Home / Transcripts / GitLab Inc. (GTLB) · December 2, 2025

GitLab Inc. (GTLB) Earnings Call Transcript

December 2, 2025

NASDAQ US Information Technology Software earnings 61 min

Earnings Call Speaker Segments

Yaoxian Chew executive
#1

Good afternoon, and thanks for being with us on the conference call covering GitLab's financial results for the third quarter of fiscal 2026. Joining me are Bill Staples, who serves as our CEO, and James Shen, our interim CFO. Over the course of this afternoon's call, we'll walk through an overview of the business, share commentary regarding our results for the third quarter, and offer guidance covering both the fourth quarter and fiscal year 2026. Prior to getting started, I'll address the safe harbor statement. I'd like to point you toward the cautionary statement concerning forward-looking statements, which appears on page two of our presentation as well as in the earnings release we issued earlier today; you can find both in the investor relations section of our website. Within the presentation and the earnings release, there is a discussion of certain risks, uncertainties, assumptions, and other factors that might cause actual results to differ from those expressed in any forward-looking statements under the meaning of the Private Securities Litigation Reform Act. As is customary, this language will govern the content of today's call and presentation. Additionally, on today's call we will discuss certain non-GAAP financial measures. Such non-GAAP financial measures exclude certain unusual or non-recurring items that management believes impact the comparability of the periods referenced. For more information about these non-GAAP financial measures, along with the reconciliations to the most directly comparable GAAP measure, please refer to our earnings release and presentation materials. I'll now hand the call over to Bill. Bill?

Bill Staples executive
#2

Thanks, Yao, and hello everyone on the line this afternoon. Appreciate you all being with us today. I'm happy to share that our third quarter performance was strong. Year over year, revenue rose 25%, landing at $244 million, which is two points higher than the guidance we gave for Q3. On a non-GAAP basis, operating margin came in at 18%, five full points above what we guided for Q3. This marks one year since I became GitLab's CEO, and each morning I feel tremendously fortunate to carry forward what Sid and the team built. Upon arriving, I laid out three points. There's never been a better moment to serve developers. We are still in the early days of AI's transformation of software, and GitLab is positioned at the core of the software development lifecycle, with the strongest and most complete platform to power that transformation. My confidence in this company and what lies ahead has only deepened. After a year on this path and hundreds of conversations with customers, I can say with certainty that we're in a stronger place today than we were just twelve months ago. We've laid the groundwork to unlock more value via AI over the next year, designing GitLab and the Duo Agent platform to stay mission-critical and to keep delivering growing value as LLMs and markets shift. I genuinely feel there's never been a more thrilling moment to be part of GitLab. AI's growth is widening our total addressable market. Barriers to creating software have fallen dramatically thanks to AI, pushing the marginal cost of generating code toward zero. Still, software is far more than code alone. Software, carrying all its embedded business processes and sensitive data, is essential to business. The world's economy depends on software. People's lives depend on software. Companies cannot risk carelessness in quality assurance, security, compliance, or governance across their software development and delivery practice. My belief is that our work becomes even more essential in a world where teams want to leverage agents to write code, given how non-deterministic AI is; for many decades, I've observed my own teams and numerous customer teams wrestling to keep up with bug backlogs, technical debt, and business requirements, all while trying to innovate. This pattern shows up everywhere and has persisted a long time. The tooling for creating software is technically quite solid, yet breakdowns happen consistently wherever people and processes come into play. GitLab has addressed this by giving teams an opinionated view backed by proven ROI. What sets us apart? We automate the full software delivery flow, covering quality, security, compliance, and governance within one process flow as part of our unified platform. Interestingly, in studying teams that use agents and new AI tools, we notice the same pattern. Agents behave strangely like people. At times they honor prompts, at times they don't. At times they produce secure code, at times they don't. What explains this? LLMs will forever be non-deterministic. That's inherent to the algorithms behind them. And each business has its own unique requirements that LLMs simply cannot guess. Suppose large language models eventually outpace people at writing code. Even then, someone—or something—will still have to check that code from the outside, and that checking will likewise be handled by agents while people keep watch over them, because businesses have complicated human needs that must be met. In our view, LLMs will keep getting more accurate and cheaper, yet systems that confirm they serve intricate business needs will remain necessary forever. Consider what that looks like right now. Tools built into IDEs—Cursor, Copilot, and Claude Code among them—have fueled a surge in code being written. We can already see the knock-on consequences in our own operations. Engagement with GitLab has risen substantially throughout our GitLab.com SaaS customer base; during the first 10 months of 2025, important activity metrics—CI pipelines, deployments, and releases—climbed roughly 35%-45% versus a year earlier, much like what competitors report. Among clients who pay us over $5K in ARR, usage proxies such as deployments and CI pipelines calculated per seat have increased 20%-40% year over year. To put it plainly, when more code gets written, GitLab becomes more necessary. According to our 2025 Global DevSecOps report, although AI speeds up coding, generating additional code does not automatically produce superior results. The AI paradox is the name we give this; in the long run, we think the winners won't be the vendors generating code most quickly, but rather those best able to maximize a customer's capacity to ship secure, high-quality software to the consumers of their business and to produce significant business outcomes via software. No one is better positioned for that than GitLab. In what way? Our end-to-end platform, which already drives full lifecycle actions for over 50% of the Fortune 100 plus hundreds of thousands of organizations spanning 100 different countries globally, has been extended, and we now deliver that identical set of capabilities natively to agents, together with shared context for humans and agents alike. This fosters stronger trust and accuracy while also helping speed up the end-to-end software delivery process that winning demands. Rather than merely creating fresh AI assistance tools and agents aimed at particular use cases, as our competition does, we have extended our platform to deliver intelligent orchestration throughout the software lifecycle, letting our partners, customers, and ourselves tackle any engineering problem anywhere in that lifecycle. For the AI paradox that most companies now face as they bring in new AI tools, the GitLab Duo Agent platform is our solution, and it will drive new revenue streams that go beyond seats. The context at our disposal is abundant. It encompasses semantic understanding of code and dependencies spanning repositories, fine-grained changes made to it over time, quality assurance tests, planning plus issue tracking and collaboration around those plans, security and compliance checks, and build, integration, and deployment pipelines—just to list a handful. As the sheer quantity of code balloons, our underlying platform grows more valuable, no matter whether a person or an agent made the change. I think the fundamental building blocks of collaborating on code are going to turn out to be formidable moats, and thanks to the Duo Agent platform, we're sitting in a strong spot even as LLMs get better and the market shifts. I'm happy to report that Duo Agent platform remains on schedule to reach general availability within the next few weeks. Now for the quarter: what stood out was ongoing momentum in GitLab Ultimate, which accounts for 54% of total ARR today and appeared in seven of our top 10 net ARR deals during this quarter. For organizations that require one unified DevSecOps platform, Ultimate is among the strongest value propositions available. Expansions at customers including Indeed, SBI Securities, and Currys were driven by Ultimate. We carried out the initiatives we talked about last quarter effectively, and those helped reinforce the company's foundation. The consistent headway we're achieving on building out our first-order motion and rolling out GitLab Duo Agent platform makes me pleased. Results outside the U.S. were also stronger. On the whole, sales cycles and win rates held steady; even so, weakness in the U.S. public sector canceled out a portion of that. Leading U.S. agencies still see GitLab as the software factory of choice and a partner they trust, yet decisions moving more slowly because of the government shutdown that followed introduced fresh headwinds during the quarter.

