Q3 2025 Earnings Call Transcript

Good day, and welcome to Carvana's Third Quarter 2025 Earnings Conference Call. [Operator Instructions] Please note, this event is being recorded. I would now like to turn the conference over to Meg Kehan with Investor Relations. Please go ahead.

Thank you. Good afternoon, ladies and gentlemen, and thank you for joining us on Carvana's Third Quarter 2025 Earnings Conference Call. Please note that this call will be simultaneously webcast on the Investor Relations section of the company's corporate website at investors.carvana.com. The third quarter shareholder letter is also posted on the IR website.

Additionally, we posted a set of supplemental financial tables for Q3, which can be found on the Events and Presentations page of our IR website. Joining me on the call today are Ernie Garcia, Chief Executive Officer; and Mark Jenkins, Chief Financial Officer. Before we start, I would like to remind you that the following discussion contains forward-looking statements within the meaning of the federal securities laws, including, but not limited to, Carvana's market opportunities and future financial results that involve risks and uncertainties that may cause actual results to differ materially from those discussed here.

A detailed discussion of the material factors that cause actual results to differ from forward-looking statements can be found in the Risk Factors section of Carvana's most recent Form 10-K and Forms 10-Q. The forward-looking statements and risks in this conference call are based on current expectations as of today, and Carvana assumes no obligation to update or revise them, whether as a result of new developments or otherwise. Our commentary today will include non-GAAP financial metrics. Unless otherwise specified, all references to GPU and SG&A will be to non-GAAP metrics, and all references to EBITDA will be to adjusted EBITDA.

Reconciliations between GAAP and non-GAAP metrics for our reported results can be found in our shareholder letter issued today, a copy of which can be found on our IR website. And with that said, I'd like to turn the call over to Ernie Garcia. Ernie?

Thanks, Meg, and thanks, everyone, for joining the call. The third quarter was another incredible quarter for Carvana. We remain the most profitable and fastest-growing automotive retailer. These data points are exciting in isolation.

Achieving them simultaneously is rare and points to an exceptional future. Achieving them by the margins we have been recently, profit margins more than 2x the industry average and growth over 40% when other public retailers are approximately flat points to something that is structurally different, something that is capable of achieving our ambitious mission of changing the way people buy and sell cars. That is exactly what we believe we are capable of, exactly what we are focused on making happen and exactly what the data is telling us we are marching toward every quarter. Q3 was another large step on the path to achieving our current goal of selling 3 million cars at a 13.5% adjusted EBITDA margin in the next 5 to 10 years.

We're getting better as we get bigger, aided by the feedback inherent in our business, the benefits of scale and our continued pursuit of fundamental gains as well as the addition of foundational capabilities. The positive feedback flywheel is spinning. The data that powers our decision-making throughout the business is growing exponentially and allowing us to iteratively improve the models powering every decision across the business. Our sales growth allows us to keep growing inventory economically, constantly broadening customer selection.

Year-over-year, our inventory turn time is approximately flat, yet our customers have nearly 50% more cars to choose from. And the benefits of scale are also allowing us to make investments that magnify these advantages. Over the last 18 months, we've added reconditioning capacity to 15 ADESA locations, allowing us to position inventory closer to our customers, reducing customer delivery time by a day in the last 5 quarters. We've developed our digital auction capability, ADESA Clear, delivering a best-in-class digital auction experience to our wholesale customers, allowing us to add wholesale capabilities to 12 of our inspection centers and counting.

Having the strongest retail and wholesale channels to sell vehicles makes us a systematically better buyer of all cars from our customers and our partners. We've also been working on making dramatic improvements to delivery capability that will show up over time. We are currently using Phoenix as a test market to optimize our finance verifications, registration processes, vehicle staging, delivery scheduling systems and staffing models for speed. As a result, 40% of customers in Phoenix are now getting same or next-day delivery compared to 10% that get same or next-day delivery nationwide.

On any given day, customers in Phoenix have about 2,500 cars available to be delivered that same day. That's worth pausing on and taking time to think through the implications. Thousands of vehicles that can be purchased in minutes and delivered in hours is a highly desirable and extremely difficult to replicate capability. Like all good things, this will take some time to optimize the rollout across the country, but it is coming.

Another set of statistics that demonstrate meaningful progress are that today, more than 30% of retail customers now complete the entire process without any interaction with the customer advocate until their delivery or pickup appointment. For customers selling their car to us, this number is more than 60%. To make this possible, our business must be vertically integrated, data must be well organized and immediately accessible. Decisions have to be deterministic and automated.

Workflows have to be concretely defined inside of systems and all that has to be wrapped in intuitive interfaces that make customers feel confident. It's hard. Our team is doing a great job, has detailed plans to keep making it better and is nowhere near satisfied. Looking forward, we continue to see opportunities for fundamental gains in every line item.

Opportunities that will make our customer experiences simpler and more fun, will make our costs lower and will make our business more efficient. Our plan is to unlock these opportunities with the same discipline that has driven our success so far. Something that has always been true in the past remains true today and that we suspect will be true for a long time is that prioritizing our opportunities is the hardest part of making significant progress quickly. With constantly evolving technology, constantly evolving customer preferences and expectations and an ambitious group of thoughtful people, new opportunities emerge faster than we are able to take advantage of the ones we previously saw.

