Carvana Co. (CVNA) Earnings Call Transcript & Summary

Hello, and welcome to the Carvana Second Quarter 2026 Earnings Call. [Operator Instructions] I will now turn the conference over to Meg Kehan, Investor Relations. Please go ahead.

Thank you. Good afternoon, ladies and gentlemen, and thank you for joining us on Carvana's Second Quarter 2026 Earnings Conference Call. Please note that this call is being webcast and can be accessed along with our Q2 shareholder letter and supplemental financial tables, on the Investor Relations section of the company's corporate website at investors.carvana.com. Joining me on the call today are Ernie Garcia, Chief Executive Officer; and Mark Jenkins, Chief Financial Officer. Before we get started, I would like to remind you that this 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 these factors can be found in the Risk Factors section of Carvana's most recent Forms 10-K and 10-Q. These forward-looking statements are based on current expectations as of today, and Carvana assumes no obligation to update or revise them. Our commentary today will include non-GAAP financial metrics. GAAP reconciliations can be found in the shareholder letter posted on our IR website. And with that said, I'd like to turn the call over to Ernest Garcia. Ernie?

Thanks, Meg, and thanks, everyone, for joining the call. The second quarter was another exciting quarter for Carvana. We sold almost 200,000 cars in the quarter. The power of compounding is clear in that number as it is almost double the number of cars we sold just 2 years ago. That sales volume puts us at just 2% market share of the used car market and 1.5% of the auto retail market as a whole. These numbers make the size of our opportunity exceedingly clear. In Q2, we also crossed over $3 billion adjusted EBITDA annual run rate for the first time. And that adjusted EBITDA isn't your typical growth company variety as is apparent based on how much flows further down the income statement. Our operating income and net income run rates were about $2.7 billion and $2 billion, respectively. In the shareholder letter, we shared some simple data related to the inventory growth and sales growth by region that looks detailed at first, but that tells a much bigger story. Over the last couple of years, we've been rapidly adding retail production capacity to ADESA sites and existing inspection centers. This has led to variation in inventory growth rates in different parts of the country. The 2 regions where we added the most production capacity, the Midwest and the Northeast, grew inventory by 57%. In those markets, sales grew in the second quarter by 54%. In the 2 regions where we added the least incremental production capacity over the last year, the West and the Southeast, we grew inventory by 17%. In those regions, sales grew by 30% in the second quarter. The middle 2 regions are also reported in the letter and validate this strong correlation. The correlation of this data is driven by the positive feedback in our model we've discussed so many times before in conceptual form. When we grow inventory, any given customer is more likely to find a car they love and conversion goes up. When conversion goes up, marketing dollars get more efficient. And as a result, our marketing algorithms allocate more dollars to these markets and more customers in these markets come to our site. With more cars closer to more customers, delivery times go down, shipping fees reduce, logistics efficiency goes up and conversion goes up again, restarting the loop as this caused us to grow inventory further. These simple data points clearly show all that positive feedback in action. It's only possible because of the machine we've built. And it drives our strategy and prioritization. Building this machine and the unmatched customer experiences it delivers is the key to our future success. And the bigger it is, the wider the moat. Our midterm goal is to build this machine to sell 3 million cars per year at 13.5% adjusted EBITDA margin by 2030 to 2035. When we announced this goal with our Q1 2025 results, we needed to grow to about 6x our scale in order to achieve it. Now 5 quarters later, we need to grow to under 4x our current scale in order to achieve it. The path is very clear, and there's a lot of execution to do. We have to keep building. We have to keep hiring. We have to keep training. We have to keep caring and we have to continually improve every part of the machine. We've been doing all those things for the last 13 years. We aren't going to stop. We are still just getting started. The march continues. Mark?

