Welcome, everyone to the RBC TIMT Conference, morning number one. Always very exciting. Very pleased to say to have Mark Jenkins, CFO of Carvana, here with us straight from Arizona. Thanks for being here.
It's my pleasure.
Nice to see you. Yeah. O bviously have my exhaustive list of questions, but I think Mark's gonna start us off with some prepared remarks. And then, for anybody in the audience who does have questions, just feel free to raise your hand and we'll try and get to as many of those as we can. So Mark, with that.
That's great. Well, thank you very much for having us. We're happy to come and talk to RBC and all the investors that are here today about Carvana. I did want to start today with just a few of the key takeaways from our most recent earnings. I think the reason we wanted to start that way is an awful lot has changed with Carvana in the last six to nine months. We're really a very different business today than we were you know six or nine months ago. And also, I think you know over the course of our last few earnings releases, we've also made an awful lot of new information available to investors and analysts that we wanna make sure everybody's aware of.
I think it allows a much deeper dive into the economics of Carvana, and also a view into where Carvana can go from here. So I'll start by basically digging into those key points. So I think I think most people are familiar with Carvana. I n our first phase of growth, we were one of the fastest-growing technology companies. I think we were one of the 4 fastest companies to make the Fortune 500, matched with only Facebook, Amazon, and Google. That was our first 8 years, which you know grew at a tremendous rate. 2022 was a challenging year for us. As many people know, interest rates increased rapidly. That had a big impact on our industry and our business.
We took a step back for the only time in our company's history on our key unit economics measures of gross profit and adjusted EBITDA margin. And then now fast-forward to 2023, and we've, you know, dramatically changed the business. If I move to the next slide, we've set company records for first quarter gross profit and adjusted EBITDA margin, second quarter gross profit and adjusted EBITDA margin, third quarter gross profit and adjusted EBITDA margin, so far this year, and that's all excluding any benefits from non-recurring items, which we have had some non-recurring benefits this year. But all those points hold, excluding those non-recurring benefits. T hat's, obviously, has us very excited about where we can go from here.
Then, you know, in the next 3 slides, I'll really focus in on some of the key points from our most recent earnings. So first, I think one of the things that has happened here in 2023 is we have really turned a meaningful corner and, you know, set some really great new high watermarks on profitability. So, I think there's the GPU story where I just mentioned, you know, we've set record GPUs for the first 3 quarters. Some of that's driven by our acquisition of ADESA, which has been a contributor to GPU, but much of it as well is driven by fundamental improvements in our core business. Most importantly, I think the last 2 quarters have really highlighted the profitability of our model.
I think that was something that investors had questions about for a period of many years, but in the last two quarters, in the third quarter, we did, you know, $148 million of adjusted EBITDA, but inclusive of $40 million of non-recurring benefits. So we did $108 million of adjusted EBITDA in the third quarter. Excluding non-recurring benefits, that's a run rate of between $430 million and $440 million on an annualized basis. And we did that in an environment that is not particularly attractive for used vehicle retailers. I think interest rates are still at multi-decade highs. You know, the industry is down overall. Used car prices are still significantly elevated relative to goods in the rest of the economy.
And a point that I'll get to later, we also did this despite carrying very significant excess capacity that we think enables, you know, strong, profitable growth in the future. So I think, you know, our $108 million, excluding non-recurring benefits of adjusted EBITDA in Q3 in a tough environment, we did just under 100, about $85 million of adjusted EBITDA in Q2, excluding non-recurring benefits. So that's two quarters in a row of very strong results. We think that answers the question on profitability and also, positions us extremely well for future growth. Okay, so moving on to the next, I think, key takeaway. Our, our current focus, and, you know, we have this three-step plan that I think people that have followed us are familiar with.
Step one was to get to adjusted EBITDA profitability. Step two is to drive significant adjusted EBITDA per car sold, and step three is return to growth. We're in step two today, and what we've really been focused on is improving the unit economics of the business, in other words, driving down variable costs per car. And the reason that we're focused on that is obviously, the more gains we make on reducing variable costs per car today, the more profitable our growth can be in the future when we turn the corner to step three of our plan, return to growth, and make sure that that's very profitable growth because of our strong unit economics and our excess capacity. And so just a couple of key highlights on that, on that point.
So in the last two quarters, we've reduced these, you know, more variable expenses, whether it's non-vehicle cost of sales, you know, that affects gross profit, or operations expenses, which are the more variable component of our selling, general, and administrative expenses. We've reduced those two combined by about $1,000 per car sold just over the last two quarters because we're focused on efficiency in step two. We think that we do see opportunities for further gains, but we think that is a really valuable transition step to going back to returning to growth, is improving those unit economics. Now, moving on to the third key point. So one of the things that's unique about our situation today is we are, we have significant excess capacity throughout our entire business.
Now, that's if you look at, you know, how much inspection and reconditioning capacity do we have? How many parking spaces do we have that are free for future storage and growth? How many extra haulers do we have in our logistics network? How much extra customer care space do we have? We have excess capacity throughout the entire business to support significant amounts of future growth, and we think that's a really powerful thing for a couple reasons. One, we think having that excess capacity makes future growth much, much easier than the growth that we, you know, had to operationalize in 2013 through 2021 because we built it, and we don't have to build it again, at least for some time.
And then secondly, we think that provides a really strong economics per unit story, because we do have very elevated fixed costs or, you know, overhead costs, which are primarily fixed today, relative to where we've operated the business in the past and relative to where we expect to operate it in the future as we utilize this infrastructure. And so I, I think that all this data, you know, we made available in our Q3 letter, but we think it tells a very powerful story, not only about where we are today, but also where we can go in the future from the standpoint of driving profitable growth. And with that, I, I'll, I'll stop.
I think those are the three key points about where we are today and what we're focused on today, and what we're excited about in the future with respect to growth and excess capacity. With that, I'll stop, and we can take some Q&A.
Awesome. Cool.
Thanks.