With AI, this is more true today than it has ever been. The future is bright. Selling 3 million cars per year with 13.5% adjusted EBITDA margin in 5 to 10 years is very achievable. There's a lot left to do, and there's an excited team ready to do it.

We will continue to aggressively pursue rapid progress, and we aren't tired. The march continues. Mark?

Thank you, Ernie, and thank you all for joining us today. The third quarter was another very strong quarter for Carvana that was driven by our team's continued focus on identifying further fundamental gains and operating efficiencies and developing foundational capabilities while also pursuing growth. We set new records for retail units sold, revenue, adjusted EBITDA and GAAP operating income. And for the first time, our annual revenue run rate exceeded $20 million, a significant milestone pointing toward the long-term scale of our business.

Moving to our third quarter results. Unless otherwise noted, all comparisons will be on a year-over-year basis. Retail units sold totaled 155,941 in Q3, an increase of 44% and a new company record. Revenue was $5.647 billion, an increase of 55% and also a new company record.

Revenue growth exceeded retail units sold growth, primarily due to higher average selling prices and traditional gross revenue treatment for certain vehicles acquired from a large retail marketplace partner. Consistent with past quarters, our growth in the third quarter was driven by our 3 long-term drivers of growth: a continuously improving customer offering, increasing understanding, awareness and trust and increasing inventory selection and other benefits of scale. Our strong profitability results in Q3 were again driven by our team's focus on driving fundamental gains and operating efficiencies as well as levering our overhead expenses.

Non-GAAP retail GPU decreased by $77, primarily driven by higher retail depreciation rates. Non-GAAP wholesale GPU decreased by $168, primarily driven by higher wholesale depreciation rates and retail units sold growth outpacing ADESA marketplace growth. Non-GAAP other GPU increased by $63. This change was primarily driven by improvements in cost of funds and higher finance and VSC attach rates, partially offset by higher-than-normalized loan sales relative to originations in Q3 2024. Looking ahead to Q4, we expect sequential changes in retail GPU, wholesale GPU and other GPU in a similar range to last year, with the latter primarily reflecting sharing fundamental gains with customers through lower interest rates.

In October, we expanded on several existing loan sale partnerships with agreements for the sale of up to $14 billion of future loan principal. First, we upsized and extended our Ally agreement for up to $6 billion of loan purchases through October 2027, an increase from $4 billion through April 2026. Second, we entered into a new loan purchase agreement with a loan sale partner for up to $4 billion of loan purchases through October 2027. Third, we entered into an additional loan purchase agreement with another loan sale partner for up to $4 billion of loan purchases through December 2027. The latter 2 agreements formalize existing relationships and establish defined expectations for sale volume and sales procedures throughout the agreement period, highlighting the significant fundamental strength of our vertically integrated finance platform.

Q3 was another strong quarter for demonstrating the power of our model to lever SG&A expenses. Our 44% growth in retail units sold led to a $319 reduction in non-GAAP SG&A expense per retail unit sold. Carvana operations portion of SG&A expense decreased by $96 per retail unit sold, primarily driven by our operational efficiency initiatives. We continue to expect Carvana operations expense per retail unit sold to decrease over time as we deliver fundamental gains and operating efficiency.

The overhead portion of SG&A decreased by $314 per retail unit sold, driven by continued leverage of our overhead expenses with greater retail units sold. Advertising expense increased by $139 per retail unit sold as we continue to take advantage of opportunities to invest in building awareness, understanding and trust of our customer offering. We expect advertising expense in Q4 to be similar to or slightly higher than Q3. We continue to see opportunities for significant SG&A expense leverage over time and as we scale, driven by both continued improvements in operational expenses as well as leverage in the fixed components of our cost structure.

Net income was $263 million in Q3, an increase of $115 million. Net income margin was 4.7%, an increase from 4%. GAAP operating income was $552 million, an increase of $215 million and a new company record. GAAP operating margin was 9.8% and an increase from 9.2%.

Adjusted EBITDA was $637 million, an increase of $208 million and a new company record. Adjusted EBITDA margin was 11.3%, a decrease from 11.7%. As previously discussed, our adjusted EBITDA is very high quality compared to many rapidly growing companies due to our relatively low noncash expenses, which we'll continue to lever with scale. We converted approximately 80% -- 87% of adjusted EBITDA into GAAP operating income, an increase from 79% last year.

As previously noted, we currently carry many expenses that support retail unit sales capacity of over 1 million units and expect our GAAP operating income to grow faster than adjusted EBITDA over time. In the third quarter, we took additional steps to further strengthen our balance sheet with a continued goal to drive toward investment-grade credit ratios. In Q3, we retired the remaining $559 million of our 2028 senior secured notes, primarily through proceeds from $539 million of equity issuance through our ATM program. Following quarter end, we also retired $98 million of 2025 senior unsecured notes due October 2025, bringing our total quantum of corporate debt retired in 2024 and 2025 to $1.2 billion.

With more than $2.1 billion of cash on the balance sheet, our net debt to trailing 12-month adjusted EBITDA ratio is now down to just 1.5x, our strongest financial position ever. Our results through Q3 position us well for a strong finish to 2025. Looking toward the fourth quarter, we expect the following as long as the environment remains stable. Retail units sold above 150,000, and adjusted EBITDA at or above the high end of our previously communicated range of $2 billion to $2.2 billion for the full year 2025. In conclusion, Q3 marked another outstanding quarter for Carvana. We remain very excited about progressing toward our long-term phase of driving profitable growth and pursuing our goals of becoming the largest and most profitable auto retailer and buying and selling millions of cars.