Bill Staples executive
#3

Third-party validation of our differentiated platform strategy is coming through strongly. GitLab earned Leader placement in the 2025 Gartner Magic Quadrant for DevOps Platforms — that's 3 years running — and also was named a Leader in the 2025 Gartner Magic Quadrant for AI Code Assistants, marking a 2nd straight year. Let me now move to our main growth goals. Objective number one: bring in additional paying customers, with particular emphasis on the mid-market and enterprise segments. We're beginning to observe instances where AI mandates push enterprises toward seeking out a solution that is future-proof. When it comes to a cohesive workflow on one unified platform spanning developer, security, and operations teams, GitLab's approach has never mattered more. Médiamétrie, the top media measurement company in France, required faster internal development processes and faced an AI mandate to bring maintenance costs down. They had already been on a free tier of GitLab for source code management, yet this first-order Ultimate plus GitLab Duo win meant replacing 7 other distinct tools right from the start while keeping ISO 27001 compliance intact. Secure AI-powered development will be made possible by Duo, letting Médiamétrie ship new measurement products and data analysis far more rapidly; we also secured a landmark deal this quarter with a global consumer tech platform, which came about because of our dedicated offering. This customer had become frustrated following critical reliability problems with the source code management provider they had been using; they have more than 5,000 developers, are renowned for world-class engineering, and hold exacting standards when it comes to reliability, responsiveness, and technical excellence. GitLab Dedicated was picked to supply the environment for their most mission-critical code repositories, safeguarding business continuity and operational resilience. Although our first-order versus expansion mix saw a slight improvement this quarter, it remains below where I want it to be; we're growing our go-to-market capacity and have brought on a new business leader to build out our global first-order team, centered on acquiring new logos. Worth noting: resourcing and ramping this team will take time, yet I believe the payoff will justify the investment. Each new customer we win today counts, because of their lifetime value. Urgency drives how we operate. Daily, we work to earn customer trust, which shows up in our best-in-class gross retention rates and demonstrated cohort growth across multiple cycles. A long runway for growth exists in our core DevSecOps opportunity, since our TAM keeps expanding and our competitive position stays strong. Choice, neutrality, and openness are things we offer in ways others do not, and that message is landing. Our competitors are actively deciding to restrict choices for their customers, whether through hyperscaler infrastructure or self-managed versus SaaS options. By direct contrast, we acknowledge that each customer journey differs and make active efforts to meet customers where they are. Objective number two: assist customers in realizing our platform's value more quickly, which in turn drives revenue expansion. A compelling pattern links our biggest expansions this quarter. Every one of them is putting money into some kind of AI tooling inside their engineering organization, yet GitLab remains the backbone of their SDLC, so we keep seeing significant room for up-tiering and attach across the customers we already have, and the stories below show what that looks like. Since 2017, a top financial SaaS provider serving small businesses has been a satisfied GitLab customer. During this quarter, they moved from premium to ultimate for close to 1,000 engineers. Because GitLab scans code right at the commit point, before anything ever leaves a developer's hands, we can take the place of several disjointed security tools. For this customer, GitLab will drive down false positives, trim manual overhead, and give developers app sec results right away. The expansion of a big European public sector organization shows what is possible among our current customers. Much like many large enterprises, they were dealing with siloed deployments spread across various groups and 4,000 developers. That created maintenance complexity along with inconsistent developer experiences. Having collaborated closely with them over the years with more than 120 stakeholders involved, they selected GitLab Dedicated and Duo during this quarter as the foundation for a modern and secure SaaS-based software delivery platform. In regulated industries and critical national infrastructure, GitLab satisfies the highest customer standards. Accelerating customer-focused innovation is our third objective. Our investment continues across three pillars: core DevOps, security, and AI. Within core DevOps, we shipped a redesigned interface along with a new intelligent pipeline repair flow, which lets developers fix issues more quickly, translating directly into faster development velocity and less time spent troubleshooting. As companies roll out AI toolkits, security and compliance rank among mission-critical customer priorities and stay key drivers of ultimate adoption; we rolled out new security capabilities, among them static reachability analysis, secret validity checks, and diff-based scanning, which bring security straight into the development process. The new Security Analyst Agent, also rolled out this quarter, makes these sophisticated tools easier to access and can offer engineers recommendations on where to focus, and through GitLab Duo Agent platform, we introduced the AI Catalog, a central spot where teams can find foundational GitLab Duo agents, best-in-class external agents such as Claude, OpenAI Codex, Google Gemini CLI, and also build, share, and collaborate on their own custom-built agents for any software engineering task. The response from customers has been strong. Many now say GitLab leads our peers in vision and in how quickly our capability evolves. This quarter, even before general availability, we closed our first few expansions built on GitLab Duo Agent platform. Our progress is fast and early results look promising, but on this massive opportunity we are just beginning. As the platform matures and we figure out the most effective ways to deliver value to our customers, pricing and packaging will probably be an iterative process. For any new investors, a reminder: a large portion of our customer base is still on self-managed solutions and might adopt some of these solutions more slowly. In February, we'll live stream a product-specific event where more details will be shared. We stand apart as the sole independent, publicly traded business that is both purely focused on cloud and neutral across models, providing DevSecOps. That gives you genuine independence. You can build in whichever cloud you prefer, using the vendors and tools you favor most, all while delivering the strongest imaginable experience to your engineers. GitLab is needed by the world now more than at any prior point. I'd like to express gratitude to our team members for embodying our values and our mission, to our customers for believing in us, to our partners for standing with us, and to the wider GitLab community. Prior to handing the call off, I also wish to thank James Shen for what he gave us throughout this transitional stretch at GitLab. Among the rock stars at this company is him, and he has performed incredibly well in stepping up as interim CFO. It's exciting to welcome Jessica Ross, our incoming Chief Financial Officer, who begins in January. Most recently, Jessica served as CFO of Frontdoor, and she brings over 25 years of experience across finance, accounting, and operational leadership, having worked at organizations such as Salesforce and Stitch Fix. Over the months ahead, all of you will get the chance to become familiar with her. So with that, I'll pass things to James.

