from last year. So thank you for making it special. We have not met in person before. So delighted to meet you in person. Thanks for making this trip. I know we talked about your background a little bit after an earnings call or so. And maybe we could start with your background. I know you worked at Microsoft. I believe you worked at Adobe. So you've been at storied software franchises. And as you reflect upon your prior background, what are the things that you experienced with Microsoft and Adobe and New Relic that are so applicable to the task at hand at GitLab. And then I want to ask about where you see the company going, but let's start with your background first.

Sure.

Is that a fair question, guys? That's good, right?

Yes. So I did a few smaller companies before Microsoft, I joined in '99, thinking that like at the time, it was probably one of the biggest software companies in the world. And I wanted to go see what it took to build a Microsoft. I thought I'd be there 3 or 4 years, ended up being there 16, 17 years. And the last 5 of which was actually getting Azure off the ground. The entire time I spent building developer tools and platforms at Microsoft. So I own basically the Microsoft web platform, everything from the web server to the developer tools with Visual Studio, .NET framework, all of those.

That was monumental. I mean that was -- I remembered as an analyst studying all that stuff.

Competing with open source. This is the Steve Balmer, open source is a....

Developers, developers.

Developers, it was all glory days. And a few things I learned from that, including getting Azure off the ground. Number one is it's one thing to build cool technology. It's a different thing to actually build a business. And particularly with the developer tools category, it's a challenging software category because it really requires you to appeal to 2 different audiences. You've got to win the hearts and minds of developers, and you've got to prove ROI and speak business value to the buyers and those who actually have the budgets to make these large purchases. And so it's difficult that way. It's also difficult because developer tools are kind of -- they're often very fad driven. They come and go, developers fall in love with one and then the next and then the next and the next and developers love to solve their own problems. So they create their own tools to solve those problems, and that leads to this fragmentation. And so it's one of the reasons I was so interested in GitLab is because for now more than a decade, GitLab had a really clear point of view that to solve developer problems and provide best-in-class productivity, you need a unified platform to do that. And it's not only building that to obviously provide the ROI for business decision-makers, but build that in a way that developers love and that they're passionate about and excited to use. That's a really challenging problem, and the company has done it with amazing success. I mean, last quarter, we reported nearing $1 billion in revenue, 29% year-over-year revenue growth, 17% non-GAAP operating margin. We released some onetime cohort data that shows cohorts getting all the way back to the inception of the company. 2016 has expanded 100x. All cohorts continue to expand about the same rate. It's an amazing business. And just seeing all of that complexity play out at Microsoft building some amazing software really, I think, helped me view the GitLab opportunity through that kind of multibillion-dollar generational business lens. That's why I want to be here. I want to help create GitLab, make GitLab go way past the $1 billion mark towards multibillion and enduring company. I mean the market needs GitLab more than ever. We're the only pure-play independent public company that's providing DevSecOps to the market.

Keep the fight on. That's great. Laudable journey.

So you said the faddish nature of the developers, I guess, designers are even more faddish. And this is a good way to get into the second question. How do you build an enduring business when you've been through a few fad cycles before? And what helps you break that curse of the fad and envision what GitLab is going to be like in the next 4 to 5 years. As you build that path, how do you overcome those breakpoints and make sure that it's generationally relevant to -- my son just started going to college. So when he comes out of college. So how do I convince him that GitLab is not that I will -- not that he will listen to me, but...