Well, thanks for that. I guess I'll start out, you know, we've, we've definitely gotten a couple, three sort of hot-button questions or common questions since the print. I think the first one, and you, you kind of, you know, I think you set it up pretty well, is that you've, as we've gone into this more difficult environment, higher rates, lower volumes, et cetera, and, and needing to sort of cut out those costs, you've kind of, you know, I think Ernie has used the word throttled at times, right? From a throughput perspective and, and everything. Let's assume things stay stable here going forward. What are the considerations as you think about returning to growth? Obviously, you've been working on these profit mandates sort of heads down for the last, call it year, five quarters or so.
What would be the things that would cause you to bring volume back?
I think we get that question frequently. And I think the number one thing that we're focused on in terms of deciding when to return to growth is our progress on our profitability initiatives and, in particular, our progress on our unit economics initiatives. D riving down non-vehicle retail cost of sales, driving down operational expenses per unit, making the business more efficient, not only purely from a unit economic standpoint, but also the projects that drive stronger unit economics also tend to be efficiency projects that make growth even easier when you decide to return to growth, because they're focused on automation and standardization of processes across the nationwide network and simplification of those processes.
I think the number one thing that we're focused on in terms of deciding when to transition to return to growth is our progress on those unit economics and efficiency initiatives.
Got it. Okay. And then I guess just from a volume perspective, what are the things you would look to affect if you were to do that?
A s we think about, okay, one step two, you're sort of ready to move on from step two onto step three, what does a return to growth look like? I think the first step in that return to growth is transitioning the focus of the business. T eam by team, all of our teams are focused on profitability initiatives, right? Making the business more efficient. I think when it's time to return to growth, those teams will transition to projects that are focused on growth. Just to use one example of that, a project that is both an efficiency project but that could turn into a growth project, is our same-day delivery pilot. W e've started to pilot same-day delivery in several markets.
I think that fits in step two, because same-day delivery does actually make you more efficient because you're more likely to do shorter distance deliveries, which is operationally efficient. But something like that, over time, can become a growth project as well as it gets rolling out, nationwide, and we really lean into it more. So, I think that's an example of the type of a project that today is an efficiency project, but could become a growth project.
But really, that, that's the first step. I think beyond that, I think getting the flywheel kicked off that we benefited so much from, from 2013 to 2021, I do think that means building selection. S tarting to grow inventory again, building awareness and credibility of the brand and online car buying in general. So that, you know, that could mean more advertising on that side of the flywheel. But I think that getting that flywheel rolling again, that we benefited so much from in 2013 through 2021, I think is a second component of returning to growth.
Got it. Okay. And then, o n that point, maybe go a little deeper, too, because you guys were ramping up volume so fast and, you were building out new org structures, you were bringing on tons of new capacity, new IRCs, new vending machines, which everybody loves. What were the kind of learnings from that perspective as you think about returning to growth? Because you mentioned, you just mentioned some of the components, but when you think about particularly getting people back in the IRCs and that kind of aspect of it, what are the challenges that you face if and when, I should say, when you return to growth?
Sure. A bsolutely. So I think there's a few learnings. I think one learning that I think is perhaps most important as we think about returning to growth in step three, is based on everything that's happened in the business over the last couple of years and all the improvements that we've made, we really think growth into an existing nationwide infrastructure will be much easier than growth, where you had to build that nationwide infrastructure alongside your rapid growth. And so when we were growing from really 2013 through 2021, and especially in the later years, you know, we were adding many large distribution centers. We were setting up new logistics routes.
We were adding many more managers of those facilities and things like that, and that's a much more difficult form of growth, and we were obviously growing incredibly quickly, 70%+ over the multi-year period. And we really think that growing into an existing infrastructure is a much easier form of growth than growing the infrastructure while you're growing.
Got it. Okay. And then you showed the chart, I think on the capacity utilization. We can't see the television. I think that was the other kind of big question we got coming out of the earnings call was, you have all of this apparent excess capacity, right? But you're also running GPU at kind of record highs, even excluding out the one-time stuff. You brought down your SG&A per car a ton.
So I guess when you think about that capacity utilization chart, what portion of the cost base does that represent? H ow to dimensionalize that, or maybe put differently, what does that represent between variable and fixed costs, right? Because, again, like, we know there's going to be cost as you build out more volume, but help us with that piece of it as it translates to GPU, to SG&A per car, if you can.
I think the most important connection between this excess capacity that we, you know, have invested in and now hold today, and the cost structure and the financials, is the way that it rolls into our overhead expenses per unit. And I alluded to this a little bit earlier, but in our Q3 shareholder letter, we broke out some details on operations expenses per unit, which are the more variable expenses. There are some fixed components, but they're more variable. Those, I would say, are not overly impacted by our excess capacity. And so, you know, when we think about the business our goal, as we grow units, we will grow operational expenses in dollar terms.
We'll seek to reduce them in per- unit terms because of some of the sources of leverage in those operational expenses. But, you know, those will grow in dollar terms as we grow. We'll seek to reduce them in per unit terms. Where the excess capacity really comes in is in the overhead expenses per unit, which are the more fixed components of the cost structure. And there you can see in some of the data we provided, you know, we're many hundreds of dollars. I think , if you just look at the data relative to some of our earlier low points and where we were, I think it's, you know, on the order of $950 or so higher on a per car basis of overhead expenses relative to our best points.
That's where we really see significant opportunity. I think, with the excess capacity that we build, we really believe that when we grow, yes, the operations expenses will grow in dollar terms, but we want to try to drive them lower in per unit terms. Then the real opportunity is in that overhead expense bucket and really levering that, in selling, general, and administrative expenses. I think that's the key to connecting those things.
Okay. And then just from a GPU perspective, you know, you've, you've ramped that a lot more than I think anybody would have thought 12 months ago. And, you know, obviously, you've had some, some write-down situations, and you need some normalization on the, on the securitizations to, to sort of wash through. But setting those things aside, what are the kind of the main inputs for further GPU leverage, if there are any?