James Shen executive
#4

Thanks, Bill, and I appreciate all of you taking the time to be with us. I'm glad to share that we exceeded our guidance on every metric, with the team delivering through a shifting landscape. Revenue in the third quarter came in at $244 million, up 25% compared with Q3 a year earlier; at present we have 10,475 customers whose ARR is at least $5,000, and those customers drove more than 95% of total ARR during Q3. Within our bigger cohort — those at $100,000 plus in ARR — we saw 23% growth year over year, reaching 1,405. Our customers span a wide range of industries and geographies, and no one customer makes up more than 2% of ARR. Regarding expansion, we closed the quarter with a dollar-based net retention rate, or DBNRR, of 119%. Total RPO rose 27% year over year to $1 billion, and cRPO was up 28% year over year to $659 million; we continue to feel good about such a robust growth rate. Non-GAAP gross margin came to 89% in the quarter. The team keeps doing well at generating operating efficiencies, even while SaaS makes up a larger share of our mix, helped in part by ongoing momentum in GitLab Dedicated and Duo. SaaS is now roughly 31% of total revenue and expanded 36% year over year. Q3 non-GAAP operating income totaled $43.7 million, versus $25.9 million in Q3 a year ago. Non-GAAP operating margin stood at 17.9%, versus 13.2% in Q3 a year ago, which is an increase of roughly 470 basis points year over year. We're making consistent headway in standing up a dedicated first-order team and expanding our quota-carrying capacity. Q3 FY26 adjusted free cash flow was $27.2 million, with an adjusted free cash flow margin of 11.1%, versus $9.7 million a year earlier. We finished the quarter holding $1.2 billion in cash and investments. A strong balance sheet and a predictable business model give us room to keep investing in our AI capabilities, platform enhancements, and go-to-market organization, while still delivering solid margins and cash flow for shareholders. On a separate note, I want to give an update on JiHu, our joint venture in China. During Q3 FY 2026, non-GAAP expenses tied to JiHu were $3.3 million, versus $3.5 million in Q3 a year ago. Deconsolidating JiHu continues to be our aim. That said, we can't forecast the probability or the timing of when it might happen. So for FY 2026 modeling purposes, we're projecting about $16 million in expenses connected to JiHu, versus $13 million a year ago. Now let me move to guidance. Although our strong performance so far this year is encouraging, the SMB weakness we flagged last quarter is still with us. On top of that, the ongoing fallout from the recent U.S. government shutdown will probably affect deal dynamics in our U.S. federal business as we move into Q4. Our guidance takes these dynamics into account. For Q4 of FY26, we anticipate total revenue between $251 million and $252 million, which works out to roughly 19% growth year over year; we anticipate non-GAAP operating income between $38 million and $39 million. Additionally, assuming 172 million weighted average diluted shares outstanding, we anticipate non-GAAP net income per share between $0.22 and $0.23. For the entire fiscal year 2026, we're looking at total revenue in the range of $946 million to $947 million, which would be roughly 25% growth compared to the prior year. Non-GAAP operating income should land between $147 million and $148 million. And with 171 million weighted average diluted shares outstanding, we expect non-GAAP net income per share of $0.95 to $0.96. Although we're not giving guidance for fiscal year 2027, I want to point out for your modeling that the premium price increase from April of fiscal 2024 has mostly been put into effect now and won't serve as a separate boost in fiscal 2027. To wrap up, our third-quarter performance has me feeling good. We're constructing GitLab for sound expansion at scale, making strategic investments in areas that generate long-term worth, while also improving profitability and producing free cash flow. GitLab stands ready for sustained success and to capitalize on a swiftly evolving market from a position of strength. I appreciate you being here today; now I'll hand things over to Yao, who will lead the Q&A.