He should, Kash. You've obviously had a lot of success. So yes, I mean, I think -- first, I mentioned the platform bet that GitLab made. There's a lot of DevOps, DevSecOps tools out there that are best-of-breed tools. They specialize in one thing. And GitLab has really specialized in a breadth approach of building a unified platform that makes the software life cycle seamless. And the reason that's been so successful, I believe, is what is it now? 15 years, we've been talking about software is eating the world. And that means every business is now a digital business and depends on software in part to grow and operate. And GitLab is really uniquely positioned to help companies do that more efficiently, more productivity than anyone. Rather than assemble the best-of-breed set of tools that come and go, you bet -- you make a strategic bet that really helps -- the developers still love, but helps them manage the process of change from software -- of software from conception to production. Now here's where some of the adjustments that we're making right now that I'm sure investors are interested in come to play, which is for years, we've had this amazing open source community that's fueled the growth. Oftentimes, customers would start with our open source product. And then developers within enterprises would say, hey, there's this amazing open source software. We're using it for free. Let's go to GitLab and get commercial support for it and let's get some of the new commercial additions like when we added security and SAST scanning, et cetera. One of the adjustments we're making right now is we're building on that kind of open source community roots to commercial with our sales-led growth approach, and we're incorporating an additional product-led growth approach to help us come at the market with AI in mind. AI is now the second, an additive secular tailwind that holds massive opportunity for developers. I mean we see it with all of the news around Cursor and Claude Code and Codex and all those developer tools, which are all taking a product-led growth approach for a couple of reasons. Sales cycles are really short, product iteration loops are really fast, and you can build something that developers love. And the innovation velocity is amazing, right? So we need to be able to tap into that to stay relevant, to build amazing product experience and to connect those AI dev tools with the system of record that GitLab is. That's the real value we provide organizations. We're effectively change management for their software. The bottleneck for software has never been software code authoring. To accelerate that's important, and it's really cool that all those AI dev tools are accelerating code creation because it lowers the barrier to creating code. It means there's more volume of code, but that code cannot go to customers unless it's version controlled, unless it's tested, unless it's secured, unless all of the security and compliance rules that the business has to maintain in order to serve their customers, unless those happen. And that's what GitLab does. We're effectively change management for software. And so we see this AI wave as a secular tailwind that's going to drive even more GitLab. But we've got to be intimate with those AI developer communities. We've got to build our own AI developer experiences and build them, integrate them directly into the system of record of GitLab. That's our opportunity and what we're really uniquely positioned to do.

Got it. I recall the very first time I met Sid, the Founder, was the company is doing $50 million in revenue or so maybe even less than that and you had this board, his digital monitor that was flashing, had an ARR update like I've never seen anything like this. And he explained to me the vision, the 9 stages of SDLC, I hope Sid is doing good. My best wishes. But in that SDLC life cycle, where do you see the most white space opportunity to gain share?

Yes. So we're investing in 3 -- we talked about this year as a team, shifting from a breadth first to a depth first approach with a focus for right now on 3 key areas. The first is AI, no surprise. And that's both a vertical investment, meaning there's specific AI capabilities that we're investing in, but also a horizontal, meaning every task an engineer can do, we are building agents to do that task. Every existing core DevOps feature, every existing security feature, we want an agent to be able to work collaboratively with an engineer to do. Everywhere in the platform that you can engage another human to collaborate. So for example, assigning issues to another engineer to work on or mentioning an engineer in a comment to get their context or their help or, for example, when you set up GitLab to automate certain processes like builds and integrations. When those things fail, they notify humans today. For all of those cases, we want an agent to be an option. So they can say, "Hey, rather than spend my precious high-paid developer resources responding to build failures, let's put an agent in front of that to be the first line of response to triage the build failure and actually recommend a fix or if possible, actually automate the fix. Everywhere I want to get more information about a code issue. Instead of just asking another human, let me first ask an agent to see if the agent can gather the context to actually recommend a fix or next steps instead of using another human." This is how we really can create 10x the kind of productivity with engineers because what we're essentially doing is we're not just giving them more productivity tools to make them incrementally faster, we're actually automating their work. We're actually augmenting their capacity to do work, where one engineer can now spin up dozens of asynchronous agentic work streams to work with them in parallel to solve these problems. It's like going from single-threaded developer to multi-threaded developers and the ability to [ paralyze ] work.

I don't think I've heard you explain this way on an earnings conference call, unless you explained...

I'm not as engaging on earnings. What can I say? I'm nervous by all your random questions.

So yes, earnings, NER, DBNER, I mean, Brian questions, not random, but CFO questions, we'll get to that in a second. But I think this kind of forum is kind of the appropriate forum to dig a little bit deeper and an earnings call does not really afford you that opportunity. But I'm glad you explained how agents play a role in that SDLC because that is a different GitLab. That's not the GitLab circa when you guys were $50 million in revenue...

Software -- I don't think software -- change management for software is ever going to go out of style. Like these amazing AI dev tools that generate code, it actually just creates more change management, right? There's lower barrier to entry, so more people are creating code. There's more code getting generated. But unless we're willing to turn our world over to the AI overlords and trust that they're going to do the right thing, which lots of articles and studies have published recently have shared that's not a good strategy for your business, human oversight is required. And I don't think that's ever going to go out of style. So supercharging engineers with agents is an amazing opportunity for us. Super excited to be part of it.

We've had that -- it's just Day 2 of the conference, Day 2 is not even over, we've had so many CEOs of software companies and other companies that are not software just continuously point to how AI is increasing their developer productivity that has not really resulted in them laying off developers. They want to hire more. They want to experience the productivity, and they want to just squeeze more product through the same developer or more developers. I want to ask you a little bit about Duo. So you brought in Duo Access by adding Duo Chat and Duo code suggestions to Premium and Ultimate tiers. What is the next 3- to 4-year path forward towards measuring more meaningful AI revenue contribution from this move here?