Sure, yeah. So we absolutely see opportunities for further improvement in GPU. I think that takes a few different forms. One, I think we do see opportunity to drive down non-vehicle retail cost of sales further. You know, we've made amazing progress in the inspection and reconditioning centers, lowering non-vehicle per unit costs through, you know, better use of technology, insourcing, process standardization, things like that. We've also made progress on the inbound transport side of non-vehicle retail costs, see further opportunities in those areas. We see a lot of opportunity in the wholesale part of the business. I think we still feel like we're in the beginning days of fully scaling our business of buying cars from customers and selling them in the wholesale market.
We think we've made some progress there, but certainly see few further opportunity. We also see opportunities at ADESA. You know, ADESA growing nicely year over year from a unit volume and profitability perspective. You know, that industry is still, you know, at relatively cyclical low point, although it has improved a little bit year over year. So certainly see opportunity there within the broader wholesale umbrella to continue to drive volume and profitability at ADESA. In other, we see opportunities as well. I think the, you know, it's some aspects of our corporate story have have caused us, when we go to the securitization market, to actually be realizing wider spreads than very comparable issuers, including issuers where our performance, you know, is equal or even better.
But we're still paying higher spreads in part because of some of the headline noise over the last year plus, in our corporate story. So we see opportunities there, I think. And then also in other, we're always working to make progress on the, you know, the data and scoring side of ancillary products and the finance platform.
Got it. Okay. And you mentioned, just on the rates and the spreads, let's say we stay in an environment of sort of flat interest rates, roughly over the next 1-2 years. What happens to that component of other GPU under that scenario?
So I would say, almost reiterate my previous point. I think we see lots of opportunities in other GPU. I would say, generally speaking, a neutral rate environment is neutral, a falling rate environment is positive, a rising rate environment is directionally negative for our finance GPU in the near term. Particularly if, you know, during the transition period where the rates are moving in one direction or the other, we do hedge part of those movements, but not all. A nd so then what I think in a neutral rate environment, which is what you asked about, you know, I think we'd really be focusing on the sources of fundamental gains that I alluded to, which include, you know, spread normalization relative to other mature issuers, as well as our internal, you know, scoring, pricing, and ancillary product opportunities.
Yeah. Yeah, got it. And then, oh, yeah. Question? Yeah.
I'm curious, Copart has had some success organically on the whole car market. Do you see them in your wholesale and ADESA business?
So I would say that, I think our, you know, 2023 has been a good year for ADESA so far. I think they're meaningfully growing marketplace volume. And then, you know, they're also seeing, you know, noticeable improvements in profitability, in large part as a result of that marketplace volume. And so I think we'll look to continue to make gains there. I think we have, you know, a lot to offer dealers over time, in the wholesale auction market. You know, I think, you know, we have very strong reconditioning capabilities. We have, you know, a first-party transport network that we believe can be very efficient today and we believe can be used to help dealers over time.
So I think there's a lot that we're very excited about. One thing I didn't mention, by the way, is we launched a new digital auction platform at ADESA to make the transaction process much more seamless for dealers who are participating in the ADESA auctions. And so I think there's a lot we're excited about in terms of our offering and where our offering can go over time for auction customers.
Specifically to Copart, do you see them competing with you at all in that space?
I think that's possible. I think they're the best party to ask about that. But I think what we're focused on, just like we're focused on in other parts of the business, is the strength of our offering, and we think our offering is very strong.
Questions? Maybe one more just on ADESA broadly. I think, you know, the timing of that acquisition was a little bit challenging just because of what was going on in the macro. And again, you know, adding so much capacity at a moment when volumes were turning south. I guess, obviously, you wouldn't have done anything differently, but maybe just talk about why, where your conviction is, I should say, on ADESA, in light of the sort of downturn in the market, 'cause that did. It added a ton of capacity, and it was just tough timing. So maybe talk to that, if you could.
Sure. A bsolutely. I think we're very happy with the acquisition. That may feel like an odd thing to say to some, but we are. We're extremely happy with that acquisition. I think the ability to be able to acquire an automated automotive asset that is, you know, that unique, and we believe that valuable. I think when you have the opportunity to acquire an asset like that, and there's such strong synergies and opportunities for the long term of your business, we think it's the right move, and we still absolutely think that. I think in terms of the some of the additional reasoning behind that, we do think ADESA's having a positive impact on our business today.
We've talked about efficiencies in wholesale vehicle sales, for example. I t's much easier for us to buy cars from customers and get them to wholesale remarketing locations. As I mentioned, ADESA EBITDA climbing year-over-year, volumes climbing year-over-year. So we're pleased with the progress that we're making in that business specifically. But then longer term, going back to that excess capacity chart we just think we think a business that has very strong unit economics, and depending on exactly how you look at it some would say, "Hey, the industry best unit economics," once you kind of dig into our GPU and our operational expenses.
Pairing that with significant excess capacity, which ADESA, helps with, and we didn't have a ton of, ADESA reconditioning capacity in those excess capacity charts because it's not built out yet. But on things like parking spaces, some inspection and reconditioning capacity, ADESA is absolutely playing a role in giving us that infrastructure. And so we just think that's a very powerful thing, as we look forward to be able to have a nationwide network built that we can grow into. I think that adds a lot of strategic value as well.
Got it. Okay. And then we talked a lot about that presumptive return to growth at some point. You guys have been very profitable, though, recently, both on a unit economics basis and an aggregate basis. Do you, when you think about the obligations you have, in terms of the balance sheet over the next couple of years, do you need to return to growth? What do you contemplate as you, as you make those choices?
O n the specific question of: do we need to return to growth? W e think the answer to that is no. Obviously, the level of adjusted EBITDA that we're already generating over the last couple of quarters, plus the opportunities that we see for improvement the business is generating meaningful adjusted EBITDA today, and we believe there's opportunity for improvement in that, even at today's volume levels. Having said that, that's not the plan. We do plan to transition to step three and return to growth. And as I mentioned, we're excited about that because we think we think it can be very profitable growth.