Yaoxian Chew executive
#5

Right now, should you wish to pose a question, kindly make use of the raise hand option found in your Zoom toolbar, or should you be connected via phone, please hit star nine. In the interest of time, we ask that everyone stick to a single question; the first one comes from Koji Ikeda of Bank of America. After that, our next question will come from Matt Hedberg of RBC. Koji, please proceed.

Koji Ikeda analyst
#6

Yeah. Hi, everyone. I really appreciate you taking my question. What I'd like to ask about is the guidance, specifically the fourth quarter outlook, and in particular subscription revenue growth. In Q3, subscription revenue grew 27%. For a model that's mainly seat-based, that's quite impressive, but it does represent a slowdown from 30% in the prior quarter; and I understand what you're saying about softness in the public sector. That makes sense to me. So my question is about the implied fourth quarter total revenue guide of 19%. Could you help walk us through a bit more on the demand environment, any federal sector catch-up that has already occurred, and pipeline coverage going into Q4, plus any extra color on how we should think about what the guide implies for fourth quarter subscription revenue growth? Thank you.

James Shen executive
#7

Thanks, Koji. Our approach to guidance this quarter was pretty much the same as the one we used last quarter. We built independent roll-ups across the field, across CRO leadership teams, and across the finance team. And the prudence we highlighted last quarter for both SMB weakness and go-to-market disruption was well-warranted, and some of that carries into Q4. On top of that, as you mentioned, we'll see some lingering effects from the recent U.S. government shutdown. Guidance, ultimately, reflects our best view of the business today with what we know, and we feel good about the guidance heading into Q4.

Yaoxian Chew executive
#8

With our next question, we'll hear from Matt Hedberg, and after that, Rob Owens of Piper Sandler. Please proceed, Matt.

Matt Hedberg analyst
#9

Matt, thanks for that. Yao, appreciate the question. Bill, during your prepared comments, you mentioned that momentum in the first-order business improved relative to the prior quarter, yet I believe you also indicated it still falls short of where it ultimately needs to land. Given that this is likely a trend spanning multiple quarters, could you shed a bit more light on, from where you sit, what remains for the team to accomplish, and perhaps what timeframe we should have in mind before some of the complete advantages from that become visible?

Bill Staples executive
#10

Yeah, thanks, Matt. As we discussed last quarter, we made the decision to bring on a global leader whose sole focus is acquiring new business, reporting to the CRO. I'm glad to report that we completed that search and brought in that person, an outstanding executive who has now come aboard and is going through onboarding, and we're starting the hiring ramp for that team, which once again will be a global team, reporting to him directly into the CRO. That hiring ramp should take a couple of quarters, we expect, with results arriving in the back half of FY 2027. On top of that, let me share about the product-led growth side. Manav, our Chief Product and Marketing Officer, has now held the seat for a quarter and has started digging in there, examining really two things when it comes to product-led growth. First, tightening the feedback loop with customers who are earlier in their journey with GitLab, and also removing friction in the customer journey to make it easier to go from a free into a paid product with GitLab. The early results there are really promising. It's exciting to see the efficiency in the funnel improving and very early results; but here again, I would expect, you know, these kinds of incremental gains, to aggregate over time, and I would expect to see that show up in terms of new customer acquisition acceleration in the later half of FY 2027.

Matt Hedberg analyst
#11

Appreciate it, Bill.

Yaoxian Chew executive
#12

All right. Let's take the next one — Rob Owens at Piper Sandler, and after him, Sanjit Singh at Morgan Stanley. Rob, you have the floor.

Rob Owens analyst
#13

Thanks. Appreciate you fellows fielding my question. I was hoping you might go a bit deeper on that Fed effect you mentioned, since it likely has a bearing on the license line. If you can put any numbers around it, that would be helpful; and then, did that also play into your retention rate, or do other factors explain that metric slipping a couple of points quarter over quarter? Thank you.

James Shen executive
#14

Rob, the key point to remember here, I believe, is that our long-term public sector thesis is still very much intact. As you know, we serve as the preferred software factory for many of this country's leading federal agencies. PubSec accounts for roughly 12% of our ARR, and we have not put a specific figure on the headwind we experienced in Q3. What I can tell you is that we did see disruption from the shutdown as well as the ongoing effects of DOGE working their way through the government, and we are strongly partnered with our customers and these agencies to help them get past these challenges.

Yaoxian Chew executive
#15

Okay, we’ll take Sanjit’s question now, and after that we’ll hear from Shrenik Kothari at Baird. Sanjit, please proceed.

Sanjit Singh analyst
#16

Yeah, can you hear me?

James Shen executive
#17

We can.

