GitLab Inc. (GTLB) Earnings Call Transcript & Summary

All right. Let's get started. Rob Owens in my last session of the 2 days. So pleased to be sitting next to outgoing CFO, Brian Robins, of GitLab. So probably a different set of questions or discussion than when we had prepared for this. But thank you for joining us. Welcome back to Nashville. Brian did his undergrad and graduate days here in Nashville.

It's a good town.

It's a great town. We're not going to talk about the football team, are we?

No. The town has changed a lot. This whole second or honky-tonk area was never -- when I was here, it wasn't here.

So let's talk about how GitLab has grown up over the last 4 or 5 years since you joined and just what you kind of came into and where we're at now. In the middle, we've had a pandemic. We've had free money and then we didn't. And just kind of walk us through the journey of this company a couple of years ago and what you exit now as you take on another opportunity?

Yes. No, absolutely. And so thank you for having me. Delighted to be here, and thank you for all of you for joining and everybody has a busy schedule. So thanks for taking your time out today to hear a little bit about GitLab. When I came to GitLab 5 years ago, this month is my 5-year anniversary, we didn't have any audited financials. It took 20 days to quasi close the books. We weren't IPO-ready. And in just a very short period of time, 11 months after I joined, we took the company public on NASDAQ. And it's really one of the most successful IPOs at the time. It was from a revenue multiple perspective based on the company's strength did really well. It's interesting to look at sort of the company and how, over the years, we've changed and really all for the better. The reason why I came to GitLab is I really do believe that having a system of record and a platform to do the entire software development life cycle is super powerful. As a CFO and write numerous S-1s, I would spend most of my time being the air traffic controller to inversion control in sort of Microsoft Word or what have you. And over the last several years, actually used Google Docs because you have -- you can track all the changes, people can work on the document the same. And it just really, really takes down the amount of time it takes to draft an S-1 and so similar to Google Docs and having a system of record do software development life cycle, we can increase cycle times by like 7x. And it's -- we had Forrestar doing total impact study. And the payback period on the product was about 6 months, and ROI was over 480% in 3 years. And so it's been fascinating to see the amount of people and the companies that have come to GitLab to experience the payback and sort of the brand of the company and so forth. So today, we have over 50% of the Fortune 100, about 25% of the G2000. When I joined the company, we just had Premium and Ultimate. And that's really all you could buy. We're just a seat-based model. It's pretty simplistic. And then over the years, we developed an agile plan SKU, that's not a takeout of Jira, but it actually is good enough that some companies are using that instead of Jira. It comes with Ultimate, but for the nondevelopers, you actually buy the SKU. We had some customers that were in highly regulated industries, and they wanted to have a single-tenant SaaS, so they could have a private network that we would host and run it for them. So we launched a product called Dedicated. Dedicated has grown 92% year-over-year. I thought it would only be for very complex organizations. But now a lot of organizations said, hey, my total cost of ownership goes down dramatically if you deploy it, you run it and so forth. And then it's been really amazing with just AI in general. If you -- I think it's been 7 quarters ago -- 7 to 8 quarters ago, AI did not enter the conversation whatsoever in the sales cycle. And then Microsoft made a $10 billion investment in OpenAI, Copilot came around, and it was the first time in my life actually that something was so led by the media opposed to technology itself and then people actually using it. And there was just a lot of confusion around it in general. And we actually got a late start to AI. And -- but we played catch up, and there's a number of reports that were put out at that time that I found really interesting. It was they thought the LLMs will be commoditized. And that's, in essence, what's happened. And so we actually partnered with Google, we partnered with Oracle, we partnered with AWS, a number of different people to actually come up with very specific AI functionality for DevSecOps. And I really feel that was the right play. And so with Gen AI, now it's agentic AI. And so now we're working on creating all these agents that can actually go do human-like things. And so it's been super fascinating. It's been an amazing journey. Over the time, the entire team has been switched out. And what made the decision so tough for me to take on another opportunity was really the people. James Shen sitting here in the front row is -- will be named the Interim CFO in the 19th when I leave, but I have 1,000% confidence in him and the team. And it's been really great to watch going from a 20-day close to a 4-day close, going from forecast accuracy of 7% to 8% down to less than 1% and the teams really matured, understands the business. And the market is huge. It's not a winner-take-all market. Talking about AI, 7 quarters ago, AI came up in every conversation. We would get tons of questions around Copilot and Microsoft. And then that sort of died down. Then it was, hey, how about this thing called Devion with Cognition Labs. We saw this like really cool little demo and how is that going to impact you, then that died down. And then last quarter, it was around Cursor and Windsurf. And then Windsurf went and did that -- a little bit of an odd deal. And so we really haven't got much questions around that. So -- we -- on this past call and what we've been trying to do is really educate the market around what's an IDE, where do they leave, where do we pick off. In 18.3, we announced a lot of integrations into Cursor, Windsurf, Codex, Claude, and so forth. And so Bill came on about 3 quarters ago, put 3 priorities in place for the company. I think the company is heading exactly where it should go. And it's going to be fun to cheer them on from the sidelines.