J ust in terms of the way we think about that, if you go back to some of the data that I was talking about earlier, we've really driven down our non-vehicle cost of sales, you know, in the inspection and reconditioning centers. We've really driven down our operational expenses per unit, which are the more variable components of selling general and administrative expenses. We've, you know, really do have this elevated overhead cost structure today. And so what we think is, when we start to return to growth, that gives the opportunity for real EBITDA leverage, and particularly SG&A cost leverage. And so I think, I think then you get to the point where, hey, you know, You do want to return to growth w hen your step two projects are done, and that has the opportunity to be very profitable growth.
Got it. Okay. And then I have to ask.
Hi, just a question. Just on your last comment, you, you'd said that you have, like, all this excess kind of overhead, running $950 per unit higher than you would need to. By keeping that expense, aren't you saying that you will return to growth at relatively short order, b ecause if not, wouldn't you cut it? I just want to, again, get how you've thought about it.
W e haven't given a specific timeline to return to growth. We're in step two of a three-step plan. The second step is continue to make efficiency gains. The third step is return to growth. I think we haven't given a specific timeline to return to growth, but I think it should be expected that, you know, it's not a long-term timeline for returning to growth. Like, we are really focused on our step two initiatives today, and then we'll transition to step three. As a result of that, to your question, yeah, we're not really looking to divest capacity. W e're very happy to keep the capacity that we have today because of our prospects for growth in the future.
I think one of the reasons that we feel very comfortable doing that is in Q3, adjusting for non-recurring items, we generated $108 million of EBITDA, right? While carrying the cost of this excess capacity, and that's a pretty darn good number. So I think there, yes, while we haven't attached a specific timeline to it, we do expect to, you know, transition to step three in the not-too-distant future, and that capacity becomes very valuable in that context.
Good. All right, we've got a little lightning round. We only have a couple minutes left. And maybe I'd even ask you to take your Carvana CFO hat off for a second, if we, if we can. So obviously everybody, I think most companies at this conference have been fortunate enough to save money on cloud expense over the last year, and some of that's, you know, in your case, idiosyncratic, right? And some of that's the big hyperscalers coming to companies and say, "Hey, let's, let's figure out a way to help you save money." So that was 2023. What does 2024 look like from an, from an industry's perspective? And not necessarily asking you from a Carvana perspective. Do you think there's more opportunity to save in 2024, like 2023, or do you think we've kind of reached a new baseline, would you say?
Sure. So I can speak for sure about that from Carvana's perspective. I think that, you know, we had. You know, one of the many areas of focus on operational efficiency of the business over the last, you know, 12-18 months, has been getting efficient with our cloud spending. I think our goal will be to continue to get more efficient in 2024, and we see some further opportunities there. And then it may be a company-by-company, you know, a question about exactly how other companies are approaching it, but our approach will be to try to gain additional efficiencies in 2024.
Got it. Okay, two more quick ones. Managed to get through the whole time without talking about Gen AI, which is a miracle. Most interesting Gen AI use case for you guys, or just that you've seen, that not enough people are talking about?
Sure, yeah. I think there's many areas where we see opportunities for AI to play a bigger role in our business. I think a few of the leading ones, certainly, customer communication is a big one. Content creation is another one, where we see opportunities. Making our internal staff more efficient through, you know, internal content creation as well as external content creation. Then in software development as well, we see opportunities. I think those are opportunities that other companies are seeing as well, but we, we do believe we see very real, tangible, potential benefits, in a handful of those areas. We also think, you know, in the longer term, being a scaled national player, we believe, allows more efficiencies, from technologies like Gen AI than, you know, having less scale, having less data.
And so, you know, we're optimistic about our ability to take advantage of that over time.
Got it. Okay, and then last question, just a few seconds left: What's the hardest decision Mark, the CFO, has to make over the next couple of years?
I think the biggest decision that we'll have over the next couple of years is how fast to grow. I think it's a question we expect to get from a lot of investors, and I think, you know, we'll want to strike a balance between rapid growth, which is something that we've achieved and had great success with in the past, but also, you know, making sure that that growth is paired with strong profitability.
Got it. We are out of time, Mark. Thanks for being here. Really appreciate it.
Thank you. Really enjoyed it.
I'm really excited for this one. I'm spending a lot of my time talking to CEOs and CFOs over the next two days, but I get to talk product here in this presentation, and so it's, I'm really looking forward to this. It's a name that we've. For those that don't know GitLab, we've known them for a long time and have been big fans of the technology.
So I'm gonna start out with some questions. If there are some in the audience, and it's a bigger group, so I will certainly leave some time at the end for folks. These guys are under a quiet period, so there's gonna be nothing about any forward-looking comments of any nature on that side. So with that as a disclaimer, David, David DeSanto, Chief Product Officer at GitLab. Maybe for those that don't know David, could you, you know, maybe tell us a little about yourself, your journey? I think you've been at GitLab for a little over four years now. But yeah, maybe just a little bit about your background.
Yeah, absolutely. So I started off my career in software development, cybersecurity, and IT, and I have been across both product and engineering organizations for the majority of my career. In 2019, I did join GitLab. I actually joined to add security and compliance into DevOps, which I think you all now know is Ultimate. Over the course of my time at GitLab, I took over products, and as GitLab's Chief Product Officer, my primary focus is on setting vision and strategy for the company, and then making sure we can bring that to a reality.
Excellent. All right, well, we got a lot to talk about here. ou mentioned your security, your focus on security. Maybe before we get into that, and GenAI, which you and I have been talking about outside here, can you talk about the importance of a DevOps platform, and why in modern software development that is so critical? And I'll caveat this. I used to be a software developer in my days prior to doing this, and I can tell you, I wish I had GitLab back then. But maybe talk about the importance of DevOps, and the real value of the DevOps platform.
Absolutely. So one of the things that companies are struggling with is doing their digital transformations, and that's because it's tied to very brittle, tool chains that they've built. And so when talking to customers, and as a former engineering leader, I can say this, you end up with dozens of tools that are needing to be stitched together to deliver software. What that ends up doing is it drives artificial silos in the organization. You get concerned about upgrading a component.