Thanks for your attention. We'll now take questions.

[Operator Instructions] The first question comes from Sharon Zackfia with William Blair.

I guess the topic du jour is kind of subprime loans. And I know we can see a lot of your subprime loan performance and your prime loan performance through various different vehicles. But can you talk about kind of the health of the portfolio, whether you foresee needing to take incremental reserves there at all? And then separately, the timing of the formalization of these new third-party agreements, kind of what brought on that timing?

Sure. I can take that one. So the very simple answer on loan performance is our 2024 and 2025 loan originations are performing extremely well, both in an absolute sense and relative to industry comparables. I think some of the chatter out there about loan performance more broadly, we think has a lot to do with the 2022 and 2023 industry-wide cohorts, which did underperform initial expectations.

I think most of the industry ourselves included tightened credit in late 2023. We certainly did, and we've maintained that tightness here through where we are today in 2025. As a result, our loans are performing strongly. I think the best evidence of that is twofold. One, just the stability and strength in our other GPU. And then secondly, the outside validation of having Ally upsized from $4 billion to $6 billion based on the performance trends that they're seeing and additionally, the addition of these 2 new purchase agreements, I think, are great validation of the strength that we're seeing given all the fundamental gains that we've made in the program over time.

So I'll start there. In terms of the specific timing of these agreements, I think it's really just a continuation and a maturation. These 2 large agreements are with existing partners who we've been selling loans to over the preceding periods. Those previous loan sales have been more on a one-off basis.

And so what these agreements do is effectively formalize the sales procedures and set volume expectations with those partners to essentially make more programmatic, the more one-off sales that we've already been doing. And so really, I think there -- the main point there is it's a maturation and a continuation of something we've already been doing, but now just in a more structured and formal way.

The next question comes from Marvin Fong with BTIG.

I hopped on just about 5 minutes ago, so I apologize if I missed this. But the OpEx, the operating expense per unit, I noticed ticked up sequentially, although it was down year-over-year. And I just wanted to understand, is that a better way to measure that metric? And can you just kind of talk about your future opportunities to continue to kind of like drive down your operations cost per unit, that would be great.

Sure. Yes. So I think it's useful to maybe break total operating expenses up into a few different categories. I think we -- let me break them into operations expense, overhead expense and advertising expense.

I think we start with advertising expense. We are starting to invest in the multiple levers of our 3-part growth driving plan, which includes continued improvements in product offering, building understanding, awareness and trust and growing selection and driving other benefits of scale. Advertising hits the second of those. And so we have been investing in advertising as part of that longer-term 3-pillar growth plan.

I think looking at overhead expenses, I do think we saw nice leverage year-over-year and some leverage quarter-over-quarter there. Within overhead, I think that's grown much, much more slowly than retail units. I think in recent periods, there's been some lumpy expenses there that we think are more transitory in nature that's caused it to be a little bit higher than it otherwise would be. But overall, we're seeing very strong performance there with nice leverage on a year-over-year basis and then to an extent on a sequential basis.

Last, on operations expense. I think another place we've seen very strong gains on a year-over-year basis. That can bump around quarter-to-quarter, just depending on the presence of onetime items or other nonrecurring expenses. So it stepped up a little bit sequentially.

But overall there, the trend is down, and we expect to drive that down further over time as we continue to drive fundamental gains in operating efficiency.

Great. And if I could maybe sneak in one more. Just skimming the real-time transcript here. So I believe you said retail GPU will be similar to last year.

And I was just kind of wondering what's sort of underpinning those dynamics? So are your recon and logistics efficiencies sort of being offset by the macro environment and what's going on with depreciation rates? Or just kind of maybe double-click on what's kind of behind the year-over-year flattish retail GPU guidance?

Sure. So I think let me first just hit Q4 seasonality. I think most listeners have a good sense of industry seasonality in Q4, but typically, it involves higher depreciation rates, both in the retail and wholesale markets. And then it's typically industry-wide, the lower period for demand, whereas other quarters have more strong depreciation rates in both retail and wholesale and stronger seasonal demand.

As it relates to retail GPU, I think what we called out is we are seeing for Q4 sort of sequential change off of Q3 that is in a similar range to what we saw last year, driven by seasonality. I think that looking at year-over-year trends, I would say that the number 1 thing I would say we feel like we've seen is I think Q2 was a bit of a strong depreciation quarter in the retail market. We attribute that to some effects from the late March auto tariff announcements. I think on the contrary, Q3 was a bit of a softer depreciation quarter on a year-over-year basis.

I think we would attribute that almost an offset to the Q2 strength. And so I think some of those depreciation dynamics would be the -- I guess, the one thing I would call out and also the thing that I mentioned in my prepared remarks about Q3 retail GPU.

The next question comes from Rajat Gupta with JPMorgan.

Just on the fourth quarter, like unit commentary, should we -- I mean, it looks like the guidance would imply a little more normal industry seasonality type numbers, at least the low end of the guidance. I'm curious, is that a change in terms of how we should think about the seasonal behavior of the business from here? I know in most years in your history, you've always grown units sequentially from 3Q to 4Q. So we're a little bit surprised by the guidance this time.