Thank you, Ernie, and thank you all for joining us today. Unless otherwise noted, all comparisons will be on a year-over-year basis. Q2 was another strong quarter, reflecting our team's continued focus on profitable growth and operational execution. We set new company records for retail units sold, revenue, gross profit, SG&A expense per retail unit sold, GAAP operating income and adjusted EBITDA. Retail units sold totaled 197,325 in Q2, an increase of 38% and a new company record. Revenue was $7.376 billion, an increase of 52% and a new company record. Revenue growth exceeded retail units sold growth primarily due to traditional gross revenue treatment for certain vehicles acquired from a large retail marketplace partner, higher industry-wide prices and a mix shift into newer and higher cost vehicles. The gross revenue treatment change will no longer affect year-over-year comparisons beginning in Q3, and we expect revenue growth to be more in line with retail unit growth in Q3. Consistent with past quarters, our growth in the second quarter was driven by our 3 long-term drivers of growth: a continuously improving customer offering, increasing awareness, understanding and trust and increasing inventory selection and other benefits of scale. Even beyond automotive retail, our growth continues to stand out. Our organic revenue growth in our most recent quarter ranks in the top 5% of S&P 500 companies, making us one of the fastest-growing large profitable companies across all industries. Second quarter marked our 10th consecutive quarter of industry-leading retail unit growth and adjusted EBITDA margin. Non-GAAP retail GPU decreased by $105 primarily driven by lapping the approximately $100 benefit from tariff-related effects last year. Non-vehicle costs were higher, primarily due to inbound transport fuel prices more than offset by higher retail appreciation. Non-GAAP wholesale GPU decreased by $158, driven by our 38% retail units sold growth outpacing wholesale gross profit. Non-GAAP other GPU decreased by $192, primarily driven by our decision to give back to customers in the form of lower interest rates as well as higher benchmark rates, partially offset by lower cost of funds, higher average amount finance and higher finance attach rates. Total GPU aligned with our expectations with a shift in allocation between retail and other components driven primarily by industry retail pricing dynamics and benchmark rate increases, respectively. Q2 was another strong quarter for levering SG&A expenses. Our 38% growth in retail units sold led to a $157 reduction in non-GAAP SG&A expense per retail unit sold, reflecting a $272 reduction in overhead expenses, partially offset by an $88 increase in operations expenses, primarily due to higher fuel prices. Advertising expense increased by $27 per retail unit sold as we continue to invest in building awareness, understanding and trust in our offering. We expect an increase in advertising expense dollars in 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 component of our cost structure. Net income was $513 million, an increase of $205 million. Net income margin was 7%, an increase from 6.4%. Adjusted EBITDA was a record $769 million, an increase of $168 million year-over-year, run rate exceeded $3 billion for the first time, marking another company milestone. Adjusted EBITDA margin was 10.4%, a decrease from 12.4%, primarily driven by increased retail revenue per unit resulting from the traditional gross revenue treatment mentioned previously. GAAP operating income was $680 million or 88% of adjusted EBITDA, an increase of $169 million and a new company record. As discussed in prior quarters, we continue to drive toward investment-grade quality credit ratios over time. In Q2, we again reduced our net debt to trailing 12-month adjusted EBITDA ratio to 1.0x, our strongest financial position ever. Our results in Q1 and Q2 position us well for a strong Q3 and Q4. Looking forward, we expect the following as long as the environment remains stable. One, a sequential increase in retail units sold in Q3 compared to Q2; and two, adjusted EBITDA of $2.7 billion to $3.0 billion for the full year 2026, an increase from $2.24 billion last year. In closing, we are excited about what our team accomplished in Q2. We remain focused on executing at a high level, delivering profitable growth and making steady progress toward our long-term 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] First question comes from the line of Daniela Haigian of Morgan Stanley.

For what it's worth, I do think the webcast is out, so you might get some questions there. First question is on what's the progress update on reconditioning operations specifically? That was something a lot of people were speaking about in the first half of the year. Can you disaggregate how much of the recent retail GPU dynamics is coming from gains there versus a supportive or favorable used car pricing environment?