From my own career, there was a point where I had to ask myself: "I have to release software next month. Do I upgrade Jenkins now? 'Cause last time I did that, it broke a bunch of my other things that are attached to it." And those are the trade-offs that organizations are trying to make. So a DevOps platform comes in, and I'll say a DevSecOps platform 'cause that's what GitLab is a way to essentially give yourself better visibility, better collaboration, and honestly, better control over how you deliver software. The best way I can put it is in the words of our own customers. About two years ago, we did a survey through Forrester of GitLab Ultimate customers.
What came out of that survey was a couple of interesting data points. The first was, on average, those customers were saying they were 7x more effective at delivering software. They got an ROI of 427%, and it paid itself back in six months. And the reason why that's important is all those things are about delivering software more effectively and also delivering it more securely, and that gives them an edge up in the market. And so when you look at traditional DevOps, which is developers pick the tools you try to stitch it together, and you compare it to a DevOps platform, you get a better acceleration, and, and you deliver secure software.
So one of the things that always intrigued me about what you guys are doing is, you mentioned it, David, you're taking a real fragmented historical market, and you're bringing a platform approach. The question I get from investors a lot of times is: "Is there a trade-off between best of breed functionality and more of a platform approach, and are your customers actually using the entire feature set within the platform versus maybe a particular module, particular use case?
L ook at it from this point of view. I get asked the question a lot, "Well, what's the value of a best in breed versus a platform? And I always say "A platform can be the best of breed." And so for us, we've really focused on five key areas of investment, obviously, source code management, code review, that's the original GitLab. CI/CD, helping people ship software. Enterprise agile planning, which you and I were just talking about.
Before we started, and our ability to bring all planning into GitLab, and then security and governance. Those are the bets that we've made that give GitLab its value, and then those other areas, we consider them better together. They're not maybe best of breed, best in the market, but because they're attached to the platform and you get that additional visibility, they end up being good enough for a lot of customers.
So are you seeing then customers maybe come to GitLab initially for a use case, or is it, "And then, it 's great for CI/CD or code development or code promotion," but then start to expand into some other things? And ultimately, does that then cause them to then look at their tech stack and say, "Well, I'm actually using more of these features. I can eliminate XYZ?
A bsolutely. And so typically, depending on the customer and the industry they're in, I'll talk about enterprise customers primarily today, but they land usually with source code management, CI/CD, and security and compliance. Because if you're going to change how you deliver software, you're gonna modernize it, you might as well grab the three most important things in the portfolio. To your point from there, over the course of their first initial contract, usually their first year, they see other tools that they're now using that feel obsolete in comparison to the platform, and that's when they move into things like planning, our artifact registry, and the other components that are around those, those key areas.
The other thing that, and I wanna get to the kind of the SecOps piece or the full dev SecOps piece, but when you think about what makes it into the paid version, either the enterprise or the ultimate, or, the Premium or the Ultimate version how do you think about w hat stays in Free and what that trade-off? But 'cause you give a lot of functionality at Free.
We give a fair amount of features for Free. T he way I look at it and how I lead the product organization is we ask ourselves, "Well, who's that feature for? What does it benefit?" I f it's a feature that is going to enable teams to be more effective, we put it in Premium. If it is something that's going to benefit organizations, we put that in ultimate. That's kind of how those two tiers work.
When going back to look at Free, we have become very strategic as to what we put in Free. Two examples would be, we wanna do free-to-paid conversion plays. About 2 years ago, we put some of our security scanners for Free. That's not the whole UI, that's not the enforced workflows, it's just the scanner. What we found out is that customers would enable that, and then within three, four months, they're reaching out, asking how they can get to a paid version of it.
And so we're very strategic in what we can put down there to drive that. The other thing is that one of the things that I loved about GitLab, and I was a GitLab user before coming to GitLab, is that community that's been built around it, and that community contributes software into GitLab itself. One of my favorite features was contributed by an enterprise customer for customers in the Ultimate tier.
By having those Free features, it enables them to do their own software development that would contribute back to GitLab. I always have to look at those trade-offs, but I think we've kind of made the goal with Premium and Ultimate being team and organization- related.
T hat's interesting, when you're thinking about then these features that are sort of, like, considered maybe more table stakes and part of the Free version, is that when you think about new feature development, is it, like, who is it trying to resonate with?
W e follow that buyer-based persona for how to build the product. And so, yes, if we're saying, "Oh, that just helps out an individual developer or will drive more community contribution," there's a tendency then to put that in Free. But if it's something that organizations will pay for and it'll drive value and specifically drive up-tiering from Free to Premium to Ultimate, then we make that strategic decision to put it in Free.
I guess from a product development standpoint, how do you think about the self-managed product, you know, development versus the cloud?
Yeah, so actually, I love this question. This is one of the things that brought me to GitLab and differentiates GitLab. Our SaaS offerings, whether that's GitLab Dedicated, our single-tenant SaaS, our gitlab.com, which is multi-tenant and self-managed, it's all the exact same code base. And so when we're looking at how to provide features, we don't have to ask ourselves, "Well, how would we provide that to our SaaS customer versus how I would provide that to a self-managed customer?" A lot of companies end up with two separate code bases because of how they built their product and that now limits them from having a good user experience, developer experience with the product. So we have that benefit that literally GitLab.com is running the same code you can download. We just have scaled the configuration of it to be able to handle the millions of monthly active users.
Are there things, though, that customers are doing with the .com version, the cloud version, and maybe this, we'll get into this from a GenAI perspective that are not maybe as scalable as from a self-managed perspective?
Yeah, so we do, starting early this year, released a thing called GitLab Cloud Connector.
So for something like AI, the self-managed customer can connect and use GitLab's cloud infrastructure to run the AI for their team. And so we're doing that to kind of come over that gap, because to your point, it can be very expensive to build some of the infrastructure. But what I would say is that what we're seeing is if you're a heavily regulated industry, you're in fintech, financial services, you're in healthcare, you're in government, you're definitely gravitating towards self-managed or you're gravitating towards GitLab Dedicated, because we can provide that to our customer, give them data residency, data isolation, and they'll be the only person on that deployment.