So curious if that is a change in how we should be thinking about seasonality? Or is it just more being conservative around the macro backdrop, anything you're seeing out there from a consumer backdrop standpoint? And I have a follow-up.

I'll jump on this one. I think it's largely more of the same. I think when you look at the last several years, Q3 to Q4 for us or for other retailers, there's a decent amount of variability in the shape that you see Q3 to Q4. And so I think we're taking that into account and we're guiding. But I think we continue to see extremely strong growth.

You saw it this quarter. We expect that heading into Q4 and into next year. I think we're on a great path and everything remains the same.

Understood. Okay. And then just on the other GPU within that, just the ancillary product penetration. It looks like you're starting to chip away at that.

Can you maybe size for us what the penetration levels are today in that business? Or how much was it up year-over-year? And curious like how we should think about benchmarking that? I mean if you look at some of the franchise retailers out there, they make right roughly $1,500 a unit or higher at a 45% penetration.

I'm curious, is that kind of a benchmark in terms of the long-term opportunity there? Are we thinking about this differently? Any thoughts there would be helpful.

Sure. I think there's a number of parts to that. I think as a general matter, I think other GPU is an area with a couple of line items underneath it. And like everyone else in the business, it's an area where we believe there are fundamental gains to be had.

And I think we've been working on those for a while, and I think we've got plans to continue to work on those, and we think there's certainly opportunity there. Just to generalize a bit, I think that's true in every GPU line item and every expense line item. We continue to feel like we've got extremely exciting opportunities and the hardest part is just prioritizing those. I think thinking about any part of our business, I think, relative to the kind of mature pre-existing automotive retail industry, I think, is a reasonable starting point.

I think as a general matter, we've always sought to outperform the benchmarks that you would see if you compared us to the outside industry. But I think in ancillary products, I think something we want to make sure that we do is we deliver very simple, high-quality value-added products to our customers. And so I think that's a guiding principle that we'll make sure that we continue to adhere to. But there's definitely additional opportunity, and we definitely plan to go unlock it.

The next question comes from Brian Nagel with Oppenheimer.

So a couple of questions. My first question, look, we -- obviously, the results you put up there on the used car side are very, very strong. The question I want to ask, I mean, there's been other data points within the space that have suggested weaker, choppier used car demand. So is there anything you're seeing that's below the reported results that suggest a more difficult demand environment?

And I guess maybe the ancillary to that is, is there something changing here to allow Carvana to capture even greater share in this most recent quarter?

I think as a general matter, I think things continue to look pretty similar at a high level is I think how we'd characterize things. I mean we're always paying attention. I think we get a good read on what things look like, obviously, looking at retail sales and also, as discussed earlier, looking at loan performance across the loan book. And I think as a general matter, things feel relatively stable.

I think we're always paying close attention. And I think, as you alluded to, I think while we don't see signs of macro weakness today, that we're very well positioned if -- I guess, when that does come to pass. At some point, there will be cycles. And I think where we are from a financial performance perspective relative to the industry and then where we are from a cash perspective and a balance sheet perspective and where we are from a consumer offering perspective and a business scalability perspective, I think the sum of all that is very good.

And so I think as a general matter, like I said, things look good. The most important ways that we measure ourselves is how are we performing relative to the industry in customer experience, in growth and in economics. If we're always progressing in those areas, we're going to be on a really good path because this is obviously a very mature industry. We know what the scale of the industry are and we know what the economics of the industry are.

So I think that's the single most important thing, but nothing notable to call out.

That's very helpful. And then my follow-up question, I guess, longer term in nature, but you mentioned in your script, just AI and to the extent to which AI is helping to enhance that consumer offering. And maybe if we could talk a little bit further about that. I mean as a consumer of Carvana now, where is AI helping my experience?

And kind of how far along are you in this process now of integrating that technology?

I think we're pretty far along. Unfortunately, every company in the world knows they're supposed to talk about their AI strategy and every investor in the world knows that every company knows they're supposed to talk about their AI strategy. So I think what we try to do is we try to put in some anecdotes that are hopefully clear that demonstrate real capabilities. So I think there's 2 chats that we put in there that are super interesting and point to what we're capable of doing.

One that we put in our shareholder letter, kind of -- it just shows a customer asking about a car and when it can be delivered and they ask for a specific color and they ask for a specific payment. And for that to be -- for this agent to be able to answer that question, it needs to be able to interact with our finance service, with our scheduling service, with our search service. It needs to be able to do a lot of things. And then it also, interestingly, in line, we drop an image of the car that is clickable and pulls them into the VDP.

It has to be able to have a designed dynamically rendered response to the customer that is sort of like a very early iteration of a dynamic UI, which I think is really interesting. So there's at least kind of 4 key capabilities there. And those capabilities exist not to serve our AI processes and capabilities, they exist to serve our entire business. The entire business is built to be automated and self-service and simple for the customer.

It's just traditionally been in more of a standard UI structure that is click and scroll. But I think as all these teams build these services and they embed all of the business logic into our systems, and then they make those systems and all that data readily accessible. It makes it very straightforward for us to build these very complex tools. I think we showed another one that's on the right of the page in the shareholder letter that is also very interesting and is interacting with very different data.

And that shows a customer that is uploading their insurance document to kind of have that taken care of prior to taking delivery of their car. To do that, we have to know state-by-state rules. We have to be able to absorb the document. We have to be able to scrape down that document, convert that to data, apply that against business rules, figure out where we're in compliance and where we're not and then articulate to the customer what they need to do.