Sure. So I think, first of all, we're working on the webcast. The team is aware of that. So thanks for the heads up. So I would say, yes, I think the team has done a great job. As we spoke last time, they worked really quickly in a couple of months, got costs back into a great place. I think that was step 1 of the plan. Step 2 of the plan was to return back to growth. Our inventory growth slowed a bit there as we were focused on costs. And I think around mid-Q2, the team started to get our kind of inventory growth moving back closer in line with sales growth. So I think that's great. And then step 3, we plan now to shift to cars that are closer to our traditional mix in terms of age of cars and mileage, et cetera. We've been moving up a little bit in car price into a little bit more near-aged cars as we've gone through over the last couple of months. So I think that's all going great and is right on track. I think the retail strength, I think, had the most -- the simplest and probably most dramatic part of that is during the quarter, the FTC put out guidance to all dealers that they were required to update their pricing to include dock fees and also any products that were required to be purchased with the car. For us, obviously, that has no impact. We haven't had dealer fees, and we don't have products that are required to be purchased with a car. But for many dealers, that was certainly a change. And as a result, over the subsequent several months, basically from April through June, many dealers were adjusting to that change, and then that was changing the underlying data that we were seeing that powers our pricing algorithms. We've been making those adjustments and kind of figuring out the implications of that. But as those changes were being made, we were following the market, and that led to some of the outsized gains. I think there's a couple of other puts and takes. But for the most part, I think things are operating as expected there.

And then my second question is on the financing piece. How do you think about the outlook on financing margins as you have changes in benchmark rates? How much of the risk is hedged? And how do you think about holding margin versus holding rates steady for consumers?

Yes. I would put that back in the context of the previous answer a little bit. So I think the way that we try to think about this is we're trying to build a big machine, and that machine delivers great experiences and it kicks off great unit economics. And I think this quarter gave a great example of the flexibility of that machine. So when we saw these price changes that were flowing through to higher retail GPUs, and then we also saw the increase in benchmark rates, we basically paused our response to the benchmark rates to make sure that we are understanding the sum of those 2 changes to have fewer moving pieces. And I think that that's the kind of flexibility that we have. But overall, I think total GPU when you add the 2 up, came in right as we would have expected. And I think, again, just kind of demonstrates the overall flexibility of the model.

Your next question comes from the line of John Colantuoni of Jefferies.

Just wanted to come back to the pricing growth rates dynamic. Can you talk to your strategy of keeping consumer-facing interest rates stable despite higher benchmark rates rather than maintaining retail prices? I'm curious if you're seeing a higher ROI on investments into rates rather than investments into pricing? And I have a follow-up after that.

Sure. I would, again go -- I think the way that we try to think about these things in general is make the machine as efficient as possible overall across all line items. And then we are separately trying to make as much progress as possible in making that machine better in the form of fundamental gains and foundational capabilities we've talked about in the past. I think other GPU is a great place to look at exactly those kind of fundamental gains and to see the types of choices that we're making. Year-over-year, I think other GPU, to some degree, as a result of the effects we just spoke about, was down just shy of $200. During that same period, we passed back over 100 basis points of rate to our customers. I think if you do the math on passing back rate to customers, it's probably a good estimate is maybe $4 to $5 per basis point that we pass back to our customers in rate. And so given all the rates that we pass back, all else constant, you probably would have expected something closer to a $500 reduction in other GPU, but we actually saw was $200. That's because there's, give or take, $300 of fundamental gains in there. And then we take those fundamental gains and we try to figure out what's the best thing to do for the business and our customers in the long run, and we've elected to pass those back over time as we think that that's the right thing to do on our path to 3 million to 13.5%.

Okay. Great. And second, with labor hours per unit approaching all-time best levels since April. I'm curious if you're expecting to see an incremental tailwind to retail GPU from this dynamic in the third quarter, given you don't record the lower reconditioning costs until the cars are actually sold.

