Image source: The Motley Fool.
Date
Wednesday, October 29, 2025 at 5:30 p.m. ET
Call participants
Chief Executive Officer — Ernie Garcia
Chief Financial Officer — Mark Jenkins
Need a quote from a Motley Fool analyst? Email [email protected]
Takeaways
Retail Units Sold -- 150,941 retail units sold, a 44% increase and a new company record for Q3 2025.
Revenue -- $5.647 billion, up 55%, reflecting higher average selling prices and the inclusion of gross revenue from vehicles acquired through a retail marketplace partner in Q3 2025.
Adjusted EBITDA -- Adjusted EBITDA reached $637 million in Q3 2025, up $208 million and a company record, though margin decreased to 11.3% from 11.7%.
GAAP Operating Income -- $552 million in operating income for Q3 2025, up $215 million, with operating margin improving to 9.8% from 9.2%.
Net Income -- $263 million, an increase of $115 million year-over-year, with net income margin rising to 4.7% from 4%.
Inventory Selection -- Customer selection increased nearly 50% year-over-year in Q3 2025, while inventory turn time remained approximately flat.
Same- and Next-Day Delivery -- In Phoenix, 40% of customers received same or next-day delivery, versus 10% nationwide, representing a major operational advance.
Retail GPU -- Non-GAAP retail GPU declined by $77, driven by higher retail depreciation rates in Q3 2025.
Wholesale GPU -- Non-GAAP wholesale GPU decreased by $168, due to higher depreciation rates and retail units sold outpacing ADESA marketplace growth in Q3 2025.
Other GPU -- Non-GAAP other GPU increased by $63, attributed to improved cost of funds and higher finance and BSC attach rates, partly offset by an elevated loan sale mix from Q3 2024 originations in Q3 2025.
SG&A Leverage -- A 44% rise in retail units sold resulted in a $319 reduction in non-GAAP SG&A expense per retail unit sold in Q3 2025.
Advertising Expense -- Advertising expense increased by $139 per retail unit sold in Q3 2025, with expectations for Q4 2025 to be similar to or slightly higher than Q3 2025.
Loan Sale Partnerships -- Expanded in October 2025 to agreements for up to $14 billion in future loan principal, including an increase in the Ally agreement to $6 billion through October 2027 and two new $4 billion agreements with separate partners extending through late 2027.
SG&A Components -- Carvana operations SG&A (non-GAAP) fell by $96 per retail unit sold in Q3 2025, primarily due to operational efficiencies; overhead SG&A (non-GAAP) decreased by $314 per retail unit sold, while advertising (non-GAAP) increased by $139 per unit.
Conversion of Adjusted EBITDA -- 87% converted to operating income in Q3 2025, up from 79% in Q3 2024.
Net Debt Position -- Net debt to trailing twelve-month adjusted EBITDA ratio reduced to 1.5x (non-GAAP) following $1.2 billion in debt retirement across 2024 and 2025, with cash on the balance sheet exceeding $2.1 billion.
Q4 Outlook -- Guidance for retail units sold above 150,000 and adjusted EBITDA (non-GAAP) at or above the high end of the prior $2 billion–$2.2 billion range for the full year 2025, contingent on a stable environment.
AI and Automation Progress -- Notable advances in customer self-service (over 30% of retail and over 60% of selling customers complete processes digitally before delivery as of Q3 2025) and AI-driven operational systems impacting verification and process automation.
EV Mix and Tax Credit Impact -- Expiration of federal tax credit led to a decline in EV share, but aggregate demand remained stable and platforms adaptively shifted inventory.
Summary
Carvana (CVNA 2.28%) achieved record levels in retail units sold, revenue, adjusted EBITDA, and operating income for Q3 2025, reinforcing rapid profit and scale expansion. Management enacted major steps to strengthen liquidity through substantial debt reduction and more than $2.1 billion in cash on the balance sheet as of Q3 2025, positioning the company at its strongest financial standing to date. Strategic investments in same- and next-day delivery, notably in Phoenix, improved customer delivery rates and showcased operational differentiation in Q3 2025. Recent agreements in October 2025 secured $14 billion in forward loan sale capacity, formalizing partner relationships and validating asset quality on the lending platform. The company maintained year-over-year cost leverage through reductions in non-GAAP SG&A expenses per unit in Q3 2025, even as advertising and operational investments grew, reflecting confidence in ongoing top-line growth and margin potential. Enhanced automation and AI integration resulted in a significant portion of customers completing end-to-end processes digitally before vehicle handoff, furthering operational efficiency.
