Okay, great. Oh, this was not for the list you're maintaining, but.
Let's go.
Yeah, thanks, Ernie Garcia, co-founder and CEO of Carvana, for being here. My name is Rajat Gupta, member of the Automotive Equity Research team at JPMorgan. Thanks, everyone, for joining, and thanks, Ernie, again, for being here.
Yeah, of course. Excited to be here.
Yeah, let's get right into it. You provided some fairly explicit long-term targets: 3 million units in 5 to 10 years, 20%-40% volume CAGR, 13.5% EBITDA margins. You provided a very helpful visualization of how you get there, 90-80 weekly production additions. You have 23 IRCs today. Perhaps in five years, that average is like 40. If we take the 180 per week addition, that's like adding 1% capacity every week to each IRC, if I did the math right. Doesn't seem like a crazy number. My question really is, if we're having this conversation a year from now, and hopefully you'll come, and the number is, say, like 40 or 50 a week instead of 180, what would have gone wrong?
Yeah. Okay, cool. I guess let me start with, I think the way that we've tried to think about these targets is, I think 10 years is approximately doing what we're doing today for the next 10 years. Doing what we're doing today is defined in absolute growth terms, not defined by a constant growth rate. That feels operationally very achievable. I think you have demand questions, which we can hit next if we want to hit demand questions. I think trying to, in five years, would be pretty unique. Very few companies at our scale compound at 40% growth rate, which would be the average growth rate for five years, once achieving this scale. I acknowledge that that is, I think, an ambitious target when viewed against history. I think it's also pretty attainable when you break it down operationally.
The number that Rajat referenced was, we've been growing for the last 12 months. We've been growing our production capacity by about 80 units per week on average over that period. During that period, we averaged 23 reconditioning facilities. By the end of this year, we expect to have around 35. By the time we've converted all of our mega sites, we expect to have around 60. If you do the math on what that implies for the last 12 months, if we've been increasing production by 80 units per week and we've got 23 facilities, that's approximately 4 units of production growth per week per facility. At the end of this year, we'll be at 35. At that same rate, you'd be at 140 cars. If you get to 60 when we open all the mega sites, you'd be at 240.
In order to get to 3 million units in five years, you'd have to average 180 over that period. I think the fundamental building blocks are very clearly there. There's a lot of execution that separates you from where we are today to growing at that rate over a sustained period of time. I think it'll be hard, but it's certainly not impossible. We think that given that we're about 1% of the market today, and even that seemingly very ambitious goal would get us to 7.5% market share, which relative to most retail verticals would be pretty low penetration, we think that it's very achievable from a global top-down perspective. I think the approximate story for demand over a long period of time has been that it has generally been there. We think that the offering that we make to customers is very desirable.
It's a simple experience. It's economically a very high-quality deal, and they feel confident afterwards and tell their friends. Generally speaking, I think we've been getting continually better at providing the experience that we provide to customers. I think the economics that we've given to customers have been pretty consistent across a long period of time. I think from 2013 to 2021, that consistent experience and offering led to very fast growth. In 2022 and 2023, we had a massive strategic shift where we changed our focus and had to get to cash flow positive very quickly. In 2024 and so far in 2025, we've experienced, again, kind of very consistent growth, which we think is driven by that core offering. We think demand should be there.
We could also kind of defend that by looking at our cohorts and looking at our oldest markets and the market share that we've got there, which is much stronger than our average market share. They are still growing quickly. I think we look at all that and we think that it is up to us to execute well, but there is no reason we cannot do it. I think overall, we are pretty excited.
Are there any one or two things you would point out about Atlanta or some of the older cohorts that have performed like the metrics you mentioned on the earnings call? Atlanta growth was similar to the overall company from 1Q 2022 to 1Q 2025. Are there one or two areas or things you could point to that's allowed Atlanta to continue to grow at that pace? Is it just the simple word of mouth? Is it inventory pools, conversion? Again, just help us get more comfortable with the demand side.
I mean, I think generally speaking, I think the entire country benefits from changes to experience quality. Things like faster delivery, more inventory pools, broader selection. Generally speaking, I think that's shared across the country. I think generally speaking, the growth that we've seen across the country is pretty similar. Of course, there's variation across time, but the last time we gave detailed information on our cohorts was in, I believe, Q4 of 2021. At that time, Atlanta had a 3.5% market share, which would equate to approximately 1.4 million units per year at a 40 million unit per year average. It was still growing quickly. 95% of the markets that we had opened, which at that time was a couple hundred markets, were growing faster than Atlanta at the same point in its life.