I think that conceptually holds. And then I think we are also at a place where our hours per unit is in a very good spot. And I think the realistic variability in those rates aren't huge dollars. So they're not dollars that we would really want you to kind of take in one direction or the other. I think there's more noise than just building a machine of this complexity at the pace that we're building it than there is kind of, I think, certainty that any given move there will flow through to the bottom line in any given period. But I think there's clearly opportunity for us there over time, and we will seek to get it as quickly as we possibly can. But I think given the way the business is performing and the unit economics that kicks off, the most important thing we can do is just make sure our costs are in a really good spot and then continue to build the machine. So I think today, in step 2 and step 3 of that inventory plan we discussed earlier, we're really focused on making sure we get inventory growth back up. It's -- inventory has undergrown sales over the last several months, and that certainly creates a headwind to just like the overall business. The team has got a great plan, and we're confident they'll catch up and hopefully surpass it in the not-distant future. But we got to make sure we do that and execute. And that, I would say, is the primary objective today in that group.

Your next question comes from the line of Rajat Gupta of JPMorgan.

So Ernie, in the past, you have hinted that EBITDA per unit is an important metric that you care about and you know investors do as well. I mean the used car industry will continue to have these pricing fluctuations like the 2-year rate fluctuations. We've had 3 quarters of lower EBITDA per unit in a row. I'm curious if you can give us a sense of when investors should expect that to return to growth? Do you think it can happen, like, later this year? Is it a 2027 story? Any color on that would be helpful. And I have a quick follow-up.

Sure. I'm going to again go back to the framework of -- I think it's all part of the big machine that we're trying to build, and we're trying to make as much progress in both growth and EBITDA dollars per unit as we possibly can. I think -- when you look year-over-year, I think this quarter, we were down about $300 in EBITDA dollars per unit. Quarter-over-quarter, I think we are up about $300. If you break it down this quarter, I think there's some pretty clear year-over-year impact. Last year, we had a tariff benefit that was around $100. Gas prices moved up pretty dramatically throughout the quarter. That probably cost us something on the order of $75 across the entirety of the income statement. I think there were a couple of things that kind of flowed through that helped to explain that. Benchmark rates moving certainly didn't help us in the quarter. So I think those things are moving around a little bit. But I think we're extremely happy that we were able to deliver 38% growth, still be there with the math that's that clean. And we did at a time when our inventory wasn't growing as fast as we wish it were. I spoke in my prepared remarks about those graphs that kind of show the extremely strong relationship between inventory growth and sales growth in all these different regions. When you kind of extrapolate that to the company as a whole, that obviously has a large impact as well. And we clearly undergrew sales during the quarter. As I said, the team's got that back on track, and we're starting to catch up, but we have not yet caught up. That puts a headwind on the business that has to show up some way, either it's going to show up in lower sales or lower profitability overall, but that's just kind of, again, part of building that machine and balancing it. So I think the team is growing the machine, scaling it, adding fundamental gains that we just discussed in other GPU. We're on a great path. And I think we've got to march and execute. I think the hardest stuff that we have to do is make sure that we're executing across every different operational part of the business. And if we do that, the demand is clearly there as shows up in those graphs.

Understood. That's clear. Just a quick follow-up. So I think 2025 was a pretty clean year in terms of the seasonal cadence of the business. In the second half of '25, you actually grew EBITDA versus the first half despite a pretty challenging fourth quarter. The guidance would imply a step down this time despite the fact that you're actually now catching up to production better in the second quarter. You're at a more better run rate on production in the second half versus the first half. So just curious like what's driving that conservatism? Or is there anything that we are missing in terms of the dynamics between first half and second half relative to last year when it looks like things are actually getting better from an execution standpoint.

Yes. I think we got to stick with guidance. I think once we start giving guidance on guidance, I think it gets complicated. But I think -- listen, I think said this in the opening of the shareholder letter and also in my prepared remarks, the most important thing is execution. And I think the -- it is very likely clear. Again, I'm going to call back to those graphs. When you look at those graphs and you really think about the implications of those graphs, it's pretty clear if we build the machine and we deliver great experiences, the demand is available. If the demand is available, it will show up in some combination of growth and improving unit economics. That's just -- the demand has to express through that. So I think the question looking forward is generally it's about our ability to execute. And I think the team has done an incredible job executing. But I also think the execution is always uncertain as we look forward several months. And so we're always going to take that into account as we're looking forward. But we're confident we're executing really well. The question is how well do we build out the machine because the machine generates the demand and creates the unit economics.