It's always been very much a product-led growth company. .

Very much so.

It was a frictionless velocity play in the early days, and I think Sid helped drive a lot of that just reputationally, but now you're growing up on that front, too. And if we look at the last quarter, 29% growth, one of the best that I saw across my space, but then there was a little caution looking forward as you maybe putting some of those final touches or changing some of the go-to-market. So maybe you can touch on how that's evolved as well. .

Yes, absolutely. So we -- last 2.5 years, we had 4 different CROs. And so we had a gentleman that was at the company, Michael McBride, he took the company roughly from 0 to $500 million plus. Then we brought in someone from Microsoft, who was with us only for 11 months. And then we had an interim CRO and so when Bill came into the company, that was the first thing he started to address. And he went out, did a search, brought Ian in. Ian is very data-driven, very fundamentally driven. And so when we talked about guidance and we talked about go-to-market, Ian's been here for a quarter, there's a number of things that Ian wants to do. And they aren't really going to impact this year from a revenue and bookings perspective, but they're the right things to set up for next year and beyond. And so when Bill and I discussed it, we're like, let's make the changes today, right? We don't have to wait for this, but knowing that it was going to cause a little disruption. And so sales at GitLab historically has really been the one individual when you land an account, you own that account for the lifetime of the account, our gross retention rate is top quartile, best-in-class. When people come to GitLab, they hardly leave. Our churn has been relatively low. The churning contraction this past quarter, contraction is way bigger than churn. Churning contraction this past quarter is the best it's been in 3.5 years as a percent on available to renew. And we actually had a number of onetime disclosures that we made this quarter to sort of address some of the concerns that were out there. But one of the things I like the most that we put in the investor presentation was the net dollar retention rate by cohort back to the inception of the company. And cohorts 10, 11 years ago, are expanding at the same rate of cohorts 2 years ago. And so it's really hard for me to think of a product-led company that actually has expansion of customers 10-plus years. at the same rate as customers 2 years ago, we actually extrapolated out our 2016 cohort just as an example. And the 2016 cohort has grown 100x in ARR -- over 100x in ARR in just 10 years. Then we put the chart in there, so you could see the wedge for every cohort. And that's really the power of the model that I love is we put so much feature functionality into the platform every month. You want to -- we have these credit values and the I in credit stands for iteration and one of the things that Sid wanted to do was he wanted to ship as much software as he could every month and iterate, get feedback and then continue to develop that. And that's really without that GitLab couldn't have been created. And so we're shipping a lot. Our security module, if you will, is really top-notch, competes with SNC, Blackdot, Checkmarx, and you get all that when you buy the platform. And so that's really what's driving the strength of Ultimate. Ultimate announced 53% of our total ARR. It's been greater than 50% of bookings for the last several quarters. And by far, it's the highest priced product in the market. But I think that's really a testament -- those -- that data that I'm talking about is a testament to the benefit that the customers are getting by deploying Ultimate.

And if you look at that oldest cohort still expanding at that rate, that feels very counterintuitive relative to, gee, we're asymptotically going to go to a couple of coders sitting in a dark room that is going to suggest things and software is going to be created. So as those customers...

That would just be magical.

That would just be magical. And look, you've got the S-1 example and working with bankers is always challenging for any side of the business that you're on. That being said, your biggest customers have like 30,000 developers or so. There are some very, very large customers when you start to talk about 50% of the Fortune 100. So there are those layers and levels of complexity. So in that customer base and with those big customers, what are they still expanding to at this point?