Yes. It's really important to understand that move in context of what we're doing with Duo agent platform. You see our existing subscription business with seats is amazingly powerful. We already shared some of the stats around that. The cohorts continue to grow. We continue to pursue that opportunity just as we have the last decade. What's new is this additive, usage-based pricing model related to Duo that we want to go after with more of a product-led growth approach where we can accelerate customer acquisition, customer adoption and tighten the product iteration loops. The way we do that is we're seeding our entire customer base, every Premium, every Ultimate customer gets an entitlement of Duo agent platform. So those features you mentioned, Duo Chat and Duo code suggestions, they're effectively the bridge between the current software we provide for humans and the software that we're now bringing with agents so that every single user can get access to Duo agent platform. It's limited use, meaning that they can only use a certain amount for it. We're currently not enforcing those limits because Duo agent platform is in beta. There's not monetization available. But when it reaches GA, we'll now have this amazing product-led growth expansion approach where we don't have to have another sales cycle. We don't have to have a required contract signature. We don't have to require a new SKU to be adopted. They can just start using. And when they hit the entitled limit, they can accept digital terms and we can begin billing them for that usage. So that's a really important -- we intentionally unlocked that to start to get our customers in love with Duo agent platform. And once we reach GA, that will all become monetized usage.

Got it. I want to bring Brian into the conversation here. When you get to that usage, when do you anticipate this usage model being a tangible contributor to growth? And can you maybe dig into the aspects and the nuances of the hybrid seat plus usage pricing model that Duo has triggered for the company?

Yes. No, absolutely. One of the great things about the model today is it's highly predictable. And we've done a good job of predicting sort of what the seat-based model is going to be. It's important to note that we aren't changing the model. So we'll still have the seat-based component. And then on top of this, we'll lay the usage component. When we think about pricing, we really look at 3 different things. One is what value we're delivering to the customer. Second is what the cost of delivery is and third, how the markets price it. The market is priced in tokens today based on usage. I think it's important to note that we'll have a package that you'll buy and then you'll get utilization above that package. And so if you think about the core of the business, we have the core, which is very large and growing. Then we'll have packages that we'll buy that will be ratable that we'll be able to know what that is, then usage above that. And so I think as we go into -- as this becomes more material, we'll build models around that usage so we can actually forecast what the consumption will be.

I'll just add to that, which is exactly right. One of the ways we're structuring that pricing model is to incentivize those commitments of usage to help customers get the very best pricing because they want the forecast and ability to budget and we want that ratable revenue. And so even though this is a usage-based opportunity that's going to scale with usage. From a customer perspective, we'll incentivize the ability to make upfront monthly commitments kind of like Datadog to get the very best pricing.

Got it. I want to talk about Ultimate as a part of this discussion. It's now a large -- larger part of ARR relative to the IPO, Brian, I don't know. Ultimate was small as a percentage of revenue. Kudos, great job on driving that share to be significantly above 50%. And security as a catalyst for Ultimate as a share of ARR has played out really well. As you look into the next 3 to 4 years, what is the conversion playbook to move more customers up the tiers that could have beneficial impact on ARPC? Who wants to go?

So Ultimate, as you're right, Ultimate is our highest price tier at $99 per month per user. And we're really happy with how Ultimate is now 53% of the total ARR. And as you said, people are coming to Ultimate primarily because of advanced security features and compliance. There are several reasons why people come to Ultimate, and we believe collectively that everybody should be on Ultimate is Ultimate, the payback period is less than 6 months, which being the highest price of the product is super hard to believe from a value proposition perspective. And the ROI is over 480% in 3 years. And so we're seeing more and more people land on Ultimate. And then we're also getting the conversions from Premium to Ultimate due to the feature functionality that you get versus Premium.

Got it. And so Bill, do you have a game plan? I know you have a new CRO coming in. Are there any specific mandates for the new CRO to ramp up Ultimate as a percentage of ARR? And what are the tactics and the playbooks -- playbook is a thing that you guys talk about your go-to-market. I asked my wife the other day, what do you say playbook? Well, [ Tommy ], don't you know? That's playbook. That's how...