Your next question comes from the line of Ron Josey of Citi.

Ernie, I wanted to go back to the execute on execute and specifically understand a little bit more about Roll Call and Leader Hub. And wondering, can you give us some insights on whether that's been fully rolled out across the IRCs? And is that what's needed for inventory to get to that level that the team needs to drive continued growth? That was question one. And then with all the comments on AI and the Sebastian being used more and more, just any insights on conversion rates given just greater usage of Sebastian?

Yes. Okay. So on the inventory plan, no, our new tools are not fully rolled out everywhere. And so I think that's certainly opportunity. I think the team has rolled out process improvements everywhere, even where the technology is not yet fully rolled out. And I think we're seeing very strong results. So you're seeing that in our HPU cost and in the fact that we started to grow inventory again in the middle of the quarter. So I think there's good stuff in front of us, and I think it's a function of how well we execute. And then I think, yes, the progress that we're seeing across the entire business, enabled by AI, whether it's just every product in the business being able to move more quickly or if it's customer experiences getting better and simpler, I think that's exceedingly clear. I think that may take the form of Sebastian, so it may happen in kind of chat or it may take the form of different kind of features that we've built throughout the website that either pull customers into Sebastien or provide information that Sebastien would have otherwise provided. I think a reasonable way to think about that is just to think about what is the customer care cost over the last several years and what's happening there? I think if the website is instantly intelligent and can answer every customer question on its own through whatever tool you put in front of them, then there would not be calls into customer care and our customer care expenses would be $0. If we just look at kind of what's happened over the last several years, 3 years ago, customer care costs went down 40% year-over-year. So 2 years ago, went down an additional 30% year-over-year. One year ago, went down 20% year-over-year. This year, we went down an additional 10% year-over-year. So I think the sum of the effects of those compounding gains is clearly showing up. And I think we've got a lot of great ideas and a lot of great work still in front of us that we need to make sure we execute on and unlock the fundamental gains that come with it.

Just a quick follow-up. You said not fully rolled out for Roll Call and Leader Hub. Any insights on like how close we are or where are we on penetration there or plan to make it fully rolled out?

Yes, we'll be rolling it out over the coming quarters.

Your next question comes from the line of Brian Nagel of Oppenheimer.

So the question I ask, maybe it's kind of basic, but just with regard to the guidance that was laid out for the balance of '26. So now you're providing at least a range of EBITDA guidance, that's new way. So the question I ask is, why issue the guidance now? And then probably more importantly, as you think about that guidance and the parameters around that guidance or maybe the thought process behind that guidance, are you signaling any type of change in the business from what we've seen here in the first half of the year, either from a sales or from a more importantly, a profit standpoint?

Sure. Let me take that one. So I think as a starting point, we're very excited about Q2 results, as Ernie pointed out, a record quarter across many dimensions. I think as we think about the guidance, it's actually the same style of guidance that we've applied the last 3 years. So typically, in the first half of the year, this is speaking back to, I think, '24, '25 time frame. We'll give some sequential color just to understand -- allow everybody to understand the seasonality and where we think the business is heading. And then when we reach midyear here, we actually give more specific guidance on adjusted EBITDA. The goal there is just to let people know some guardrails around what we're expecting in the second half. And that really is our philosophy. And again, basically the identical philosophy that we've applied the last couple of years. I will say that, overall, I think we're feeling very good about the trajectory of the business. I think, like I said, this is a very strong quarter, driving 38% retail unit growth. That is against an industry backdrop where the industry is down, call it, on the order of 4 points year-over-year. So I think the 38% growth is a notch more impressive in light of that industry backdrop. The other thing that I just get very excited about this quarter is the fact that in 2 large regions, representing approximately 1/3 of the country, we grew 54%, and we're growing at 54% in those very large regions despite the fact that at the company level, we're almost ticking over $30 billion annual revenue run rate. We've ticked over in Q2 over a $3 billion adjusted EBITDA run rate, more than $2 billion run rate of net income in the second quarter, which is a good quarter. But that just says we're at real scale here. And the fact that we're at this scale and level of profitability, and we've got major regions that are growing at 54% to me, just got me very, very fired up this quarter. I think why is that happening? It is building the machine. It's basically all the sources of positive feedback. So we grew selection in that region. That did allow customers to choose cars that were closer to them. In addition, as we had more cars, it made sense to market more in those regions. And as we marketed more, we drew more customers to the site who then converted on the cars that we had available. And it just serves as a really good example of how the whole model works and why it's so valuable for us to just continue to march down our execution path, continue to build this machine that involves ramping production, building last mile and multi-car logistics capacity to connect that production to consumers, continuing to drive the customer experiences. And I think one thing Ernie may have mentioned, but we had a great quarter on customer experience as well. We've seen that marching up as we've been growing at these levels. And so when we put all that together, it just -- it was a really, really great quarter, and it was a quarter where I think we just have data points that give us very high conviction about the growth trajectory that we're on and the long-term sustainability of that growth trajectory. So those would be some of my thoughts.