Garcia described a path to selling "3,000,000 cars at a 13.5% adjusted EBITDA margin over the next five to ten years" (non-GAAP) and emphasized that achieving high growth and profitability simultaneously indicates "something that is structurally different."
Jenkins stated that loan performance for "2024 and 2025 loan originations are performing extremely well" and cited the Ally agreement expansion as external validation.
Garcia noted that 98.5% of used-car sales are still with traditional retailers, supporting the potential for continued share gains given Carvana's model and economics.
The company cited substantial technological progress, including "ambient agents" according to Ernie Garcia capable of detecting and deploying code fixes without human initiation.
Industry glossary
ADESA: A wholesale vehicle auction platform integrated into Carvana's logistics and retail business.
BSC: Ancillary product related to Carvana's "other" Gross Profit per Unit (GPU) line, typically referring to bundled service contracts.
GPU: Gross Profit per Unit; a key non-GAAP metric for overall per-vehicle profitability, segmented into "retail," "wholesale," and "other" in Carvana disclosures.
SG&A: Selling, General, and Administrative expenses, measured on a per retail unit basis for operational efficiency analysis.
ADESA Clear: Carvana's digital auction capability for wholesale vehicles through ADESA.
Full Conference Call Transcript
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-Ks 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?
Ernie Garcia: Thanks, Meg. 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 two times 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, 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 3,000,000 cars at a 13.5% adjusted EBITDA margin in the next five to ten 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 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 eighteen 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 five 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, taking time to think through the implications. Thousands of vehicles 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 is 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 to 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. 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 3,000,000 cars per year with a 13.5% adjusted EBITDA margin in five to ten 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?
Mark Jenkins: 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 billion, 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 150,941 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 three 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 leveraging 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 BSC 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 two agreements formalize existing relationships and establish defined expectations for sale volume 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 leverage 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. The Carvana operation 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%, 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 non-cash expenses, which we'll continue to leverage with scale. We converted approximately 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 1,000,000 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 twelve-month adjusted EBITDA ratio is now down to just 1.5 times, 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, pursuing our goals of becoming the largest and most profitable auto retailer and buying and selling millions of cars. Thank you for your attention. We'll now take questions.
Operator: Thank you. The first question comes from Sharon Zackfia with William Blair. Please go ahead.
Sharon Zackfia: Hi, good afternoon. The topic 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 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? Thank you.
Mark Jenkins: 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 upsize from $4 billion to $6 billion based on the performance trends that they're seeing. And additionally, the addition of these two 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 two large agreements are with existing partners. 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 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.
Operator: Okay, great. Thank you. The next question comes from Marvin Fong with BTIG. Please go ahead.
Marvin Fong: Great. Thanks for taking my question. I hopped on just about five minutes ago, so I apologize if I missed this. But 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 the better way to measure that metric? Can you just kind of talk about your future opportunities to continue to kind of like drive down your operations cost per unit? That'd be great. Thank you.
Mark Jenkins: 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 three-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 three-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 one-time items or other non-recurring 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.
Marvin Fong: 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 now just kind of wondering what 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 kind of maybe double click on what's kind of behind the year-over-year flat flattish retail GPU guidance?
Mark Jenkins: 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 one thing I would say was 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. 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 about Q3. Retail GPU.
Marvin Fong: Okay, perfect. Thank you so much, Mark.
Operator: The next question comes from Rajat Gupta with JPMorgan. Please go ahead.
Rajat Gupta: Thanks for taking the question. Just on the fourth quarter, like unit commentary, should we expect, 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, Ed, 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 Q4 to Q4. 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, of consumer backdrop standpoint? Thanks. And I have a follow-up.
Ernie Garcia: Sure. I'll jump in 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, 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, everything remains the same.
Rajat Gupta: Understood. Okay. And then just on the other GPU within that, you know, just ancillary product penetration, looks like you're starting to chip away at that. Could 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 I mean, you look at some of the franchise leaders out there, they make roughly $1,500 a unit or higher at a 45% penetration. I'm curious, is that 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. Thanks.
Ernie Garcia: 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 everywhere else in the business, it's an area where we believe that our gains to be had. And I think we've been working on those for a while. 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 at a general matter, we've always sought to outperform the benchmark that you would see if you compared us to 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.
Rajat Gupta: Okay, great. Thank you, and good luck.
Operator: The next question comes from Brian Nagel with Oppenheimer. Please go ahead.