Okay, that's. You mentioned a couple important things that I just took, you know, when I hear you talk about financial services or government, those, these are highly regulated industries.
They are.
Can you talk about how GitLab supports customers with some of these more complex regulatory requirements?
Yeah, absolutely. So one of the things that's unique about GitLab is that security and compliance is built in from day one in the product. So if you are in one of those regulated industries, you automatically get the benefits of our compliance. That compliance functionality includes who can merge what code, the visibility to know if there's been a compliance violation. You can set global-wide security scanning and execution policies, and that then allows them to have the visibility they need to go into an audit. One of our customers, who's a call them a voice SaaS platform, you know, recording, dictating, and so forth. They shared with us that they were able to get through their last SOC 2 audit very quickly.
Every time they were asked a question, "Hey, how did you handle this?" or, "Where is this configured?" They could just point to GitLab and say, Oh, here are all those logs. It's captured here. Oh, I can filter it by that for you o r, "Oh, yeah, we have this enforcement here. It can't be turned off. And so, that's how we're supporting them. What I can't stress enough, though, is that GitLab Dedicated part of it. When we launched it a year ago, about 18 months ago, we knew we had something based off our initial demand. But as time has gone on and we've brought that to general availability, we're now extending what regions in the world can be run in, we're seeing a lot more demand for it. Regulated customers don't want to manage their environment themselves.
They want someone to manage it, and they can't go onto a multi-tenant environment, especially if they're a government agency. And so that gives them the ability to have the benefit of everything you could do self-managed, but not have to manage it yourself.
Yeah. It's sort of the best of both worlds.
Yeah, I mean, honestly, when I was a GitLab customer five years ago, I would've looked at it right? 'Cause we had the same thing. I worked for a company who sold into government agencies around the world. We always had to show them we were being secure, and things were stored locally.
F or GitLab Dedicated, i s it 1,000 seats?
It's a minimum of 1,000 Ultimate seats.
I s there an infrastructure cost in addition to the seat count?
Correct, yeah. I t's kind of billed as one SKU for the customer, but it ends up including GitLab Ultimate seats. They select their cloud region, then there's costs associated to that. And then, yeah, there's infrastructure, storage, compute that go with that.
Are you seeing non-regulated industries look at Dedicated as well?
We are. Predominantly, it's mostly been regulated. But no, we do have some customers who are just in, we'll call them high-tech software who are wanting to have their source code better secured. One of the things about GitLab is, it's not just that we sell security and compliance, but security is actually baked into the product. You're more secure using GitLab's source code management than you would be with another product. And so we're seeing some of those customers go, "I really want that because now I don't have to manage it myself, and I know the source code's secure.
Interesting. I have a few more questions. I wanna talk about the Ultimate tier, and then ultimately, we're gonna talk about some of the GenAI capabilities you've got, and then we'll open it up to questions here in a second. But let's talk about Ultimate. Let's talk about DevSecOps. 'Cause, you know, when you guys went public, you were a DevOps company, and now you're a DevSecOps company.
Correct.
It seems like that was a big part of your initiative four years ago.
It was.
Talk about a little bit more about that progress towards really driving this Ultimate on security and why customers are resonating moving to that.
M aybe to give a little context about myself, I mentioned formerly in cybersecurity. I worked for a company where we were saying we were shifting security left, and in my conversations with GitLab, and I'll say predominantly my long interview with Sid, our co-founder and CEO I realized we're actually, we weren't shifting security left. And so that brought me to GitLab, 'cause I was like: I can actually help organizations do better than I can before.
And so then we asked ourselves: If you're gonna truly do that, what does it mean? And I can't stress enough, as a former security person, you don't realize how much your terminology is confusing to a developer.
And so we focused on moving our security scanners to run at code commit time. That means that we're scanning that delta code change. It means the developer doesn't have to wait hours sometimes, and I'll say this, my last security company I worked for, our scanner could take 24 hours to run. Developers can't wait that long. And so we could scan smaller codes of or chunks of code, and we made it developer-friendly. We added in training. We added in prompts to help them understand it. We get to AI, we can talk about how we're now using AI to help with that as well. When that was successful, and I would say that was mid-2020 into 2020, we realized we had something.
And so, to the question about highly regulated customers, that's when we started asking ourselves: If you're a compliance person, you're a compliance manager, compliance director, like, what do you actually need to know that you have a secure software supply chain?
And so that's when we started building out what we call Govern, which is our governance controls across the entire organization. And so that, combined with the security scanning, makes Ultimate the value that it is today. A developer once, who was working on GitLab Ultimate, said, "GitLab secure and protect," in this case Govern, "make GitLab Ultimate," and their play on words is saying they're providing this value that organizations can really take advantage of.
What I will say is that, as we start to look at what does it mean, and I know you wanna talk a little bit about the future for security at GitLab. What I would say is that, if you end up becoming a GitLab customer, you're always working towards that Ultimate. Because you might land with Premium, you're focused on source code management, CI/CD, and then you realize there's requirements you have to meet w is that visibility and those controls which Ultimate offers.
So let's talk about that. You know, and maybe almost like a day in the life, because historically, you know, you might sell a seat to a developer or maybe a business ops folks person. Are you now selling them to who, I use the word Ultimate, but who's, like, the Ultimate landing spot then?
We did a good job with that word, right?
It's great word.
You can't, you can't talk about GitLab without saying Ultimate.
Yeah, without saying Ultimate.
So it depends on the industry they're in. If it's technology, if it's fintech, we have a tendency to see the CTO or CISO be the one who starts that conversation. And the reason why they're asking that is they say: "Hey, look, we have to meet all these requirements. I've got 100 tools," and that's not a joke. Like, I actually talked to a customer who had 120 tools that made up their ability to try to deliver software.