Your next question comes from the line of Sharon Zackfia of William Blair.

I had a GPU question given the degradation we've seen kind of over the past year, which I think has run between $200 to $150 a quarter. Do you expect that to narrow as we get into the third quarter? It seems like you might have still some other GPU compression, but perhaps the retail side is now getting better. I just would love to get some color on how you're thinking the third quarter might shape up there.

Sure. I think at a high level, we're going to stick with our guidance on this. I think we -- you can look at our results. There's a bit of seasonality in the different GPU line items that is kind of reasonable to assume could be somewhat similar in the future as to what it was in the past. But I think we're going to stick with our guidance and try to stay away from giving too much detailed line item by line item color.

And can I ask a follow-up? Just given the West and the Southeast are lagging in production increases. As you think about the ADESA conversions, if I remember correctly, part of the real estate advantage there was that there were quite a few ADESAs that were in kind of very favorable locations in the West, particularly in California. What does the slate look like for conversions geographically kind of over the next 18 months?

Yes. I think we've got opportunities all over the place. And I think the team has a very clear kind of build-out plan, and it works basically 2 ways. One way is where is the optimal place for us to put inventory on the map given where we have opportunity. And then one way is what are the inspection centers or regions where we've got the management teams that are executing at the highest level. And I think that we are constantly trying to balance those 2 things out. So I think, for example, the 2 regions where we grew inventory the most, the Midwest and the Northeast, the Northeast is definitely pretty heavily impacted by the ADESA footprint that we were able to open up. The Midwest is also impacted, but that was a place we already had some strength. I think looking forward, we plan to balance those 2 considerations, but all of it, we plan to open up. And then in addition, we also are just beginning work on a fresh build site as well. So the way the team determines where fresh builds will go is they look beyond 3 million and they say where are the gaps in the map that would be optimal for us to fill in. And then those are the sites that we look at there. So I think we'll continue to kind of run that -- I'd say the first way is basically conceptually. It's just like what is the best place in the map to grow inventory? And the second way is practically, where are we executing the best. And I think we'll try to continually balance those 2 considerations. One thing I want to add to Mark's previous comments because when Mark gets fired up, I get fired up, and I kind of think everyone in the whole company gets fired up. So I'm going to go back to those charts for a second. Not only is kind of the inventory growth driving sales, it really truly does drive the entire machine. In those same regions where we have more production and then we have more sales, we have more cars that we buy from customers. We have marketing that is more efficient than the rest of the country. We have profitability that looks very similar to the rest of the country. There's not variability in profitability across those markets. So anyway, I do want to keep pushing on that because I think we try to make this point that execution is the most important. And I think the evidence that the variability in growth rates around the country provided us is very helpful. I think the evidence has always been there across time, but to see it laid out so clearly in the same moment in time and to see it across the entire business playing out that clearly, we think is really exciting. And so that makes it, again, an execution story, and we just got to make sure we execute.

Your next question comes from the line of Andrew Boone of Citizens.