Brian Nagel: Hi, good afternoon. So a couple of questions. Hey, how are you? Good. My first question, look, we obviously the results you put up here on the used car side 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 suggests a more difficult demand environment? I guess maybe the answer to that is, is there something changing here to allow Carvana to capture even greater share in the most recent quarter?
Ernie Garcia: I think as a general matter, I think things continue to look pretty similar at a high level as I think how we 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, 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, I think that we're very well positioned if we don't 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 scales 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.
Brian Nagel: That's very helpful. And then my 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. Maybe we could talk a little bit further about that. I mean, as a consumer of Carvana now, where is AI helping my experience? How far along are you in this process now? Integrating that technology?
Ernie Garcia: I think we're pretty far along. I'm not I think, unfortunately, every company in the world knows they're supposed to talk about their AI and every investor in the world knows that every company knows they're supposed to talk about their AI strategy. So what we try to do is we try to put in some anecdotes that hopefully clear, that demonstrate real capabilities. So I think there's two chats that we put in there. That I think 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 the 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 four 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 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 two 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 if 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 can run into a bug on a website and it can automatically 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 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 two years ago. 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.
Brian Nagel: That's very helpful. Congratulations and thank you.
Operator: The next question comes from John Colantuoni with Jefferies. Please go ahead.
John Colantuoni: Great. Thanks for taking my question. 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.
Ernie Garcia: 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 that 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 are interacting with and what it 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 is adapted in a way that in the numbers is basically not something that you really see or needs 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 comeback and we'll be well positioned for it when they do.
John Colantuoni: Okay, great. And, you announced a 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?
Ernie Garcia: Thanks. Yes, thank you. I appreciate the question. I think it remains early. It'd be a bit to comment. So we're going to kind of stick focusing on the core business and stay tuned for the future.
John Colantuoni: Okay. Thanks. Appreciate it.
Operator: The next question comes from Chris Bottiglieri with BNP Paribas. Please go ahead.
Chris Bottiglieri: Hey, thanks for taking the question. First, was hoping to delve into the same-day delivery test, which sounds pretty exciting. The logistics per unit went up for the first time in ten quarters, which tells me people are using it significantly. But can you just frame and obviously it's going to pay for itself, there's a sales lift. But can you frame for us how performance in Phoenix is doing versus a control market? That hasn't seen this type of increase or whatever you can tell us to because it sounds like this might be an area you're going to invest in 2026.
Ernie Garcia: Sure. Well, I think, first of all, clearly is a very clear relation 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's 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 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.
Chris Bottiglieri: Got you. Makes sense. And then wanted to parse the other GPU commentary out a little bit more. So it sounds like attach rate was up. Probably benefited from rate cuts because you perfectly hedge. There was another Fed rate cut in Q4, so that should be another tailwind, I mean, mitigates lapping that. But it sounds like you're going to reinvest that into the consumer proposition just offer lower rates to consumer. Just wanted to, a, confirm that and b, how do you think of that beyond when there's no more rate cuts and you kind of lap this into '26?
Do you feel like the rates go back up or how you think about the value prop to consumer on financing? Once the rates start going down?
Mark Jenkins: 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 Q4, 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.
Chris Bottiglieri: Got you. Okay. Thank you.
Operator: The next question comes from Daniela Marina Haigian with Morgan Stanley. Please go ahead.
Daniela Marina Haigian: Hi, thanks for taking the question. So first clearly, Carvana has built a strong digitally enabled mousetrap in this dealer business. But how do you think about from new entrants such as Amazon that also have warehouse and logistics capabilities? How does that feed into your I guess, return on ad spend? And then also on that same line is what is the biggest gating factor in your near-term growth curve?
Ernie Garcia: 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. And try to make sure that we're paying close attention to every line item in the business and doing all the hard work that is 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 two to 2.5 times 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, 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 in a growth perspective relative to the industry, how are we doing in customer experience versus the industry and how we're 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 gates the speed at which you can run. And that's why we're doing all this work to make sure that we stay way 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 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.
Daniela Marina Haigian: Great. Thanks, Ernie. And I guess on that piece of your moat of what you've built out on this business, you also have a very low capital intensity in building this out. I think a lot of your fixed costs are already embedded. So as you think about making progress towards that 3,000,000 unit target in five to ten years, what do plans look like to expand production capacity beyond that 3,000,000 and what the capital requirements to get there?
Ernie Garcia: I think our eyes are as big as anyone out there. 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 ensure that we get to the 3,000,000 and the 13.5. But I think there's no question that there's opportunity beyond that.
And I think that you can probably look at our past and how 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 given specific guidance and expectations on that today.
Daniela Marina Haigian: Thank you.