That's like coaching the team. [indiscernible] a place to run. Yes, speaking of playbooks, so coming back to your question, Ian is our CRO. He's been in seat now -- Q2 was his first quarter and a phenomenal CRO, amazing strategic mindset. We are really looking at the mid- to long-term opportunity with GitLab, which we believe is massive, while executing every quarter to maximize our results. And as we think about reaching that $2 billion-plus revenue scale and the need to -- the desire to continue to be a high-growth company, it's clear that we have to run a really clear playbook of how we drive that customer value journey. And you just go back a couple of years ago, the customer value journey was really simple, land on Premium, upgrade to Ultimate. Worked fantastic for us. Well, the world is a little more complicated these days. We've got Premium, we've got Ultimate, we've got Dedicated. We've got Duo. And now including Duo agent platform with this new usage-based component, we need a playbook to help every seller and the solution architects and people -- customer success people beside them to really understand how to guide the customer on that value journey, how to get the most value from the product, when to upgrade from Premium to Ultimate, how to add more Duo capability. And so that playbook has to be really well defined. Otherwise, you have soccer or football or whatever players running in all different directions, right? So that's part of what he's doing. And like Brian said, ultimately, we want everyone on Ultimate. We want everyone using Duo agent platform. And -- but we have to meet the customer where they are. Some are early in their DevOps journey. Some are ready to go out of the gate. And so that playbook has to be flexible as well, depending on where the customer is and the customer problems we're trying to solve.

Yes. And with that said as well, Ultimate sort of speaks for itself, right? And so one of the things that we included in the investor presentation this quarter was the net dollar retention rate by cohort since the inception of the company. And I can't think of too many other companies where you have cohorts from 10, 11 years ago that are still expanding at the same rate of cohorts 2 years ago. And that really is the power of the model. Sometimes they add more seats, sometimes they're doing a tier upgrade and so forth. And so to me, I think that's -- for us, we go into the -- we go into our customers with a very consultative sales approach and find what's best for them. As they adopt more stages, which they get when they buy it because we just charge per seat, they see the benefit that accrues to them. And then having that system of record across their entire software development life cycle really pushes them to Ultimate.

Yes, the blessing is that you give good detail, which we really appreciate. But the curse of it, if it is more seats, then people like, oh, yes, your ARPU went down, right? And if it's more ARPU, oh, the seats went down, AI is taking away jobs, right? So we're caught in this narrative that anything that is good can be spun as a negative narrative. And I'd like to say the numbers and narrative are important for a stock. And if the numbers are good, but the narrative is not good, it doesn't help the stock, right? But hopefully, we get through this. I know you published a study. It was based on 428 customers or so that said they plan to grow their developer headcount. Can you tell us a little bit more about how that study was done, when it was done? Was it done at a time when the Claude Code and the latest models from OpenAI, they're all kind of rapidly evolving and getting better and better. And so what is the math behind this? People ask me, what is the P x Q? I'm like developer science is art. It's not like P x Q. But what is your take on this whole developer jobs because you've got a unique vantage point here watching. That's who you sell to. That's your market.

Yes. One of my favorite quotes from the quarter was actually Matt Garman at AWS, who quoted something that Bill Gates used to say when I was at Microsoft all the time, which is like that's the dumbest idea I've ever heard. This notion that people are going to fire their developers and replace them with AI, like who do you think actually makes AI work for your company? It's these technical people called engineers that actually are able to harness AI and help your company take advantage of this. So I don't believe developers and engineers are going to be going away anytime soon. I think there's actually going to be more demand for them forever. In terms of the study you mentioned, Brian and I after the last call, where we were like barraged with like, oh, AI is going to kill your seat model and remove engineers and everything.

Not from me...

All those questions...

Not from me.

We said like we sat down, we're like, how can we bring data to this conversation to help investors understand the real dynamics with AI and the real opportunity. And so we did a number of things, including some of those onetime disclosures we already talked about. We also engaged a third party to do a survey of our customer base. That survey just wrapped up a couple of weeks before earnings. So it was like a month-long process before earnings. So it's very recent, about 400 customers. And we asked them some pretty simple questions. like, number one, do you see AI increasing, decreasing or not changing your use of GitLab. 91% said they expect it to increase their use of GitLab. We asked them in the next 12 months, do you anticipate your headcount growing, shrinking or staying the same in the context of this AI environment. And there, again, I think 88% expected to grow or stay the same, of which 78% of those said it was going to grow. So this notion that like engineers...

is it 78 out of that -- 78%?

88%.

Okay. So it's only 10% that are going to keep it flat.

Right. So it's like this notion, at least we're not seeing it. Our customers are not saying it. This notion, this hype is like the market is going to about to collapse or something is, I think.