And all that has to happen in an automated way. There's a number of systems that are required to do that. So I think those 2 chats, chat is one possible interface, but they sort of reveal the brain behind the chat. And I think across the business, there's very interesting things happening everywhere.

We're generally a very technology-forward company with a lot of ambitious, curious, excited people. And I think as a result, we tend to adopt technologies very quickly. I think another very different but extremely fun anecdote that is pretty recent from inside of Carvana is our team calls these ambient agents. I don't know if that language is an industry term or it's just what they call them, but I think it's descriptive.

We now have some agents that basically have triggers, so they don't require a human to pose a query. They just have triggers that can be data informed. A customer can run into a bug on a website, and it can automatically kick off an agent that then knows to go investigate that bug, try to figure out what's going on and then inform us what might be wrong. We recently had a version of that, that was triggered by one of the triggers that we set and no human kicked it off.

It went and identified a bug. It suggested a solution. It wrote code. It sent it over to a person and then that person approved the code and the code was deployed.

That's really like that's basically sci-fi from the perspective of 2 years ago. And I think that's also indicative of what's going on inside the company. So I think we are structured to benefit from this. And I think that we've got a lot of very high-quality people that are working very hard to make sure that we take full advantage of it.

And I think that we're well on our way.

The next question comes from John Colantuoni with Jefferies.

I just wanted to start with the EV tax credits. Given your mix of EVs is greater than the industry average. Can you give us some perspective on how you see the elimination of the federal tax credits impacting demand for used cars in that space? And how you're making any necessary adjustments to minimize the impact on Carvana's growth trends? And I have a follow-up.

Sure. I think as a general matter, the expiration of those credits clearly mattered and clearly shift customer selection. I think the evidence so far is pretty clear that it's just a shift in preference of vehicles, not a change in aggregate demand, at least not one that is noticeable. So I think our system is well positioned to handle that.

We've got -- our system is, for lack of a better description, sort of listening all the time to what our customers interacting with and what is that they want. And then we're making sure that we replace the cars that they want based on the actions that they're taking. And so we kind of have a system that pretty naturally adapts. And I think that what you'd expect, we have seen, we've seen a reduction in EV purchases as a result of the expiration of that credit.

And I think the system has adapted in a way that in the numbers is basically not something that you really see or need to be called out. And then I think as a general matter, I think we continue to be believers in EVs. I think all these new technologies go through their positive moments and their tougher moments. And I think it is true that EVs are a very high-quality fundamental technology that's early in their curve.

And we expect over time that they will make a come back, and we'll be well positioned for it when they do.

Okay. Great. And you announced sort of a second franchise dealership acquisition last month. Can you talk about the results from your first foray into physical dealerships that made you acquire a second? I'd be curious if your findings suggest that this could be an area of investment for you in the coming years.

Yes. I appreciate the question. I think it remains early. It would be a bit premature to comment. So we're going to kind of stick to focusing on the core business and stay tuned for the future.

The next question comes from Christopher Bottiglieri with BNP Paribas.

First, I was hoping to delve into the same-day delivery test, which sounds pretty exciting. The logistics per unit went up for the first time, I think, in 10 quarters, which tells me people are using it significant. But can you just frame -- and obviously, it's going to pay for itself if there's a sales lift, but can you frame for us how performance in Phoenix is doing versus the control market that hasn't seen this type of increase or whatever you can tell us because it sounds like this might be an area you're going to invest in '26.

Sure. Well, I think, first of all, there clearly is a very clear relationship with speed and conversion, just like every other e-commerce business. I think that's something that we've seen across the business across time, and it's something that we continue to see in the business today. So I think that, that is a reason to focus on this and build this capability out.

I think from a bigger picture perspective, we also just think it's tremendously differentiating and very exciting and kind of strategically important to be able to do something that other companies just simply can't do. And so we -- in my prepared remarks, I had my dramatic pause where I asked you to contemplate what it means to be able to buy thousands of cars in minutes and have them delivered in hours. But I really do think it's useful to think about what that means and what that looks like as it feeds back over time, and we get more and more inventory pools closer to more and more customers and those inventory pools get larger and larger and customers have more selection and we automate more and more of our processes and the speed and ease gets simpler.

We think that we're building a machine that is qualitatively different and structurally different than any other machine that's out there. And so there's certainly -- we would expect for there to be conversion tailwinds as we continue to work on this in Phoenix. And then once we feel like we're in a really good spot, start to roll it out to more locations. But more importantly, we think it just continues to separate us as a completely different business and a completely different offering to consumers that will enable us to have completely different kinds of results over a very long period of time.

Got you. That makes sense. And then wanted to parse the other GPU commentary out a little bit more. So it sounds like attach rate was up.

You probably benefited from rate cuts because you don't perfectly hedge. There was another Fed rate cut in Q4. So that should be another tailwind that mitigates lapping that. But it sounds like you're going to reinvest that into the consumer proposition to offer lower rates to the consumer. I just wanted to, a, confirm that.

And b, how do you think about beyond when there's some more rate cuts and you kind of lap this into '26? Do you feel like the rates go back up? Or how do you think about the value prop to the consumer on financing once the rates stop going down?

Sure. Yes. So I mean, I think we talked about some of the drivers of strength in other GPU. I think strong loan performance, strong performance on loan sale monetization and cost of funds.