Operator: The next question comes from Jeffrey Francis Lick with Stephens. Please go ahead.
Jeffrey Francis Lick: Good evening. Congrats on a great quarter, guys. It was a great pause, by the way, Ernie. So I appreciate it. Was Meg's idea. So please give her credit.
Ernie Garcia: Absolutely. Absolutely.
Jeffrey Francis Lick: In terms of the sourcing environment, was just curious if you could comment on any evolution 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, just a color on that.
Ernie Garcia: Sure. Well, I think there's 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. We're also structurally advantaged to be able to buy cars that are wholesale. 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 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've added reconditioning capabilities. And so those 15 now represent both wholesale capable and retail capable sites. We have six sites that are just retail. Those are the eight 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 decision channel and a great retail decision 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 very well, I think we continue to be in a great spot to take advantage of that.
Jeffrey Francis Lick: And when you get to the retailing of 2,000,000 to 3,000,000 cars, do you think the proportion of where you source will change much or will it be pretty much the same?
Ernie Garcia: I think we'll see over time. I think it's early to call our 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 can 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 centrally 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 five or six years, those two 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 the fundamental that largely this market is customer 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 two 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.
Jeffrey Francis Lick: Great. Thanks very much and good luck on the rest of the quarter.
Ernie Garcia: Thank you.
Operator: The next question comes from Andrew Boone with Citizens. Please go ahead.
Andrew Boone: Thanks for taking so much of the question. Ernie, you talked about scale in the beginning of prepared remarks. And then you mentioned automation multiple times. 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? Thanks so much.
Ernie Garcia: Sure. Well, I'll 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 depths 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's that requires that's a very different kind of offering that we think is very strategic 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 and 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.
Andrew Boone: If I can sneak in one more quick one. How do you think about the guardrails of expanding same-day delivery? Or how do you guys think about making sure that rollout is smooth? And what are the that you guys are involving to make sure you guys are containing what may be the cost of that? Thanks so much.
Ernie Garcia: Sure. Well, I think like anything in the real world, I think, you know, 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 looked 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 customers and we can roll out that same-day delivery capability in more spots. So I think it's going to be a multifaceted, multi-step approach over the next several years. But I think step one is proving out that we can do it at meaningful scale. I think that box is checked. Step two is making sure that we really nail it in Phoenix. And then step three will be to continue to roll it out from there.
Andrew Boone: Thank you. Thank you.
Operator: The next question comes from Michael McGovern with Bank of America. Please go ahead.
Michael McGovern: Hey, guys. Thanks for taking my question. 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 two, especially since your unit guidance implies some deceleration. Is there any notable deceleration from lower income cohorts specifically?
Ernie Garcia: 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 storylines out there that point in that direction. And I think in our data, we can look at sales data, 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.
Michael McGovern: 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?
Mark Jenkins: 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.
Michael McGovern: Helpful. Thank you. Thank you.
Operator: The next question comes from Michael Montani with Evercore ISI. Please go ahead.
Michael Montani: Yes, hi. Thanks for fitting me in here. Joined a minute late, but I did want to ask if you discussed at all the advertising expense, sounded like that might be expected to go up. And I just wanted to see if 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.
Mark Jenkins: 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 three key pillars of our long-term growth strategy.
Michael Montani: Okay. And then just on the wholesale GPU, were you signaling that could step down quarter over quarter by about 25% to 30% the way it did last year? And if that's the case, was just thinking there could be opportunities for improvements given some of the enhancements you've made in terms of processes. So want to make sure I had that right or if there's anything else to dig into from a depreciation perspective, etcetera to know?
Mark Jenkins: I didn't so I think we really think of sequential changes on a per unit basis what we called out. And so I do think that their 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. And 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.
Michael Montani: Okay. Thank you.
Operator: Last question comes from Christopher Alan Pierce with Needham. Please go ahead.
Christopher Alan Pierce: Hey, thanks. Can I just ask one big picture question? The 3,000,000 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?
Ernie Garcia: Sure. I've not I would say at a high level, the timelines we provided there were five to ten 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 to be done across the entire business to make sure that we're buying cars, reconditioning cars, delivering cars to customers, 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 drive that we think will dictate when we achieve that goal.
Christopher Alan Pierce: Okay. Thank you. Thank you.
Operator: That's all the time we have for questions. I would like to turn the conference back over to Ernie Garcia for any closing remarks. Please go ahead.
Ernie Garcia: Great. 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.
Operator: The conference is now concluded. Thank you for attending today's presentation. You may now disconnect.