Report contents
Industry landscape
Where the company competes: thousands of sellers, one national delivery machine
Carvana is a used-car retailer that buys, reconditions and sells vehicles online, then delivers them to customers' homes. It sits in the middle of a long chain. Cars come from consumers selling directly to Carvana and from wholesale auctions run by Manheim, ADESA and others; they pass through reconditioning; they reach buyers who usually finance the purchase. Carvana originates those loans and sells most of them into the asset-backed market. It also runs a wholesale channel, disposing of trade-ins and aged inventory at auction, and sells ancillary products such as vehicle service contracts and GAP waiver coverage (a policy that pays the gap between an insurance payout and the loan balance if a car is totaled).
The core market is US retail used-vehicle sales: a used car, plus financing, sold to a consumer. Almost all of Carvana's revenue comes from selling the metal; wholesale and ancillary products supply the rest. That definition matters because it excludes the businesses that make the industry's economics look better than they are. Wholesale auctions earn fees on other dealers' cars, and franchise dealers earn service and parts margin after the sale. Carvana earns the spread between what it pays for a car and what it can sell it for, minus reconditioning, transport and the cost of customer acquisition.
Who captures the margin. The product is close to homogeneous and the price is public: every listing carries a number, and shoppers compare it on Autotrader and CarGurus in minutes. A retailer therefore has little room to charge above the market for the same car. Upstream bargaining power is also limited, because auction supply is dispersed, though buying directly from consumers gives Carvana a lower-cost source and a car it can inspect before it commits. Downstream, the buyer's monthly payment matters more than the sticker, so the financing rate and the ancillary attach carry much of the profit. That is why the profit pool in this submarket sits with whoever can source, recondition and turn inventory cheaply, and fund the resulting loans at a low rate. It is a wholesale-to-retail spread business with thin percentage margins on a large ticket.
Geography: national inventory, local delivery. Carvana sells only in the United States. Inventory is priced and marketed nationally, but delivery is not: cars ship free inside a radius and carry a fee outside it, and title and registration rules vary by state. So the reach of the model is set by the physical network of inspection and reconditioning centers, not by the website. The 2022 purchase of ADESA's US auction sites gave Carvana a set of locations it has been converting into reconditioning capacity, which is also where its wholesale sales happen. Scale, in other words, is national in purchasing and advertising and regional in logistics.
Market structure: fragmented, with a small online fringe. Measured against the whole US used-vehicle market, which runs in the tens of millions of transactions a year, Carvana is small: it retailed 596,641 vehicles in 2025, the company reported in its annual report. That is a low single-digit share. The sellers around it fall into four groups.
- CarMax, the largest used-only chain, with a national store footprint and its own origination platform.
- Franchised dealers such as AutoNation, Lithia and Group 1, which control the trade-in funnel and sell millions of used units through their lots.
- Independent used dealers, numbering in the tens of thousands, which dominate locally on price and carry high funding costs.
- Private-party listings on Facebook Marketplace and Craigslist, which take no intermediary margin at all.
The online-only field has thinned. Vroom shut its e-commerce operation and Shift filed for bankruptcy in 2023, leaving Carvana, CarMax and Sonic's EchoPark as the main national online sellers.
Barriers: low to sell a car, high to do it nationally. Apply the counting test and the answer is clear: there are far more than five meaningful competitors, so entry at the dealer level is cheap and no participant sets the market price. The narrower arena Carvana competes in, national online retail with in-house reconditioning and next-day delivery, is a different matter. It requires capital for inventory, hundreds of acres of reconditioning space, and a labor force of roughly 25,000, and it requires the operational discipline to keep a car's days in inventory short while promising a delivery date. Two well-funded challengers tried that and failed, which says something about execution difficulty without proving the position is unassailable.
The structure points to a specific question for the moat: if the product is priced publicly and nobody can charge above the market, Carvana's returns have to come from buying, reconditioning and turning cars more cheaply than independent lots and franchise dealers can, and from attaching finance at a competitive rate. Whether that cost gap is wide enough to be durable is what the next section tests.
Competitive advantage
Does the company have a moat? A cost edge, not customer captivity
Overall rating: Moderate. Carvana's returns look like a moat. Over the four quarters to September 30, 2026, $2.39 billion of operating income on $27.5 billion of revenue, both summed from the company's quarterly reports, supports an estimated after-tax return on invested capital of 16% to 29%, depending on whether trade payables count as capital. What defends those returns is a per-car cost and inventory-turn advantage, not a hold on the buyer. The third quarter shows the difference. Retail units rose 39% to 217,385 while gross profit per unit fell $365 to $6,997, the company reported in its third-quarter shareholder letter. Handing back roughly $79 million of gross profit in a single quarter (365 × 217,385, estimated) to keep volume growing is what an operator without pricing power does.
The used-vehicle market fails the concentration test outright: tens of thousands of independent lots and franchised dealers sell the same models, and nobody sets the price of a five-year-old Camry. Excess returns in a market that fragmented have to come from cost. Carvana has now posted 10 straight quarters of margins around twice the industry average, by its own account in that letter. Whether that gap is a barrier or a head start is the question this section has to settle.