There's some positive trends we're seeing in finance attach. I do think those are driven by lower rates. We're also seeing some positive trends in ancillary product attachment rates as well. I think our viewpoint, and you'll notice we had a record in other GPU this quarter.

It's our highest level ever. That's really driven by these fundamental gains that I was just pointing to. I think in Q4, our plan is to pass these fundamental gains back on to customers. So other GPU in we think we'll end up looking something much more like Q4 2024 rather than Q3 2025. And I think that's something we feel really great about.

We really have driven meaningful fundamental gains in the finance and ancillary products platform, and that gives us an opportunity to pass some of those gains on to customers, for example, in the form of lower interest rates.

The next question comes from Daniela Haigian with Morgan Stanley.

So first, clearly, Carvana has built a strong digitally enabled mousetrap in the dealer business. But how do you think about competition from new entrants such as Amazon that also have warehouse and logistics capabilities? How does that feed into your, I guess, expected return on ad spend? And then also on that same line is what is the biggest gating factor in your near-term growth curve?

I think as a general matter, we try to think about making sure that we're delivering the best customer experience as we possibly can. We try to make sure that we're paying close attention to every line item in the business and doing all the hard work that's necessary at the detailed level to constantly make sure everything gets better. And we try to focus less on any given competitor. I think that served us very well over time and brought us to this place where depending on what profit metric you're looking at, we're 2 to 2.5x as profitable as the average automotive -- the other average automotive retailers.

And I think that to me, that's, I think, probably like the most important single way to look at this. I think there's a question about what new entrants can look like over time and how many there will be and what scale they will come at and what approach they will take. And those are all fair questions that we can all speculate on. But I think there are also facts that today, 98.5% of used cars and 99% of cars in some total are sold by traditional retailers that have the economics that we discussed earlier that are materially different than ours and aren't super well positioned to build a machine that looks like ours.

And so I think to the extent that we just stay focused on ensuring that we've got a scalable business that's delivering great customer experiences with different economics, I think anything that is likely to come is unlikely to be powerful enough to change that 98.5% or 99% of the industry that looks the way it looks today. And so that's where I think continually, at least from my perspective, my personal favorite metrics are how are we doing from a growth perspective relative to the industry, how are we doing in customer experience versus the industry and how are we doing in economics versus the industry?

Because I just think that when you have a capital-intensive business, that requires lots of work and lots of scale to deliver good customer experiences, you're competing against the industry, and it's unlikely the entirety of the industry can move very fast in light of all that capital investment that's necessary. You see all the things that we're having to do to move at the speed that we're moving. And I think what I would say is the simplest reduction of kind of what is the constraint. It's basically just the sum of effort across this large complex business where you're moving things and you're organizing people, and it's a lot to do.

And I think that, that gates the speed at which you can run. And that's why we're doing all this work to make sure that we stay in front of ourselves. not certainly not what you asked about, but relevant, I think, in this conversation. We talked a lot about what we're doing with ADESA and ADESA Clear and our inspection centers, and we kind of put this new concept in the shareholder letter about having retail capabilities, wholesale capabilities or retail and wholesale capabilities at all these different centers across the country. That's the kind of work that you can see in that graph, that's us doing work as fast as we can that is very complicated to be able to unlock those capabilities over time.

And I think it's positioning us well for a broad future, and it's indicative of the kind of work that's necessary to scale a business like this.

Great. And I guess on that piece of your moat of what you've built out on this physical business, you also have a very low capital intensity in building this out. I think a lot of your fixed costs are already embedded. And so as you think about making progress towards that 3 million unit target in 5 to 10 years, what do plans look like to expand production capacity beyond that 3 million?

And what are the capital requirements to get there?

I think our eyes are as big as anyone out there. And I think the opportunity in front of us is very, very large. And I think there's no doubt that the goal that we're chasing today that is time bound is our current goal, and we expect to have other goals beyond that. I think it's premature to talk too much about those other goals because I think that we've got several years here of hard work to make sure that we get to the $3 million and the 13.5 million.

But I think there's no question that there's opportunity beyond that. And I think that you can probably look at our past when we've attacked problems in the past to get a sense of what that future could look like. But I think it's early for us to be giving specific guidance and expectations on that today.

The next question comes from Jeff Lick with Stephens.

Congrats on a great quarter, guys. In terms of the sourcing environment, I was just curious if you can comment on any evolutions of that. I know your relationships or how you're doing business with some of the commercial rental providers has changed a little bit. And then also just as you grow into sourcing 600,000, 700,000, 800,000, 900,000 units, buying from people's driveway, just any evolution there would be helpful and just the color on that.

Sure. Well, I think the most important fundamental there is what we alluded to a bit a moment ago, it's just making sure that the business is structured to be a structurally better buyer of cars so that we can be a better partner to partners out there, and we can give very exciting bids to our customers. And I think if we divide the types of cars in the world into wholesale and retail with retail being defined as a car that we're well positioned to retail, it's very obvious that we are deeply structurally advantaged in buying cars that we are well positioned to retail. I think as one of the many benefits of partnering up with ADESA is that it put us in a spot where we're also structurally advantaged to be able to buy cars that are wholesaled.