Ernie, I wanted to go back to the last comment that you just made about production and more inventory coming online and what that means. You guys made a decision earlier this year to reinvest a point back into financing costs. And so I guess just from a higher level, like why is that the right decision versus investing in labor or some other checkpoint that you guys have to drive more production? Like why did you guys make that? And then how do we think about what you're saying today and where you guys allocate investments and costs going forward?

Yes, I think that's a great question. So let me answer like the kind of labor versus anything else first. I think the reality is like the financial returns on growing the machine are extreme. It's not a financial question. Like if we could write a check and have the machine be bigger, it would be very straightforward that we would want to do that at extreme speed. It's much more about the execution of building out the facilities, hiring people, training people, making sure people execute well, the people care. That is the thing that I think is much harder, and it's the thing that unlocks much larger returns and more enduring returns. As a general matter, we are working to go as fast as we reasonably can there. I think occasionally, you will see bumps in the road. I think kind of Q4 into early Q1 in recon, we had some bumps in the road. And I think the kind of idealized execution would never have a bump, but I think real-world practical execution, I think the best you can hope for is that when you hit a bump, you recover quickly. And I think that team is recovering very quickly. So those execution type investments, we want to make as quickly as we possibly can, and it's just about making sure that we execute. I think that's the big question there. On financial kind of investments, I think that's a very good question that is, I think, interesting and points to at least an apparent contradiction a bit in some of the things that we're saying. So we've invested a bunch back in the customer offering, right? We've given fundamental gains back to customers. We've done that at the same time that we are constrained, right? We're showing you these graphs. We're constrained. We're telling you that inventory grew less than we wish. Why would you kind of give money back to customers in the face of those constraints? And the answer is that we expect to relieve those constraints, and we're trying to make sure that we build the business in the best way that we can over the long term. I think any time we do like customer-facing optimization, you can kind of do it 2 ways. You can view what are the starting economics that we've got and what are elasticities and what are the smart things that we should do to maximize the value to customers and to ourselves. If we were not constrained at all, we could press a button, get car and the whole system just kind of worked. And that's one way to kind of do the math. Another way you can do the math is you can say, okay, we're constrained. And as a result, all fundamental gains should just flow straight through to us. I think in '22 and '23, when we were making sure that we put ourselves in a spot where we were completely financially independent, it was very clear what choice we should make. I think where we are today, where we're extraordinarily financially independent, where we're generating enormous cash flow that is accruing on our balance sheet quarter after quarter, and we're compounding and growing. I think we are in a position that we're excited to be in where we feel like we can make longer-term choices there and do what's right for the customers. So it is likely the case that giving back to customers in the form of rates over the last year is not something that we've been fully paid back for in growth. The fact that our inventory is tight sort of makes it clear that we didn't get fully paid back for in growth because if inventory gets tight, your conversion rates go down. So we probably didn't get fully paid for it. But it puts pressure on the machines. It will cause us to build. And then we got to make sure that we execute, and we're in a great spot anyway. So we're going to try to do what's right for us and our customers over the long run.

And then just as my follow-up question, I wanted to ask about the cash generation. You guys are now at 1x kind of leverage. What should we be thinking about or considering before you guys start to think about capital returns or some other form of return of cash...

Sure. Well, I think the most important things that I think of when I think about what to do with our cash, it really comes down to investing in the business and generating returns and building the machine for the long term, just to give an example of that. So our trailing 12-month operating income was just over $2.2 billion. We generated that $2.2 billion of operating income on only about $7.5 billion of net operating assets. And so we're talking about on the order of 30% operating return on net operating assets, which is a pretty great business and a business that you really want to invest in. And so that's a little bit of a different angle on some of the things that we're excited about in the business, including our growth, including our customer experiences, including our industry-leading margins. But the way that we're now earning returns on the capital that we invest in building the machine, I think, is also very exciting. And so first and foremost, where our head goes on, hey, what do you want to do with the cash that you're generating? It's invest in this machine that we're building that delivers great customer experiences and has shown the capability to grow very quickly in a sustained way over a long period of time. So those are my major thoughts on that.

Your next question comes from the line of Jeff Lick of Stephens.