I think the right mental model for the way that we've grown over time is that we've continually made the offering better. We've continually made enhancements to the website and grown inventory, and that increases conversion. Those are shared across markets. Another very important kind of driver of growth is that customers have to experience it and tell their friends that it's good. That component happens locally. I think that it's clear in the data that that's an important component, which I think is medium and long-term good. Because if growth were driven just by your national offering, then what you would see is you'd see as you open new markets, you would see them just jump to your market share that you enjoy in your older markets, and they would instantly be at that level.
You have to kind of earn your increased market share beyond that. What we've continually seen is when you open a market, it ramps up monotonically in time, but it does not just jump to the higher market share. I think that is because consumers, in their mind, they are confronting a trade-off where they know that buying a car can be an experience that they are nervous about, and they know that they are anxious about getting a bad deal, and they want to go try something new. They also are kind of like, "I buy a car once every six or seven years. I do not want to do anything dumb." Sometimes the default can be do what everyone else has always done. I think consumers sit there and have that battle in their mind when they are evaluating whether or not to go to Carvana.
I think as more and more of their friends and family buy a car from Carvana, it gets easier to say, "Okay, that's the way that I do it." I think that plays out over time. The time scales are actually long, given that the transaction cycle for buying a car is a five to seven-year transaction cycle, which is like another—I am going to just keep going on random tangents here. I think that is actually good. I think that people in e-commerce have developed a heuristic that I think is smart in many ways that faster transaction cycles are better. I do think that conditional on being able to observe the growth rate, I think a slower transaction cycle is actually better because you have more latent demand that is on the come.
If you have very, very rapid transaction cycles, you very quickly move to kind of the level of sales that your offering supports. If you have a very long transaction cycle, it takes time for those transactions to occur. You are always kind of behind the curve of consumer demand, which I think is a net positive for us.
No, I think that's a good way to put it. It's very clear. The other side of this equation, unlike a new car industry, you cannot manufacture used cars. To sell more used cars, you need to source more used cars. How do you help solve the supply side of this equation? I mean, 3 million, again, it's a small number relative to the size of the market, but it's still a big number. No other retailer has done that under one hood. Is supply a challenge? Should we be worried about that, either from consumers or from auction? Is it more localized? How do you solve for supply? That should not be a constraint to hit these targets.
Let me start with something that hopefully provides a bit of credibility to then my very conceptual and very convenient answer that I'll ultimately provide. I think the credibility is if we go back in time six years and we listen to all the conference calls or listen to any of these conferences, whatever else, the question back then, we were approximately one-sixth of our current scale. All of the questions were, "Will you be able to sell all your receivables at greater scale? Will you be able to source cars at similar margins?" Those were all of the questions. I think they were all good questions.
I think they all stem from this idea that if you want to take more of the market, like the incremental customer is more expensive in some way, maybe it's marketing dollars or acquiring the car or whatever it is, they're going to be more expensive in some way. I think a mental model that has been more predictive so far of what's happened in our history and we expect to be predictive in the future is there's already 40 million transactions that are occurring. Those 40 million transactions are ultimately just customers trading cars with other customers through the mechanism of many different dealers that all have very similar cost structures and profit goals. You have a very stable market that sort of provides buttresses for what your economics are going to be because many of these dealers cannot absorb losses.
As we grow, we may go to auction one day, let's say that's where we're going to buy our incremental car, and raise our hand on a car that another dealer expects to ultimately be able to sell to a customer. We may kind of argue about that car in the form of bids. To the extent we do that intelligently, it's not just that one car that gets bid up, it's all the cars that get bid up, right? While we're arguing about that, that means that everyone's shared input cost just went up by the same amount. Everyone has the same underlying model. Most dealers don't have the ability to take losses. It is very likely that will be passed through.