And then I think when we do the further work to unlock both wholesale and retail capabilities at the same locations, I think we become a better buyer again because not only are we well positioned to dispose of both types of cars, we're also well positioned to reduce the expenses that are traditionally inherent in the system that take the form of time and extra shipments and cost. And so I think we are doing the work today to go unlock those capabilities. In that graph, we've got 74 sites. We now have 41 that we label as wholesale only.

That's the original 56 ADESA sites minus the 15 where we have added reconditioning capabilities. And so those 15 now represent both wholesale capable and retail capable sites. We have 6 sites that are just retail. Those are the 18 inspection centers that we had prior, minus the 12 where we've added ADESA Clear, which is a digital auction capability.

And then we've got 27 that are both, which is the sum of the 12 inspection centers that have ADESA Clear plus the 15 integration sites where we've added reconditioning capabilities to ADESA. So we now have 27 sites where we're well positioned to handle any type of car very efficiently, not just because we have a great wholesale distribution channel and a great retail distribution channel, but also because we can do both more efficiently. So to me, that's the structural thing that we're doing. And I think the more progress we make there, the better position we're going to be.

And then I think we continue to make progress with our partners. And I think there will probably be more to talk about there over time. But as long as we position the business for it very well, I think we continue to be in a great spot to take advantage of that.

And when you get to the retailing of 2 million to 3 million cars, do you think the proportion of where you source will change much? Or will it be pretty much the same?

I think we'll see over time. I think it's early to call a shot there. I mean I think at like the simplest, most fundamental level, most of the used car market is customers swapping cars with each other. And then they just do it through many different mechanisms.

That's not strictly true because you do have off-rental cars and then you do have cars that flow out of fleets. I think you could call kind of off-lease cars, something sort of in between because it is a customer car that makes it to another customer. But generally speaking, cars are just moving through some elaborate mechanism from one customer to another. So we think having the business of buying cars from customers is essentially important.

I think there's many forms that can take over time. And then we think having a business of being able to buy cars very efficiently from business disposers of cars from fleets is also centrally important. And the machine is being constructed in a way where we feel like we're an advantaged buyer regardless of where cars are coming from. And then I think if you look at buying cars from customers, that is a slightly different offering than selling cars to customers.

But now for probably 5 or 6 years, those 2 brands have grown pretty much in lockstep, whether you're looking at the percentage of cars that we're retailing that were sourced from customers or if you're looking at our wholesale to retail ratio, they bounce around a little bit. But generally speaking, they've been pretty consistent. And I think that just speaks to, a, the fundamental that largely this market is customers swapping with each other. And so there's a similar sized market to buy cars as there is to sell cars; and b, the fact that those 2 businesses are growing at about the same rate as we continue to grow our brand in a way that benefits both sides of the business.

So like I said, I think we're well positioned either way. I think it's early to call our shot. We'll hope to succeed in both areas.

The next question comes from Andrew Boone with Citizens.

Ernie, you talked about scale in the beginning of your prepared remarks. And then you mentioned automation multiple times as we've gone through this call. Can we just step back? And can you just talk about what are the biggest opportunities that you have to increase automation as you do gain scale?

Like what are the key variable costs that you guys can really drive down that still remains in the model?

Sure. Well, I'll point again to the anecdote because I just think that they communicate very well. So what I would say is if we look back to that chat that we demonstrated in the -- or that we showed in the shareholder letter that shows like an insurance document and the customer interacting with that, I think that gives you a sense of the kind of thing that can be done. It's basically just expanding automation at greater depth so that the entirety of the process can happen in an instantaneous way where very clear instructions are given to the customer.

They know exactly what to do, and they're able to complete a task with no latency. I think that's been part of a progression over years as we've kind of built those capabilities. We first have to learn all the rules across different states. We have to make sure that those are written down and codified.

We have to figure out what our particular rules are going to be for our business. We then have to go and find a way to get data to get uploaded into our site. And initially, that data is manually looked at and checked against business rules and the customers approve. And then you add the ability to scrape the data off of the document that's passed to us and then the tools are built to make that simpler.

And then you scrape the data and you automate the checking of that data against that business logic. And so to me, it's just that constantly deepening level of detail of automation that is very valuable. And I think that's certainly valuable to reducing costs. I think what we at least tend to think is the maybe even more exciting thing is just being able to differentiate the offering.

Just making sure that customers can know exactly what to do and shop with extreme confidence and get offers that are amazing. Going back to being able to sit there in minutes, look at thousands of cars can be delivered in hours. That requires -- that's a very different kind of offering that we think is very strategically valuable, but requires a lot of things happening in the background. So I think every part of the business, finance verifications, registration, customer care, every part of the business, continual gains in reconditioning and things that we're doing there, things in logistics across the business.

I think there's opportunities to make every single workflow simpler and more automated, more repeatable and more scalable. And I think we've just been continually doing that work over and over. And hopefully, you feel like you see it showing up in the results.

If I could sneak in one more quick one. How do you think about the guardrails of expanding same-day delivery? How do you guys think about making sure that rollout is smooth? And what are the profitability metrics that you guys are involving to make sure you guys are containing what may the cost for that?

Sure. Well, I think like anything in the real world, I think the first thing you have to do is you have to kind of aim for something that's hard and then you start to see what are all the constraints in the system that are causing you to be limited in what you're able to achieve and then you have to go attack the biggest constraint and then move on to whatever the next constraint is that emerges. And I think that's why we're working really hard in Phoenix right now to attack those constraints one at a time. And I think we've seen a lot of progress there.