Sanjit Singh analyst
#18

Thanks. Bill, regardless of the ongoing discussion about models priced per seat, I believe we can all agree that an enormous amount of software is being built and launched, especially at this moment. You referenced certain metrics concerning activity and usage on the platform, and those numbers sit well above the revenue growth you're currently delivering, at least for now. So, Bill, this is somewhat a longer-term question, but what ultimately is the answer—how do we bring platform activity into alignment with the revenue growth you'd like to achieve? Is the solution essentially Duo agents, or should we be considering something further out over the medium term?

Bill Staples executive
#19

Thanks for that, Sanjit. The way I see it, over the medium to long haul, the solution really comes down to how we move away from being purely a seat-based subscription company and toward a model that blends seats with usage-based pricing as we roll out the Duo Agent platform. During my prepared remarks, I noted that we're right at the edge of this — within the next few weeks we'll announce general availability and roll out pricing. That's going to let us capture revenue from what happens after AI generates code, driving AI acceleration throughout the software lifecycle and resolving the AI paradox we brought up earlier in our prepared remarks. Beyond that, we're pursuing incremental innovation layered on top of our premium and ultimate SKUs, which deliver added value to customers at an incremental cost — and that fits into our FY 2027 roadmap too, opening up fresh monetization possibilities.

Yaoxian Chew executive
#20

Alright. Up next, we have Shrenik Kothari from Baird, then Raimo Lenschow from Barclays. Shrenik, please go ahead. Hi everyone, this is Zachon speaking on behalf of Shrenik. Appreciate you taking the question; I have one about Duo. You've regularly stressed how architecturally significant Duo is and how critical it is to the mission, but how are you actually monitoring the monetization of Duo-specific capabilities, now compared with just core DevSecOps or CI/CD functionality? And then perhaps what percentage of new ACV involves Duo or features related to Duo? Thank you.

Bill Staples executive
#21

So yes, at present, within our Duo Pro and Duo Enterprise offerings, monetization happens via add-ons priced per seat, and while we haven't disclosed how much revenue those products generate specifically, they remain at an early phase. Earlier this year, what we did was make a shift — pivoting away from an innovation roadmap driven by use cases toward one driven by the platform, which means we've layered AI capabilities into our core platform at each level, enabling an agentic way of working with AI that assists customers in addressing countless challenges throughout the software lifecycle. Customers gain the ability to pick best-in-class AI tools — for instance, offerings from Amazon, Google, OpenAI, and Anthropic — and also to build their own agents with Duo technology, in order to tackle, once more, any category of engineering problem spanning the software lifecycle. That has now spent a couple of quarters in beta and is approaching general availability, and upon reaching general availability, usage-based pricing will be introduced. So both the innovation itself and the launch of that monetization stream are at quite early stages, yet the tone and dialogue with customers — as I've talked with them over the last year regarding AI — keeps getting stronger, with growing enthusiasm for the platform strategy we're pursuing now; thus I'm genuinely thrilled about what lies ahead. I do view this as widening our TAM and delivering remarkable new value to customers.

Yaoxian Chew executive
#22

Okay, now we'll take a question from Raimo Lenschow at Barclays, and after that, Howard Ma from Guggenheim. Raimo, please proceed.

Raimo Lenschow analyst
#23

Hello, I appreciate you fitting me into the schedule. Bill, I have a question for you — regarding, well, the softness among small and midsize businesses, that clearly affects everybody, and it's simply, as you know, not something within your control. Still, is there any action you might take — say, on the SKU side, or similar — that would let you compete more effectively in that segment? Is anything like that feasible, or is our only option to sit tight until sentiment in the broader market there gets better? Thanks.

Bill Staples executive
#24

Sure, SMB accounts for a really tiny portion of total revenue for us, and it isn't an area we focus our business strategy or go-to-market efforts around. That said, startups and smaller organizations that are expanding do matter to us, especially when it comes to building awareness and getting early adoption going. Our main play there, I'd say, has been offering a genuinely strong free offering — our open-source packages plus the free tier on GitLab.com — and we've watched solid uptake across an extremely wide community. I genuinely believe Duo Agent Platform opens up fresh chances to turn those free users into their first paid relationship with GitLab, since we're putting AI on top of those no-cost offerings over the next 12 months. Clearly, that's a route we haven't taken so far with the Duo Pro and Duo Enterprise add-ons, but once Duo Agent Platform hits GA, I think we'll give it a hard look. And I believe it appeals to customers regardless of size who'd like to begin their GitLab experience on a free DevSecOps platform yet are eager and ready to spend on AI, because they grasp both the enormous worth and the expense involved in providing it.

Yaoxian Chew executive
#25

Thanks. Our next caller is Howard Ma with Guggenheim, and after that we'll hear from Mike Cikos at Needham. Howard, the floor is yours.

Howard Ma analyst
#26

Thanks, that's great. During the prior quarter, you provided a figure showing that seats are expanding at a double-digit rate compared with a year earlier, and that this had been speeding up. What I'd like to know is whether that pattern remains true, and how does the growth in seats appear if Duo seats are taken out?

James Shen executive
#27

Howard, we provided that disclosure only once, in the previous quarter, so you could reason through and grasp how seats behave within the business. We're pleased with the seed growth achieved in this quarter, but regarding the particular details shared in the prior quarter, we won't offer any comment.

Yaoxian Chew executive
#28

Thanks. Our next query comes from Mike Cikos at Needham, and after him we'll hear from Kingsley Crane of Canaccord. Please proceed, Mike.