To the extent that we're right and we do sell the car, next month, that other dealer that we were arguing about the car with is not going to raise their hand for that because they do not expect to sell it again next month. I think when you're in a mature market with mature unit economics and a bunch of players that have shared economics, if you can displace them in demand, your best expectation is that your economics for the incremental unit are the same as your economics in the previous unit. I think that that has played out over time. Whether we're talking about our ability to go sell financed receivables, today, consumers are buying 40 million cars. Many of them are financing those cars, and that is being financed through various channels.
There are people that want to own receivables at that scale. To the extent we go take market share, the natural holders of financed receivables are going to come find us, or we're going to find them, but we're going to find each other one way or another. I think the same is true of cars. To me, I think as long as we execute the way that we would like to, I'm sure there will be little bumps and we'll tell a bunch of stories along the way. Generally speaking, we should be displacing similar businesses that have similar economics, and we would expect our economics to be unimpacted, at least in a meaningful way.
Understood. Maybe on that supply side, how important is it getting access to, or the first look to the lease returns as part of that 3 million target? I think on the earnings call, you mentioned you expect the business mix or inventory mix to look pretty similar even at that 3 million. That would imply lease cars probably is not that important, but just curious how you think about the off-lease supply supporting that 3 million target. Maybe you want to touch on the franchise acquisition if that's going to help eventually.
Yeah. What I would say is I think as a general matter, we want to have access to every channel of supply that exists, and we want to try to structure our access to those supply channels in the most efficient ways we possibly can. I think that's part of what our marketplace offering has been. I think that we'll continue to seek to find ways to take what we think is an advantaged retail channel paired with an advantaged wholesale channel to make us a great place for natural disposal of cars to dispose of cars, whether that's consumers or rental car companies or off-lease or whatever else. I think making sure that our system is as efficient as possible disposing of cars makes us the best possible buyer, and that gives us access to all that volume.
I think there's only been one window in time that I can think of or I'm aware of where off-lease volume was differentiated in its quality, and that was post-COVID when there was very rapid car price appreciation, and therefore those cars that were coming back off-lease were extraordinarily valuable. I think that that was a boon to franchise dealers, and that was hard on independent dealers. For the previous 25 years plus where I've been paying attention to automotive retail, generally speaking, those cars were still making it through auction. You still had access to them, and it was more a function of just what was the quality of your retail disposition channel. That kind of dictated how you get those cars. I think we want access.
We'll try to build creative solutions to those problems and make sure that we have access to lots of cars. I don't think that there's anything particularly unique about lease. On the franchise dealer question, we'll just keep kind of giving you the stiff arm. One day, we'll definitely tell you something.
Understood. On the margin targets, it was interesting that you decided not to give a gross margin versus SG&A number, but more an EBITDA number. I'm curious internally, how do you look at your economics? Are you trying to solve for individual line items like retail GPU, finance GPU, or are you looking at retail and finance as one bucket, or are you just looking at EBITDA per unit as one bucket? Help us think if the way you're looking at these metrics has changed internally as a company.
Yeah, let me give an example. I believe that when we went public in 2017, we made one of our key metrics total GPU, right? We added it up across the transaction. To a lot of people, I think that struck them as odd, but the rationale that we provided at the time, which is the same rationale that we would stick with today, is that the consumer makes a single decision. The decision is, "I want to buy a car." When they want to buy a car, it is likely they're going to trade in a car. It is likely they're going to get financing for that car. When they make their decision, they're going to have awareness of what they think their trade-in should be worth. They're going to have awareness of what a rate should be and what a price should be.
Ultimately, what drives their decision is going to be, "Okay, my monthly payment is going up by this much, and I'm getting that car instead of that car." That is the actual decision that they're making. There are all these different line items that play a role in putting together that transaction. What matters to the customer is what their payment delta is and what the car changed, right? That is what matters. We have always viewed it as the used car system is a system that does not manufacture products. It just helps consumers trade cars with each other, and the consumer makes a single decision. The economics are sort of a single thing. What you want to do is make sure that you have access to all the different places where money could flow through.
Because if you do not, you are susceptible to your competitors marketing in ways where they make way less money in the line items that you have, and they make more money in the line items you do not have. As long as you have coverage of all the different places where money can flow, you are kind of fine and you are basically indifferent about where the money shows up. It is tremendously simplifying, I think. For example, our team has put together graphs of if you take the average of all of the public automotive franchise dealers over the last 25 years and you graph their EBITDA margin over a 25-year period, what you basically see is it averages four the entire time. In 2008, it goes down to three when you control for goodwill write-downs. It goes lower than that if you do not control for goodwill write-downs.