They're like: "I have no clue where it is. Things are broken. I need to be able to talk to auditors." And so that ends up pulling them towards that. Now, as we talk about, like, who inside that organization gets the most excited it's been very interesting. GitLab used to be a very bottoms-up opportunity. It landed with a single team. That was usually a developer. The developer convinced the other developers in the team and the EM to start using it, engineering manager, and it's actually shifted to almost a top-down.
W hat that customer did, and it could be the T-Mobile use case on the website, where they ship software every seven minutes now using GitLab. I’ve heard you’ve all heard the UBS story a bunch, so we won’t play with that, but that's another great example. And they say, "I need to do that, too." And I think that's where our customers are great. I see our customers as partners. They see themselves as partners. Everyone says that. Everywhere I've worked, we say that, but this is actually truly the case. Some of our customers do become the best salesperson, 'cause they'll say, "Hey, I was talking to my coworker or friend, a confidant, whoever it is, that works at this other company who has a similar problem that I had." And he's like, "Let me introduce you to GitLab.
And so I think that's, that's where it is.
Top-down, bottom-up now. It's the SecOps piece that may be driving a big part of that.
That's driving a lot of the top-down for sure.
I nteresting. We could sit here and talk all day on just this, but for the sake of time. Let's move on to some of the generative AI capabilities. I guess from a high level from a productivity perspective, do you see a difference between, we've been talking about AI forever. I know the answer partially, so it's leading the witness, but the real difference between AI and generative AI and what that means from a GitLab perspective.
W e're primarily using generative AI at GitLab. What I would say is, instead of really talking about the differences between those, I think what would be great is to talk about how GitLab is doing it differently. Because everyone's saying they're including AI, and that, that's true. I, I was just at KubeCon last week, it's big, ops conference, and you couldn't walk five people without someone telling you they had AI in their product.
So GitLab, even though we're source available, we talked about free, we're an open core company, we're very transparent. We end up in a situation where these are heavily regulated customers. These are people who have intellectual property they can't have go public, and they come to GitLab to help them secure that. And so when we were looking at AI, we had to ask ourselves: What would make GitLab the best usage of AI and do it the GitLab way? And so we, we defined three tenets. The first was apply AI to the entire software development life cycle. We recently just did our DevSecOps survey. If you've not read them, go read them. They're amazing. They're full of lots of great detail. We had over 5,000 respondents.
What we found out through that survey is that only 25% of the time is spent writing software. If that's the case, then if you just accelerate that one part, the developer part, you're not really helping your organization get better. The second was we want to be privacy and transparency first. What that means is that from a privacy standpoint, your code stays your code. We don't use it to train or fine-tune our models. The transparency part is that you could literally go to GitLab documentation, it's docs.gitlab.com, and you'll see every model we're using and how it was trained. And so that has allowed customers to feel more comfortable adopting it. GitLab is trusted by more than 50% of the Fortune 100 to secure their intellectual property. That's an outstanding number, right?
You talked about security started essentially four years ago in the product, and now we're the trusted provider.
Then the final one was, we really wanted to focus on best-in-class AI, and what that means is picking the right model for the right use case. And so, we currently have between 12 and 16 models we use to power GitLab for our AI features. And so if you combine those all together, that's how GitLab is doing AI differently, and it, it's really helping people adopt GitLab Duo, which is our suite of AI-powered DevSecOps workflows.
Today, it also manifests itself in Code Suggestions as an add-on module. Do you see multiple add-on features in the future that are kind of AI based? H ow do you think, again, about what makes it into the base, an Ultimate tier versus an add-on?
That's a great question. So, just an update for everyone, if no one saw the press release last week, we actually announced Code Suggestions will be GA in December. Very excited about that. We've had a lot of customers turn it on and use it and give us feedback. And also, last week, we announced that this week, with on the 16th, which I think is Thursday, I know, honestly, I've been all over the world. I can't remember what day of the week it is. The Duo Chat functionality will go into beta, which means it'll be available to a lot more customers and be monetizable.
Is that an additional add-on?
It's part of the existing add-on we announced which has an introductory price of the $9.
But it'll also be available to customers in Ultimate. So to answer your question, we're looking at where the AI feature fits and who it benefits, and that's driving where it goes in the product. Today we have nothing that's gonna be in the Free tier related to AI. But things like our vulnerability summary, which I mentioned a couple of minutes ago, a customer shared how that allowed them to drop some of their training material.
That feature will explain the vulnerability in natural language, give an example of the code being exploited, give an example of that code being fixed in the developer's programming language. And so that feature is gonna go in Ultimate 'cause it's built on top of GitLab Ultimate. And so that's how we're looking at it. It's like, where does the feature fit, what's it built on top of, and what's gonna give our customers the best boost?
So then, it's a great sort of like dichotomy of how you think about adding it. Do you see a world in the future where there might be two, three, four different add-ons that are, you know, sort of very specific to, you know, a use case?
Yeah, I mean, it's a great question. Some of that's probably forward-facing stuff that I can't talk about in the quiet period here. But what I would say is that we're finding ways to monetize the product in a way that is best, for our customers.
That includes, we shared with our customers, we have a Plan add-on that's available this week, that's helping them bring in the non-technical people into GitLab Ultimate. And so we're always looking for ways to do that. We realize just Premium, just Ultimate, can be very restricting.
I'm gonna ask the competitive question, then we'll open it up for the group here. You know, it's the question that I get all the time. Y ou and I were talking ahead of time. W e wanted to focus on your functionality versus the competition. But let's talk about competition.
We absolutely can talk about competition.
Well, how do you think about, 'Cause to me, I see competition with Free. I see competition with GitHub, and I see competition with, you know, fragmented providers. Is that kinda how you think about it?
I'm less worried about the free part. M aybe a couple of years ago, Free had a lot more value and functionality in it. I'm less worried about that. I can be honest, like, we've done a really good job making it a free-to-paid conversion play. We're seeing the impact of that, even through the recession here and, the pandemic, like, we still were growing it's 'cause people were converting.