Mike Cikos analyst
#29

Thanks a lot, everyone. I'd like to circle back briefly to the public sector piece, and I appreciate you sharing the details and the challenges caused by that shutdown. Could you assist us in breaking down, when comparing the guidance for Q4 against the results for Q3 that you just reported, do you anticipate that the public sector drag stemming from the shutdown and DOGE will in fact build up or grow in size as we consider this quarter ending in January? And once more, I realize we're getting into, I suppose, very granular territory here, but I simply wanted to understand your thinking regarding the assumptions you're making as we examine this outlook. Many thanks.

James Shen executive
#30

Yes, thank you, Mike, for asking. As you know, I won't speak to the exact size or whether it's bigger or smaller from one quarter to the next. What I can tell you is that we continue to observe ongoing impacts from the shutdown. As you know, the federal government of the U.S. doesn't come back to life instantly, and we are partnering with our clients on those deals and renewals that slipped out of Q3 and into Q4.

Yaoxian Chew executive
#31

Alright, now we'll take a question from Kingsley Crane at Canaccord, and after that, Derrick Wood with TD Cowen. Kingsley, please proceed. Thanks a lot; it was good to learn about the Duo Agent expansions that came even before GA.

Kingsley Crane analyst
#32

I realize those who invest with us can't wait to observe what Duo Agent does out in the market. Which product proof points indicate that GA readiness has been reached, and beyond that, how should we think about the pace at which adoption will build during fiscal 2027? Thanks.

Yaoxian Chew executive
#33

Kingsley, would you mind asking that again? Your audio dropped for us. Are you able to hear me?

Kingsley Crane analyst
#34

Yeah, is this coming through?

Yaoxian Chew executive
#35

Yes — right when you got to "investors are excited," we stopped hearing you, so could you go ahead and ask that again?

Kingsley Crane analyst
#36

Absolutely — so the market is clearly enthusiastic about what Duo Agent will do, and what it will eventually mean. I'd like to dig into which product proof points you're looking at to determine GA readiness, and then how we should frame the 2027 ramp. Thanks.

Bill Staples executive
#37

Yes — to gauge readiness, we've established several criteria. The most important ones are reliability, performance, and how stable the platform is overall when it comes to satisfying what our customers expect. We're also assessing how well the agents respond and how responsive they are — both the ones we develop and the custom agents our customers create themselves — and whether those responses are high quality. Then, last, we clearly have to make sure we satisfy our own demanding security bar and don't release anything with vulnerabilities or put customers at risk of any vulnerability. So beyond customer adoption and usage, there are several quality-related criteria we track. As I said, we believe we're getting close to satisfying every one of those criteria within the next few weeks. When that happens, we will announce general availability. As for adoption and usage, predicting the exact pace over the coming quarters is difficult. I'll simply say again what I've said before: 70% of our revenue comes from self-managed customers, and those customers need to upgrade before they can use Duo Agent Platform. Getting most of that customer base moved to a new version frequently takes several quarters. So compared with a purely cloud SaaS operation, we'll see some sluggishness there. I'd also add that I'm quite enthusiastic about the chance to bring Duo Agent Platform to the public sector, especially since that's come up in discussion today. A lot of AI tools available now depend entirely on models hosted in the cloud, but Duo Agent Platform provides cloud-hosted models as well as the capability to operate in fully air-gapped settings or with custom self-hosted models — a setup many of our public sector customers already run. So we're eager to get it into those environments as those customers are in a position to adopt it.

Yaoxian Chew executive
#38

Alright. Moving on, we have Derrick Wood with TD Cowen, and then Cyrus Motika from Wells Fargo. Derrik, please proceed. Derrik, can you hear us? It appears your line is muted.

Derrick Wood analyst
#39

My apologies for that. All right, is this coming through? We hear you. Yep, fine. Appreciate it, everyone. James, would you break down for us what sits inside the net revenue retention rate — seats against tier upgrades against price yield? And I believe in last year's Q4, you had one really big seat expansion deal. Could you give any color on how we should think about what that means for NRR as we lap that sizable deal and reach Q4 this year?

James Shen executive
#40

Derrick, I'm glad to walk through how DBNR broke down this quarter, and I'd also like to address this particular disclosure in a wider sense; so Q3 looked a lot like Q2, with seats making up a bit more than 80% of the mix. Yield accounted for roughly 10%, and up tiering provided the rest. As our business keeps changing, this disclosure is going to matter less, partly because we've moved from being a two SKU company to having multiple SKUs, and partly because we're adding usage-based business, which Bill mentioned, on top of our seat-based business. So in the quarters ahead, we plan to share more on this subject.

Yaoxian Chew executive
#41

Thanks. Up next we've got Cyrus from Wells Fargo, and after that Jason Celino from KeyBanc. Cyrus, the floor is yours.

Ryan MacWilliams analyst
#42

Hi everyone, Ryan MacWilliams here, filling in for Cyrus. You've moved beyond the price increase, and you're putting additional capabilities into the higher-tier plans. Competitors offering premium tiers, such as Cursor and Claude Code, charge $200 per month for those offerings. Given that, as AI plays a bigger role in DevOps as a whole, do you see a chance to raise prices on GitLab's premium tiers? Or is monetizing it through usage the more probable path? Thanks.