It's -- when we land with a customer, typically, it's been a bottoms-up adoption. And so we'll land with a division department. And then we'll go from -- and the persona as a developer. And we have a -- between us and Microsoft, we -- the amount of license -- we have an estimated 50 million users on our free tier I think they say they have like roughly 130 million. And so if you combine the 2, it's like 3x -- excuse me, 3x amount of knowledge workers in the world today. But typically, it's a bottoms-up adoption because every developer has a GitLab account, they got a GitHub account. And whether they worked in a university, whether they're a hobbyist or coding. And so when people go into these companies, they're very familiar with GitLab. And so typically, it's bought on a credit card, it's 50 to 150 licenses. It's the developer who has it. And then the next division department will come up. And so you'll go around sort of that's why we have a person that sort of lives with the account. So as you'll go around, you'll continue to hit division department, division department with the persona of a developer. And then you'll get into the operations people. And so you'll still expand and sell more seats. And typically, they would land on Premium. Now more people are landing on Ultimate, but there would be a premium bottoms-up led and then when you got to the security department, theye're like, hey, this is great, but it doesn't have SaaS, DAST, fuzz testing, container scanning, vulnerability assessment, all the stuff that we have in security, but our Ultimate does. And so then we would typically see an upgrade to Ultimate and then we would expand more seats and continue to hit the Dev, the Sac and the Ops. When you look back at the net dollar retention rate for all the years, the biggest area that people are expanding is seats. And so more people are adopting the product within the company. UBS is a great example. UBS today wall-to-wall implementation, they went out and did a massive RFP. They were on Azure and some other platform that I won't say. And they selected GitLab because of the technical ability and what we could deliver. And they deployed 9,000 licenses in 9 months, 18,000 licenses in 15 months. I don't know how many they have today, but thousands and thousands of licenses deployed. And the beauty about GitLab, you save in 4 different ways. So the first area you save is you can actually get rid of point solutions. And so instead of having 15 or 18-point solutions in the entire software development life cycle, you can actually cancel or just end of life, a number of those. And so you get paid really quick -- paybacks really quickly for that. The second thing is, usually, there's a team within these organizations that are actually doing the UI and putting all the point solutions together. And when there's updates or whatever, they got to go make sure that they're, in essence, creating a platform within the company. And so when you get rid of the point solutions, you don't need as big of a team or if you go all in on GitLab you actually don't need the team. And so it's the second area of savings that we see from companies. The third area is really on developer productivity. If you go from a 2-month release cycle to a 2-hour release cycle, you have just way more time to do more work and you become way more efficient because you have the system of record where everybody can work on it at the same time. And so that's the third area. And the fourth area is, a lot of these applications that are getting developed are revenue facing. And so you can imagine if you can get them out in hours versus months, that as well. And so those are the areas that we see companies get the payback from the product.

And then the spirit of the UBS, which in our estimation is then going to be a high 7-figure, low 8-figure type of customer. They're a big customer. So in the spirit of UBS, let's talk about PQC, price times quantity, but then we got to add in consumption. And how you think about that pricing model of, say, $10 million customer that they're not waking up with now $15 million bill and the relationship goes sideways.

100%.

So we're at a massive transition in this industry. We're moving to even more of a consumption model soon, so good luck with that. But how do you think about these transitions relative to those large customers and how GitLab is going to -- what they're going to do to assuage customer concerns?

Yes, 100%. So I can -- let's unpack that a little bit. I'll talk about 2 different ways. One is, how do we view pricing at GitLab and so really, there's 3 different things we look at. We look at what it costs to deliver and so we've included some of our AI features like Duo Chat and Code Suggestions in Premium and Ultimate because the cost of delivery is relatively low. The second thing we look at is what value are we delivering. The more value you deliver, obviously, the more you can charge and so we try to assess how much value we're delivering. And then the third thing is how it's priced actually in the industry today. And AI is heavily based on tokens. Token cost has actually dropped dramatically over the last 2 or 3 years. I was at another conference earlier this week and heard token costs went from like $33 down to like $0.09, like over just a very short period of time. And so those are 3 things we look at in pricing. I think when we talk about the consumption model at GitLab, it's very important to note that we aren't changing the business model. We're actually adding another element to the business model. So the way I like to think about it is really in 3 layers. The core layer is Core GitLab and that will be a seat-based model. It won't change. UBS will continue to pay us per seat. All of our customers will continue to pay us per seat. The second layer will, let's call it, the AI core level. And so even though most everything is based on consumption, people don't like models where they get surprised. And so our sales team will go in and work with the customer to understand what the consumption level will be and they'll buy a package. And that package will include -- it will be query-based or token-based or something, and they'll be able to budget and plan that and then the third sort of level that I like to think about is AI variable. And so when you go over that package base, then there would be a variable cost associated with that, and that will be sort of the highly variable component. We're building some feature functionality and flags into it to where we'll see when they go over, we'll alert them. And the per token variable amount will be more than the package amount. So then the sales team will go in and try to upsell them to a bigger package, so they have more consistency on their forecasting. And so I think that different than, I think, the company that you're referencing is pure consumption. This will actually be more in sort of 3 different layers and I don't think it will be as surprising to someone like UBS because they'll have a real-time meter you can see, and so we'll know where exactly they're at on the consumption-based components.