We do see GitHub, obviously. There's really two main people in this race for DevOps, and that's the two of us. We see them as a developer platform, and that's using their own words. They, they've shared they wanna focus very much on the developer, the developer only, and that developer experience. We then separate ourselves as that enterprise platform that brings in everyone, and so that's kind of how we see that.
But to the last part of your question, a lot of times, I see the point solution as the competitor and not so much another platform, and that's because we're going into these companies of fragmented tool chains, and they have, again, sometimes over 100 tools that they can get rid of using GitLab, and it's helping them understand that GitLab's just as good or better for that use case. And so Plan is a great example of that. This is the year that we're seeing customers migrate off of Jira in favor of GitLab Plan.
That's not just small startups or little companies. This is large enterprises who are saying, "There's a better way to do plan, and we're gonna use GitLab.
Interesting. So when you see, I guess, the double-clicking on the GitHub piece, the functionality is quite different, right? Maybe explain a little bit where 'cause even just from an Ultimate tier perspective, just, you know, level set kind of how when you, when you see the competition with GitHub, why do you win then?
Yeah. So it's purely on the DevSecOps platform component. What we're winning on is that GitLab provides a secure SCM. GitLab provides scalability that's not always there with competition, and because we're supporting everyone in the process, you get a much better visibility. The one thing that's been very surprising to me is that customers adopt our Value Stream Analytics day one. That's their ability to look at their entire tool chain and see where their bottlenecks are, and that's giving them visibility. And so if you are using GitHub, those things are add-ons you buy through their marketplace.
GitLab is included from day one. And so, for us, it's, I need better CI/CD, I need scalable CI/CD, I need secure source code management, and I need those security scanners, the compliance controls, all just built in. The one thing that I've been very impressed by with our customers is they're very vocal with us. So they may tell you something different, Matt. They may go, "Well, it's really this other thing that I made a decision on," but really, if you compare the two, it's a developer platform versus a DevSecOps platform.
If you look at the number of paid seats, we don't know exactly what you guys are or even what GitHub is, but it's still... We're scratching the surface on the total number of seats out there, right?
O ne of the things that makes me excited about GitLab's future is that this is still very much a disjointed tool world, and there's a lot of upside for that. And so GitLab has had a lot of success. In the four years I've been here, almost four and a half years, a bunch before I was here, too, and it's because people need to transform how they deliver software. During the IPO, Sid said, "Everyone's gonna have to become a software company to stay competitive." I think everyone's gonna have to become a security-aware and AI-aware company to stay competitive.
You could do that by buying 100 tools, or you could buy the leading platform. The way I like to tell everyone, and you all may have seen this as well, but don't take my word for how great GitLab's DevSecOps platform is. Look at the Gartner Magic Quadrant for DevOps platforms. We were a leader. Look at the Forrester Integrated Software Development Platform, where we were the only leader. And so that is our customers telling you through that survey, it's the industry telling us that there's a better way to deliver software and it's very much a lot left to go.
I'm gonna pause here. Are there questions for the group out here? Go ahead. I t's coming right up here.
Question for you on the planning opportunity. When you look at the per-seat price, Jira tends to be less expensive for the non-technical individuals, which, at least when we talk to some customers, seems to be an obstacle to adoption. How do you think about that as a way to unlock potential competitive wins?
That's a great question. So when we've been looking at how to position GitLab's plan against something like Jira, we have to look at it from a feature-by-feature standpoint, and I think a great example of that is the cheaper Jira license you're mentioning is really focused just on team planning. GitLab has OKR functionality, business objective tracking. That, you'd have to go buy Jira Align or another tool on top of that. And so when you start looking at the organization-wide visibility, GitLab, even in Ultimate, becomes a very good selling point for customers. Now, I mentioned a second ago that we announced the plan add-on. That's also gonna help, to your question. As that becomes available this month, it's gonna give customers an option to bring in, say, their program managers or people who they don't wanna buy a full Ultimate license for.
Thanks for the question.
That's a great question.
Yeah, any, any others out here?
Yeah, someone else raised their hand. I don't remember who that was.
Okay. Well, we'll keep going here. We only have a couple minutes left anyway. The one thing that i t always strikes me that observability should be hand in glove with DevSecOps as well. How do you think about that opportunity as well, right? If you're helping to build, deploy, manage, shouldn't you be observing it as well?
H ow I look at it is that if you're gonna be successful delivering software, you need to have the loop closed which is observability, analytics, and so forth. And so, not much of a secret, we acquired Opstrace to bring observability into GitLab. We've now shipped both error tracking and tracing inside of GitLab. We also looked at it as that that's part of getting the feedback, and we need to be able to close that loop with actual user feedback. And our customers asked us for a way to essentially measure that. Their proposition was, "If we build, secure, and deploy the application using GitLab, should we not be able to get the analytics of that application back into GitLab?
W e built a unified analytics stack that observability or Value Stream Analytics , Product Analytics are on top of, and it's allowed us to then continue to expand what we can do. But absolutely, I think you have to close the loop with that information, whether it's incident tracking with observability or user feedback. If not, you don't know what to build right in your next release.
T here's a lot of things to be excited about. It's when I think about the opportunity for GitLab, I think of multiple drivers. I think of free- to- play, I think of Ultimate, I think Code Suggestion, I think Dedicated. Well, if I ask you, what is the single or just underpenetrated market opportunity, what's the single biggest thing that you're excited about?
It's that last one. So as a former engineering leader and seeing what's needed to securely deliver software, it is truly an untapped market. We put it at $40 billion. If you add up us and, and GitHub, we're not to $40 billion.
So there's a lot of untapped potential there, and I think it gets unlocked with questions like: How do you bring in Plan better? How do you bring in observability better? And I think that's gonna allow GitLab to continue to grow. That's why I'm excited about the future of the company, and I think that over the course of however long, you're gonna see us mature those areas and bring in more into GitLab.
Excellent. Well, David, from all of us at RBC, thank you for your time. And we're excited about the future product roadmap here at GitLab, so thank you, man.
Yeah. Thank you very much, and thank you all.