Phoenix looks like other markets in the country just several months ago. And now we've got 40% of customers getting same or next-day delivery compared to approximately 10% in the rest of the country. So we've obviously made rapid progress there. I think we will continue to try to progress in Phoenix.

And then undoubtedly, the next step is going to be to roll that out to other inventory pools that are near large population centers, and we'll prioritize that intelligently. And then I think that's another place where you can kind of see the entire playbook. And then as we add more integration sites with ADESA, so we've got retail capabilities in more spots, then we can have inventory pools in more spots that are closer to more customers, and we can roll out that same-day delivery capability in more spots. So I think it's going to be a multifaceted, multistep approach over the next several years.

But I think step 1 is proving out that we can do it at meaningful scale. I think that box is checked. Step 2 is making sure that we really nail it in Phoenix. And then step 3 will be continuing to roll it out from there.

The next question comes from Michael McGovern with Bank of America.

There's a lot of talk out there about the K-shaped economy where you have lower income cohorts of consumers seeing relatively more pressure relative to higher income. Curious if there's anything that you've been able to see on that front, demand trends between the 2, especially since your unit guidance implies some deceleration. Is there any notable deceleration from lower income cohorts specifically?

I really don't think we have anything interesting to say there, and apologies. I think there's no question that -- there's a lot of story lines out there that point in that direction. And I think in our data, we can look at sales data or we can look at credit performance data, and we can kind of try to cut it in many different ways. I just don't think that there's interesting super validating data points that we can point to for that story.

I think what we see tends to look a lot more consistent than that particular story, but we'll obviously continue to pay attention, and we'll follow the data where it goes.

Got it. And then in Phoenix specifically with same or next-day delivery, I'm curious, can you discuss kind of what you expect for GPU or EBITDA per unit or anything that you could give us kind of what that investment, if you will, looks like to deliver the cars more quickly? Or is it pretty seamless since you have that infrastructure there built out already?

Yes. I would say it's really more the latter. I think that the same-day delivery is really more about a technology investment at this stage and a process investment, making sure that it's a complex transaction. And in order to have same-day delivery, you need to nail every single aspect of a complex transaction accurately and in a short period of time.

And so that's a technology focused investment. It's also -- it requires strong processes from the operators that are executing that. But that's really like the main investment. There's some incremental investment in staffing just to make sure that you have the capacity available to execute same-day delivery.

And so you have a little more slack capacity there. But that's not a very large dollar amount in the grand scheme of things. The way that we're executing same-day delivery today, it's really more about a focus, a technology and process investment more so than a cost investment.

The next question comes from Michael Montani with Evercore ISI.

I joined a minute late, but I did want to ask if you discussed at all the advertising expense. It sounded like that might be expected to go up. And I just want to see is that quarter-over-quarter or on a per unit basis? Any color that you could provide on that one? And then I had a quick follow-up.

Sure. I can take that one. The outlook that we gave for advertising expenses on a dollar basis for it to be similar in Q4 to this Q3, maybe slightly higher, but similar to slightly higher. And again, where that comes from is really just continuing to invest in building our brand, building awareness, understanding and trust of our brand is one of the 3 key pillars of our long-term growth strategy.

Okay. And then just on the wholesale GPU, were you signaling that, that could step down quarter-over-quarter by about 25% to 30% the way it did last year? And if that's the case, I was just thinking there could be opportunities for improvements given some of the enhancements you've made in terms of processes. So I just wanted to make sure I had that right or if there's anything else to dig into from a depreciation perspective, et cetera, to know.

So I think we really think of sequential changes on a per unit basis, what we called out. And so I do think that there's seasonality in a multiple of the GPU line items. I think in wholesale, that takes the form of higher wholesale depreciation rates in Q4, lower auction volumes in Q4 than other times of the year. So we do typically see a seasonal pattern there.

And we called out something seasonal, something similar to last year sequentially on a per unit basis.

Last question comes from Chris Pierce with Needham.

Can I just ask one big picture question. The 3 million unit goal, I guess, is that strictly around what determines when you could hit it earlier or later? Is that about adding recon scale personnel? Or is it about -- did you consider end markets or credit cycles or anything?

I'd just love to know kind of big picture, how you came up with it, what drives it and what could pull it forward or push it back?

Sure. I would say at a high level, the time lines we provided there were 5 to 10 years, which correspond to 2030 to 2035. And I think the fast end of that is approximately 40% compounded growth and the slow end of that is approximately 20% compounded growth. I think as a general matter, we view that as largely driven by our ability to continue to execute is probably the biggest determinant of that. There's a lot of work that has to be done across the entire business to make sure that we're buying cars, reconditioning cars, delivering cars to customer long leg and last mile, handling customer questions and just scaling the entirety of the business.

So I think there's a lot of work in there. And I think our execution is the primary driver that we think will dictate when we achieve that goal.

That's all the time we have for questions today. I would like to turn the conference back over to Ernie Garcia for any closing remarks. Please go ahead.

Great. Thanks. Well, thanks, everyone, for joining the call. Carvana team, another awesome quarter. Thank you guys so much. You really have a lot to be proud of. I hope you are proud. I hope the high fives fly, and then let's come back tomorrow and keep it going. We have a lot more work to do. So thanks to all of you.

The conference has now concluded. Thank you for attending today's presentation. You may now disconnect.