When you control for non-cash goodwill write-downs, it goes down to three. Basically, that's the story, right? It's been completely flat. If we zoom into if you had ChatGPT go analyze all of the public automotive retailer conference calls for the last 25 years and you said, "What are the key themes?" You'd get 100 themes. You'd get people talking about economics moving from the car to F&I and moving from here to there. You'd have gas guzzlers and Hurricane Katrina and this recovery and that recovery. You back up and you look at the EBITDA margin, it's just completely flat.
To me, I think that the biggest force—we can talk about all these little things that go on, all the line items—the biggest force is that you have an enormous mature market with mature unit economics, and you have tens of thousands of players with the exact same cost structure that do not have cash balances to absorb losses. At the end of the day, it is a pass-through of a fair return on that capital. That is basically what it is. If you just back up to that level and you say, "That is what this should be if we execute well," the only things that now move are what is our experience versus everyone else, what is our ability to get access to revenues compared to everyone else, and what are our expenses compared to everyone else.
For us to do that well, we have to be super dialed into all of the different line items. We have all kinds of interesting projects we're working on in inbound transport and outbound transport and vehicle data and pricing and finance and wholesale. All of that reduces at the end of the day to, can you do things on the revenue side better than everyone else, on the cost side better than everyone else, and deliver an experience that people like more? If you do that, you win. The equation is basically, what is the industry-level profit, which is very flat, and then what are those three deltas? To us, the more that we can kind of reduce the conversation, I think the easier it is for us to talk about important things.
Because I think all of us as people and as analytical people, we tend to look at the numbers that are moving. Whatever number is moving right now, we want to look at and understand. That's useful because you make sure that you're not missing anything. It also sometimes you can zoom in too far and you can miss the bigger picture. The bigger picture is, I think, much simpler than people think, and I think very positive.
That's helpful. You gave yourself some flexibility in those margins in your letter that at some point you want to give yourself margin flexibility to grow. Can you talk a little bit about where you think you might need to invest or add some slack to get that growth? Is it GPU? Is it SG&A? Where does price fall into that equation? Just help us through maybe rank ordering some of the things you could do that might have the highest return on growth.
Yeah. First, what I would say is I think we've been on a path for a couple of years. We've been increasing our rate of growth and improving our economics continuously. I think that from where we sit today, we have another long list of very interesting projects that we're kicking off that we think have significant fundamental gains, and we don't think our offering has changed. As a result, we would expect our best expectations that demand growth would continue in a way that is similar to the past. I think to the extent all of that is right, there are exciting versions of the future that are very possible where we don't need that flexibility.
What I also think is true, though, is that we have 1% of the market, and we're currently around twice as profitable from an EBITDA margin perspective as the average outside of us automotive retailer. The most important thing we can do by a long way is displace players in this very large, very mature market with unit economics that we understand where the three deltas were very positive. In light of being 1% and having that be the most valuable thing and just recognizing the reality of the world being a dynamic place, we want to make sure we preserve flexibility and that we communicate what are our priorities because the world makes you trade-offs sometimes. When we face those trade-offs, we will trade off within reasonable margins. We'll trade off toward growth. That's not a signal for where we are today.
That's what we think is smart to manage the business with a 5-10 year horizon.
Got it. I wanted to touch on the lending business briefly. Your gain and sale margins, the progress made over the last 18 months has been quite phenomenal. I mean, there's always been some benefits on the gap between rates and EPRs, but loss expectations are still high. What factors would you attribute to this improvement? Is it just residual pricing, more demand for the resids? Is it the new partners coming on? Help us walk through that and the sustainability.
Yeah. I mean, first of all, we're a 12-year-old business that has been rapidly increasing in scale, and many of the things that you do in finance get much better as you have more scale. Just a very simple one is your best credit score that you can build is definitely a function of the volume of data that you have. We're about to roll out our eighth credit scoring model. That's a very simple fundamental gain. Many of the things that you do in finance, you structure your loans. You have to decide for any given customer what is the minimum down payment, what is max monthly, what is max term. Those are all choices that are informed by data that you get better at over time. You have verifications processes.