Do you think there needs to be a wholesale shift to consumption at some point? Or do you still see the PQC type of model?

With GitLab, it's really interesting, right? So when I was interviewing with Sid to come to the company, I said we're giving so much value away on a seat basis, and you really want people to buy the blades in the platform. And so if we actually sold by the blade, I think we would -- we could actually get more value from the customers. and it would be more aligned with what we're delivering. But then when I came into the company and really my entire career, I've actually worked very closely with CROs on the go-to-market and so forth. I realized that the market is just so big. It's a $40 billion TAM, growing very quickly, now with AI and system developers and everything else, we believe seats are going to actually grow more. And when I looked at the -- when we went public, our gross retention rate was like 97%. And so when I looked at all this, I realized the #1 thing for GitLab to be successful was to go land more customers. And if you actually made the pricing model super complex, either by blades or all consumption or something like that, the sales process would get linked in, it would be more complex, it would be way more difficult. And so the seat-based model, and we got 90% non-GAAP gross margins. And so it's -- we're delivering it very effectively as well. I think is the right model to capture market share. And then I think with the add-on of AI that will be consumption-based, that hybrid approach, I believe, is the right approach.

In Q2, you grew your R&D high teens on a year-over-year basis. Why doesn't GitLab in and of itself, find more efficiency here. And we're going back to the number of developer type of argument. Obviously, you're getting better productivity out of those developers, but..

Yes. So I think we've done really well on sort of operating leverage in model. Since I've been in the company for 20 quarters, sales and marketing [indiscernible] improved every single quarter. We were able to get the company non-GAAP positive. We -- year-to-date for the first 2 quarters, about $150 million of free cash flow production, greater than a Rule of 40 company. And so I think we are -- one of the things that Sid and I said at the IPO, and it's really stuck, is the #1 thing we want to do at GitLab is growth, but we're going to do that responsible. And so since I've been at the company, there's been like really a number of different phases in the company. So got to the company we're private, hundreds of millions of cash on the balance sheet, company was growing in excess of 100%. But we acted responsibly, and we made the investments in the right area. When we went public, investors, Wall Street and Journal saying grow at any expense, right, grow, grow, grow, it's all based on revenue, but we still acted the same. Then we had a sharp correction in the market and COVID came and we still acted the same. And so I think the -- as we look at laying out the plan and we look at sort of creating shareholder value, the #1 area we look first is the growth, but we will do that efficiently. The 2 areas that we're investing most in today are sales and marketing and R&D. Sales and marketing capacity curve is super important. We meet on it actually weekly. We never want to get behind. We have enough capacity for this year. And going into next year, enterprise ramp time is roughly 9 to 10 months. So very, very critical. And in R&D, we're investing in AI, in security and SCM, CI/CD because we want to remain best-in-class in that because we land with the developer, then we can actually expand in the business with operations and security and so forth.

We've got time for 1 question if there is one. You know he always has a question. Go ahead.

Can you expand on that point of developers and not like [indiscernible] have you started to see that at all like on [indiscernible]..

Yes. So great question. There's a lot of questions about a seat-based model and will seats go way down or will go way up. We've been saying GitLab, we think the market will get bigger. There's been a lot of third-party articles around this that AI has actually created more code, but it's taken more time for security fixes, bugs, review time and so forth. We haven't seen like the citizen developer yet. I think the AI tools that people are using are really at the beginning stages but I do believe, like in the future, especially with agentic AI, the things that you'll be able to do like anybody could be a developer. And I think software is eating the world. Every company is a software company in some way. And so I think you're going to see the complexity of software go up. I think the way that people actually are operating within companies are going to change. And I think you'll see a much bigger market in the future due to [indiscernible] developers and the impact that AI has on that.

Excellent. Well, that's all we have time for, Brian. Thank you.

Thank you, Rob. Great to see you.