Bill Staples executive
#43

Our approach is indeed to monetize via usage. In my view, the correct strategy over the long run for monetization involves a pricing structure that delivers an even trade of value against cost. What I observe from rivals regarding AI pricing is genuinely inconsistent, and it has shifted quickly. I anticipate some change when it comes to what we charge, yet instead of adding yet another per-seat charge, as I mentioned before, we will shift toward a model based more on consumption, in which buyers can commit funds in advance for usage so they obtain the most favorable rates. That commitment, however, represents a shared pool of usage available to every user. Having piloted this and rolled it out to customers, they are quite enthusiastic about the model. And in the end, I think once buyers feel that enthusiasm and recognize the value, they purchase more as time goes on.

Yaoxian Chew executive
#44

Alright, now we'll take Jason Celino with KeyBanc, and after that, Steve Koenig from Macquarie. Jason, the floor is yours.

Jason Celino analyst
#45

Great, thanks. Bill, during your prepared remarks, you shared a few interesting figures regarding the deployment activity happening across the platform. I can't recall the precise percentages, but what portion of this heightened activity do you attribute to customers building applications for AI — that is, the underlying development work — versus what comes from improved productivity thanks to AI coding tools? I hope you see the distinction I'm drawing, but I'm curious about what you're observing.

Bill Staples executive
#46

Yeah, watching AI coding produce downstream effects on the platform is exciting. I'd say a combination of factors drives it — likely both dynamics you just described. Yet in the end, a software team isn't merely focused on the code it produces; it's focused on the innovation being delivered to customers. That encompasses the complete software lifecycle needed, spanning everything from planning those changes, to testing, to integrating, to deploying, and to confirming they satisfy security and compliance requirements. That's precisely what GitLab handles. And since code volumes keep rising, since engineers can tackle more projects more quickly, we observe that acceleration across the remaining stages of the software lifecycle. Up to now, none of those stages had AI acceleration. That's the work we're undertaking with GitLab Duo Agent Platform. And when we deliver acceleration across the whole lifecycle, I think we'll start seeing the monetization benefits discussed earlier on this call, since customers also want those advantages — to speed up not only code generation but the whole software delivery process.

Yaoxian Chew executive
#47

All right, our next caller is Steve Koenig with Macquarie, and after him we'll take Miller Jump from Truist. Steve, please proceed.

Steve Koenig analyst
#48

Okay, thanks. Yeah, can you guys hear me okay?

Yaoxian Chew executive
#49

Sure, we can.

Steve Koenig analyst
#50

Excellent. All right. So, to follow up on what was just asked, Bill, I get that the platform's worth spans the entire software development lifecycle in a unified way; however, I'd like to know, once you roll out Duo Agent Platform as generally available and people start picking it up, which parts of the lifecycle do you expect to see the quickest gains in productivity from? And then, if I may squeeze in one more, I'd appreciate additional detail regarding the softness in SMB. Is that driven more by macro conditions or by something in your execution? Many thanks for taking my question.

Bill Staples executive
#51

Yes, I'll take the first question about GitLab Duo, and James, perhaps you can handle the SMB one. So regarding GitLab Duo Agent Platform, what matters most to grasp when comparing it with other AI tools is that our approach is genuinely a platform one for AI, which means customers are able to leverage what the platform offers — now AI native — to orchestrate actions using AI tools across any category of engineering problem. For instance, we've watched customers benefit from it when it comes to assisting them with planning and documenting what they intend to tackle in advance, before any code has even been produced, to assist with analyzing bugs and with triaging and prioritizing the work that has to get done, in code. We've also watched them benefit from GitLab Duo Agent Platform to write and to review code; we've watched them benefit from GitLab Duo Agent Platform to perform security analysis, to perform prioritization according to the advanced characteristics that we capture through our security scanning capabilities. We've watched them benefit from it when it comes to troubleshooting pipelines that fail when, you know, code doesn't pass the quality standards, security standards, or other compliance guardrails that a company has established, so it truly spans everything, and that's what makes it so thrilling because, you know, having now spent many decades in software engineering, the process of software engineering is extremely complex, and there are any number of ways things can break down. And what customers will be capable of doing with Duo Agent Platform, rather than waiting for a human to engage in a manual process to recover from any one of those failure classes or any of those work tasks, is now apply an agent that can automatically work on their behalf to triage, to analyze, to debug, and to recommend a fix or even automate a fix. And we think that's what will bring incredible value to our customers. James, on the SMB question?

James Shen executive
#52

Okay, so just briefly regarding SMB — that's a weakness tied to a particular segment, and we've been pointing it out across several quarters at this point. Let me mention a handful of things. First, as Bill discussed, we've got a really robust no-cost option, and within this segment we're noticing sensitivity around pricing. So sensitivity applies both to price and to spending in general. And when these clients approach their renewal dates, there's considerable examination and review of licensed usage. SMB makes up only a minor portion of what we do — approximately 8% of ARR — and for our guidance we're taking the position that this softness will carry on through Q4.

Yaoxian Chew executive
#53

All right, we'll take the next question from Miller Jump at Truist, and after that, Mark Cash from Raymond James. Miller, the floor is yours.

Miller Jump analyst
#54

Hey, great. Thanks for taking the question. In the prepared remarks, you all said that every one of your biggest expansions this quarter involved customers who use some kind of AI tooling. So I'm curious — at this point, do most of your customers use AI tooling, or does that point to a smaller subset of the group? And was there any difference in what drove growth for those accounts when it comes to the seats, customer yield, and up tiering that you discussed for the broader business? Thanks.