The more verifications that you impose on any given customer, the lower the expected conversion is, but the better the expected performance is. That is something that gets better over time. You can just get strictly better by building better technology that allows you to do more verification faster, but you also can make trade-offs in terms of what you are doing there. All of these things are continually getting better, and there are teams that are working on all of these projects. I think we also, I am nowhere near objective on this particular point, but I think it is true. I think auto loans are a mispriced asset, and I know that is a very arrogant thing to say, and I believe that it is a mispriced asset.
I believe there's a growing body of evidence that those who buy them buy more of them disproportionately to those who buy other assets. I think that that allows them to become slightly less mispriced over time. I think that that's been a 10-year journey as well. My personal take is there remains some room in that journey. We'll see. I think the team's done an awesome job there, and I think the vertical integration of the business lends itself to providing very simple customer experiences and capturing more of the economic gain that exists across the transaction. I think that you've seen that show up pretty continually.
We had a little bit of a, if you looked at our curve going way back in time, going back since whatever the first years of data we went public in 2017, so I do not know if you provided, I cannot remember if you provided two or three years of data, but if you can go back to 2014 or 2015, you would basically see a straight line up. In 2021, you hit an early peak because there are lots of things that are going on then that are good. We came down and we are kind of back up. Generally speaking, you have seen a pretty linear line that looks a lot like our EBITDA margin. If you zoom into any of our line items, you will generally see that improving line over time.
Yeah, yeah, for sure. I have one more question on balance sheet cash flow to return to the audience. Say you get your targets in 5 years or 10 years or 15 years. You're going to be generating tens of billions of cash flow over that period. What's going to be the use of that cash?
Can't we just sit on it? We're going to be generating tens of billions of cash flow and maybe a high five for a second? What I would say is I think that's a fun and stark reality versus the types of questions that we were getting a couple of years ago, which is nice. I think it's likely true as long as we execute the way that we plan to. I think there are many places we could put that cash for now. I think we're trying to just kind of reduce net debt, which is being reduced very, very rapidly. I think that's a smart thing to do. To the extent we run out of good ideas, we'd be able to buy back stock. I think generally speaking, we've got plenty of ideas.
Would you consider just owning resids more, using the balance sheet for the finance company, things like that?
I wouldn't want to constrain us today, but we'll see. We'll try to do smart things.
Fair enough. Just want to open it up to the audience here if anyone has a question. Anyone here? No? All right. Okay. We'll give you a mic.
We got a question on.
Oh, you do? All right. There we go.
I think I can anticipate the answer, but I believe the company brought a new car dealer, Stellantis, and just kind of curious what the strategy is around.
Oh, wow. Come on. You can't anticipate the answer. Yeah, sorry. It really is. I don't mean to be evasive on that, but it is very early. We don't have much to share just yet, but we're testing and we're learning. As we have something to share, we'll share it. Yeah, forgive us for the non-answer on that one.
Right. The other kind of the follow-up on the margin question, it does seem like the company has enormous leverage potential to go beyond that 13.5% margin, which I think you've somewhat alluded to. Can you talk about, I mean, it's still going to go dramatically above that margin. I guess how do you think about the brand? For instance, if you were able to just offer strategically lower prices than the whole industry all of the time, or can you just talk about how you're thinking through what to do with the excess margin that is likely on the come?
Yeah. I think we're pretty excited about that. I mean, I think the best reduction of that is I think we said in the shoulder letter, our goal is to become the way people buy cars, right? I think to become the way, you just need to be, it needs to be a very simple conversation people can have with each other, right? It's go to Carvana. They have the car you want. The deal's fair. It's super easy. Car's great. If you don't like it, you can return it. If you say that confidently to your friends enough times, I at least don't understand why we're not selling a huge percentage of the cars over a long period of time.
I think when you start from the place of having approximately double the EBITDA margins of the rest of the industry, and you start from a place of not having fully levered into your fixed costs and clearly having more opportunities in front of you, right? I think it's always hard to know exactly how much opportunity you have in front of you, but roughly speaking, if we convert EBITDA margin to EBITDA dollars per unit, we improved by around $2,000 two years ago and around $1,000 in the last year. If you can improve that much that fast, there's more left, right? We can argue about how much more is left, but there's more left. I think that our growth, I would say there's been three periods of our company's history. There's 2013 through 2021, which is pretty straight, full-on sprint growth.