Probably after the earnings call in the follow-up calls, you guys are like laughing behind what is it -- how does this even happen, right? I hear this a lot. Tech support, I understand, we call deflection is a real problem, and that's been a very underserved market, high turnover in that demographic. They don't really unfortunately get paid that much. So it can have an impact. But developers, I'm on the same page with you. Let's talk about the competition.

One other quick point. In that survey, we also asked what AI tools do you use because of this notion that like these tools compete with GitLab. Well, guess what? Claude Code, Cursor, Copilot and Duo, all in use pervasively across our customer base. So again, from a competitive standpoint, it's actually more of an and than or.

Yes. So Bill, you've been through developer tools wars and survived a few, won a few. As you take a step back, is this more of an adjunct thing. AI, there are so many AI foundation models. A couple of years ago, there was 2, now it's like 6, soon to be 7. And so we're going to have a proliferation of foundation models being able to generate code like crazy. Do you care who wins or you do because whoever wins, you -- how does it benefit GitLab? Does it matter to GitLab who wins the code wars?

Yes. No, I don't think it matters to us. One of the beautiful things about GitLab is, again, we're the only pure-play independent public DevSecOps company on the planet. We stand for independence, not only in terms of the cloud you choose to run in, you can run in your own cloud, you can run in any of the hyperscalers. You can take GitLab anywhere or you can run it in our SaaS environment. We stand for independence on LLM. You can choose any LLM provider. We provide all of the popular ones built in and continue to expand there. Not only that, we also support self-hosted environments. So you can run GitLab and Duo, including Duo agent platform when it's GA in a completely self-hosted, air-gapped environment where you have no connection to any public LLM provider. You can run your own models and Duo agent platform works great with that, too, which is an underserved market that we're really excited to go after, by the way. And from a dev tools perspective, we also now have shared we are providing native integration with all the popular AI dev tools. So Cursor joined us in a partnership announcement last month. Anthropic, OpenAI, Amazon, Google, all announced and did partnerships with us to integrate those tools natively into GitLab. But I just did a LinkedIn post, if you're at all interested in seeing the demo of what this looks like...

Follow him on LinkedIn.

For a developer. It's like literally 30 minutes before this session, I'm like I get this question all day long. I want to show developers and all of you what the experience looks like. It's seamless. It's seamless. Within GitLab, you can use Claude Code, Amazon Q, Codex, all within GitLab seamlessly and get that developer tool of choice. So again, we stand for independence, we stand for choice. And that's the beautiful thing. We don't care who wins. We're meeting customers where they are.

Exactly. Anybody has a question? This is fascinating, by the way. This is...

You asked the guidance question, so Brian...

Brian is not going to be around to deliver the guidance next quarter. So I mean...

Very, very capable to.

Yes. Chad, go ahead.

Everything you're laying out is pretty compelling. I think what you -- yes. Everything you guys are laying out, I think it's pretty compelling, and I think it's pretty clear that there's plenty of good stuff going around you guys. I think the confusion that people have is not only kind of where the Cursors and Windsurfs go from left to right, but then I think people have a hard time maybe understanding the competitive set a little bit because we hear about GitHub obviously, and it's probably your largest competitor. So maybe if you can take it however you want, maybe those 2 pieces, the left to right thing and then the competitive set because you are, as you said, kind of the clear and obvious independent player.

Yes. Speaking of our largest competitor there, I -- having been at Microsoft a long, long time, I can say I feel for the team getting absorbed into Microsoft. And I know the team that they're becoming part of. I wish them luck. I do think it changes the dynamic for the team and for customers. GitHub has remained fairly independent up until this point, and we'll see what the future holds there. We certainly see the community reacting to that and worried about being locked in increasingly into the Microsoft Cloud and developer tools ecosystem and AI. And that's why we are so outspoken about independence and choice because that's what we hear customers really want. In terms of the competitive moat or ability for other entrants to come into this business, I mean, think about who else has tried. Google and Amazon, as examples, have both tried to enter the DevOps market and both offered core value propositions that GitLab and GitHub offer, and both of them have exited. So could somebody else try? Sure. It's software. Anybody can build the software. But we have a proven business model, a proven platform with ROI. We've got incredible community around us. And we're taking all the steps needed to build a generational company to scale this thing to multibillion. So competition is inevitable, but I think we've got all the right things to create an enduring company. That's why I'm here. That's why I'm so excited to be here.

We wish you really well on your journey, and we wish you well, Brian. We'll miss you certainly on this stage, but maybe on a different stage. But congrats on all the achievements. Congrats to you, Bill, as well, and looking forward to the years ahead.

Thank you so much.

Thank you so much for your attention, guys.