Bill Staples executive
#55

Yeah, I think last quarter I shared this. We ran a customer survey in which we looked at a few different questions about customer AI tool usage and their forecast regarding, you know, increased GitLab usage as well as seats. In that survey, we did see fairly pervasive use of AI tools alongside GitLab. It's important to remember it's an and not an or. Many times I've heard investors call some of these other AI tools competitors. While there's clearly some overlap between what we're doing and what they're doing, ultimately customers view them as complementary, because they serve a variety of different use cases and support one another. So yes, I believe AI tool usage is pervasive across our customer base, with many of them implementing multiple AI tools as part of their current AI strategy. And I believe we're in a good position with Duo Agent Platform to capture our fair share of that demand, because we're solving inherently different problems than other AI tools on the market.

Yaoxian Chew executive
#56

Thank you. The last question is going to come from Mark Cash with Raymond James, and then I'll hand things over to Bill for some closing remarks. Mark, the floor is yours.

Mark Cash analyst
#57

Yeah, thanks. Bill, to expand on what was just asked — during this quarter, we've noticed bigger companies, OpenAI and Google among them, rolling out additional DevSecOps capabilities, and smaller AI-focused firms doing the same. I'm just wondering how you plan to put that $1.2 billion in cash to work, along with the very robust free cash flow we're seeing, to perhaps sharpen GitLab's edge across the software development lifecycle when competing with those companies. Thanks, I appreciate it.

Bill Staples executive
#58

Yeah, that question really has two aspects, I'd say. One is how I view GitLab's competitive position against both small and large vendors. The other involves capital deployment. James, perhaps you could address that second aspect. I'll handle the first one.

James Shen executive
#59

Yeah,

Bill Staples executive
#60

honestly, when you break down what goes into an agent, every single one really has four components. One is an LLM, and just like nearly every other vendor out there, we give access to all the main foundational LLMs available on the market. Those costs are only going to keep dropping. Their quality is only going to keep rising. There isn't much room for differentiation there. It's basically the electricity powering any electronic system. The second component of an agent is the prompt, which guides the LLM toward solving the problem. That's defined, you know, using human language. And once again, the IP value here is pretty shallow. Tons and tons of open source prompt libraries exist and are out there for anyone. There are only so many ways to instruct an LLM to solve a problem. And, you know, we give Duo agents excellent prompts right out of the box, but we also let customers customize and extend those prompts. So there's a bit of extensibility beyond what competitors currently offer. But those first two components, I'd say, are mostly commodity. And it's truly the latter two components that set GitLab apart and that even the big AI vendors can't equal. The first of those is context. We offer not just the system of record for all the source code our customers store, but every change made to that source code over time, all the testing and quality validation of that source code, the security of that source code, all the associated plans and bug tracking and everything else. All that context feeds agents the ability to reason and make good decisions. And almost no other competitor has the breadth of context we have as part of our unified platform approach. And we think that's a lasting differentiator over time. The fourth component that agents have or are made up of is tools to actually take various actions on behalf of users. And here again, our unified platform really stands out as a strength because anyone can generate code. It basically means generating strings that get written into text files, right? But when it comes to full lifecycle software engineering, you're talking about far more sophisticated operations, everything from planning and testing and securing and integrating and deploying code that requires a rich set of capabilities that have to be integrated one with another. And we've delivered that to humans for over a decade now. But with Duo Agent Platform, we're unlocking all of those rich capabilities for agents as well. So agents can take those actions. And that's the really exciting differentiator and value we provide our customers that I really don't think either small or large AI competitors can match.

James Shen executive
#61

Right, and regarding where we stand with cash — that $1.2 billion in cash and investments genuinely places us in a strong spot given current market conditions. Also, we've built a solid history when it comes to fiscal discipline, and we're always evaluating the best possible paths for allocating capital that will maximize value for both our customers and our shareholders.

Yaoxian Chew executive
#62

Alright. That brings our question-and-answer portion to an end. At this point I'd like to hand things over to Bill so he can share some final thoughts. Bill, the floor is yours.

Bill Staples executive
#63

Hello, and thanks to all of you for being on the call today. Having reached the one-year mark in my time at GitLab, I feel we are performing better than at any prior point, guided by a plan aimed at responsible growth at scale. As we already discussed, we are adding to our sales capacity within our field organization and putting money behind a dedicated first order team so we can capitalize on our expanding TAM. Product innovation and differentiation are picking up speed too, and we are proving we deserve to shape what comes next in software development with AI; we stand right at the edge of announcing general availability for our agentic AI platform, which will shift our business model away from being purely seat-based toward a hybrid of seats plus usage, opening fresh routes to bring value to our customers; all of these amount to truly meaningful structural upgrades at GitLab. One item that stays the same: we continue to be dedicated to investing and building for responsible growth that drives shareholder value. I'll end the call the same way I began it. There has never been a more thrilling moment to be part of GitLab. James and I are in Phoenix, Arizona this week, and we will be presenting at the UBS Global Technology and AI Conference. We hope to see you there, or somewhere else during the quarter. Thanks once more, and good night.

Read the full transcript via the API

You're viewing the first half of this call. Get the complete GitLab Inc. transcript - plus 131,000+ transcripts from 7,000+ companies, speaker segments and full-text search - through the EarningsAPI REST API or hosted MCP server.

Get an API key View API docs →

For developers and AI pipelines

Programmatic access to GitLab Inc. earnings transcripts and 131,000+ others is available through the EarningsAPI REST API and the hosted MCP server. Quarterly plans from $162 - full transcripts, speaker segments, full-text search, and the /api/v1/transcripts/recent polling endpoint for ETL pipelines.