There is 2022 and 2023, which is massive strategic change, total retrenchment, focus on just getting cash flow positive as soon as you possibly can. There is 2024 and 2025, which are growth again. I think when you see that growth, that growth is happening with our consumer economic offering quality being pretty similar. The offering that we've provided the customers, paired with the speed at which they're telling each other the story of their Carvana experience, has led to the company-level growth and the cohort-level growth that we've seen. My hope and belief would be, to the extent we hold that constant, it would lead to a lot of growth in the future as well.
I think to the extent we're able to go unlock the fundamental gains that we think we can unlock and pass a better offering on to customers, that would be incremental growth power beyond the growth power that we've seen for the last 12 years. I think that that's really exciting too. I mean, I think our eyes are really big, and I think we've got a lot of work to do. There's a lot of work to do to unlock those fundamental gains. There's a ton of work to do to go actually scale the system and execute at the level that we need to. We will definitely hit bumps along the way, but we're positioned unbelievably well. We've got a business model that scales really well.
We've got a team that's been through brutal times together, which I think demonstrates the ability to solve problems as they come and also suggests a closeness that I think is also a valuable thing as you're trying to manage through heavy growth. I think we're starting in a great economic spot. I mean, to me, it's like I would have said at virtually every point in our company's life, except for maybe a day or two in 2022 or 2023, I would have said it's the best day of our life. We're back to where I think every day is basically the best day of our life. I just think we're in a good spot.
Thank you. Any other questions? I just have a couple more here. I wanted to talk a little bit about AI and how you're starting to leverage that within the company. What internal dataset do you think is more underutilized or undervalued today within the company and which could potentially become a differentiator long-term for Carvana, given how much you've collected over the last 10/ 12 years?
I think the most valuable dataset is the cross-transaction dataset. I think I would again go back to the thing we discussed earlier, which is a consumer makes a decision to buy a car, but to buy that car, their currency is likely a car that they brought and a finance receivable. They may also want a warranty. I think the benefit of being a single vertically integrated system is all of that is available via API. All that data we can just call and pull in, and we can pair it with our scheduler API. You can have a customer that can say, "I want a car for $400 a month that can be delivered on Thursday. I've got a family of four. What's good for me?" You can answer that question.
You can answer that question in a very informed way because you can access all of your previous data, so you can figure out what people are likely to want. You also can call all those services to give someone a complete answer to that question that a retailer or a finance company or a third-party software company that's used to value your trade-in would not be able to do by themselves. To me, I think there are many valuable datasets. Everything we do, we're vertically integrated. We've got all of the transaction data. We've got all the clickstream data. We've got all the finance data. We've got the repeat customer data. We've got loan performance data. I think pairing all that together into a single experience simply, I think is probably the most differentiated thing that we have.
Understood.
AI massively accentuates your ability to do that.
Right. Yeah. Right. That makes sense. Any questions? No. I had one question, a little off-topic, just culture and philosophy related. Is there an internal debate, maybe one or two debates in the company, that keeps resurfacing at the leadership level but never gets fully resolved?
I mean, I think the most recurrent debate has always been how fast should we grow. I think that generally speaking, I personally have been on the side of let's go faster. I think generally speaking, the operator's been on the side of let's make sure that we're moving at a pace that is kind of balanced and thoughtful and where we make sure that all the operations stay very tight. I think this tension is valuable because I think you find better solutions when you argue with people you respect, and then you have to push on each other's assumptions. I think pressure is valuable.
I think that one of the interesting things that I'll for sure take away from 2022 and 2023 is we did better in a way that we could not have done on our own without the world telling us that we were stupid over and over again. I think that makes you better. I think putting pressure on yourself and finding ways to internally generate that pressure is important. I think we also debate how do we do that because I think we've had two and a half years now where I think people stopped calling us as stupid, and every once in a while they call us smart. I think that that's much less helpful to productivity.
I think trying to make sure that we all internally stay as hungry as we can is, I think, another thing that we are putting energy into today and I actually think is hard. I think being successful for a long time is not normal for people. I think it is much more normal to let success go to your head and be a little less effective as a result.
Understood. I think there are no further questions. I mean, I think that's a good way to end.
Perfect.
Thanks, Ernie.
Thanks, everyone.
Thanks everyone for joining.