Market position: about 2% of a market nobody controls
Market position: Moderate. Management put the Q3 2026 run rate of 870,000 units at roughly 2% of the used market and 1.5% of all vehicles sold. Running that 2% claim backwards implies a used market near 43.5 million units a year and puts 2025's 596,641 retail sales close to 1.4%, an estimate built on management's figures rather than an independent count. Carvana is not the largest used-car seller in the country; franchised dealer groups and CarMax sell at national scale. It is one of three national online sellers left after Vroom exited e-commerce and Shift filed for bankruptcy in 2023, and the fastest-growing of them.
Stability: eight quarters of gains after a 21% retreat
Stability of position: Moderate. Eight consecutive quarters of roughly 40% unit growth against an industry the company describes as flat is documented share gain, and Carvana has doubled in two years. The series is not monotonic. Revenue fell from $13.60 billion in FY2022 to $10.77 billion in FY2023 while inventory was cut from $1.88 billion to $1.15 billion, and the present expansion is two years old. Two years of gains from a 2% base cannot separate a durable cost advantage from a well-funded push. What the record does establish is that the volume line moves with management's own decisions as much as with competitive wins.
Profitability: high returns on a two-year record
Profitability: Fairly Strong. The company's statements show the swing clearly.
| USD bn | FY2021 | FY2022 | FY2023 | FY2024 | FY2025 | TTM Sep 2026 |
|---|---|---|---|---|---|---|
| Revenue | 12.81 | 13.60 | 10.77 | 13.67 | 20.32 | 27.45 |
| Operating income | (0.10) | (1.49) | (0.08) | 1.00 | 1.88 | 2.39 |
| Net income | (0.14) | (1.59) | 0.45 | 0.21 | 1.41 | 1.71 |
The TTM column sums the four reported quarters. FY2023's positive net income arrived in a year with an operating loss, so it says little about the retail business. Return on equity of 59.9% in FY2025 is an artifact of a $3.44 billion equity base and a fourth quarter whose net income of $0.86 billion exceeded the first three quarters combined. The cleaner test is the ROIC estimate: NOPAT of about $1.89 billion (operating income taxed at the 21% federal statutory rate) against invested capital of $6.5 billion — equity of $4.35 billion plus debt of $5.22 billion less cash of $3.05 billion — gives 29%. Count all operating assets net of cash, $12.1 billion, and it falls to 16%. Both clear the 15% lower bound for an advantaged business; neither is a long average.
Margins: twice the industry, and narrowing
Margins: Fairly Strong. Gross margin was 18.9% in Q3 2026 against 20.3% a year earlier, and the adjusted EBITDA margin slipped 1.3 percentage points to 10.0%. Those absolute levels are unusual for car retailing and imply real scale spreading fixed costs. The direction is the wrong one: revenue per unit rose about 2% while gross profit per unit fell 5%, which means the cost of putting a car on the lot and getting it to the door rose faster than the price it fetched.
Overall rating: Moderate, capped by duration and the lack of a captive customer
Overall rating: Moderate. Profitability is the strongest dimension and carries the most weight, which keeps the rating above Weak. Two things hold it below Fairly Strong. The high returns are two years old rather than ten, and they rest on a cost position in a market with cheap entry, so they have to be re-earned car by car, quarter by quarter. Second, the one stretch when Carvana stopped growing, 2023, produced an operating loss. Scale without customer captivity is the weakest of the durable advantages: an entrant with capital can build the same centers and buy the same shoppers, which is roughly where CarMax and the franchise groups already stand.
Where the advantage comes from and how long it can last: reconditioning density sets cost per car
The company's own operating disclosure carries the weight here. Independent share counts for used-car retail lag by a year or more and cover a market too fragmented to measure quarter by quarter, so the per-unit economics in the shareholder letter and annual statements are the best available evidence. Three sources of advantage are worth testing: cost per reconditioned car, whether the buyer is held, and whether fixed costs spread with volume. One number tests all three — gross profit per unit — because a scale advantage has to show up as more profit per car as volume rises.
Cost advantage: a real per-car gap that is narrowing
Cost advantage: Moderate. Carvana buys at auction and directly from consumers, reconditions in its own inspection centers, hauls on its own logistics network and delivers to the door. Cost per car falls as a center runs closer to capacity and as delivery routes in a metro thicken; a truck dropping three cars on one loop costs less per car than one dropping a single car. The 2022 purchase of ADESA's US auction sites handed it buildings to convert, and the conversion is why the fixed-cost story has any teeth.
What the position is worth today: adjusted EBITDA of $800 million across 217,385 units is about $3,680 a car (estimated). If management's claim of roughly twice the industry margin holds, the 5-point gap on $8.04 billion of quarterly revenue is about $400 million a quarter, near $1.6 billion annualized (estimated). That is the prize, and it explains why two funded challengers tried to copy the model.
The protection is structural only in part. Buildings, trucks and a reconditioning process take capital and several years to replicate, and Vroom and Shift ran out of money trying. But the buyer can leave in a click, and a rival with stores and a trade-in funnel can build the same local density. The adverse fact is in the most recent quarter: cost of sales per car rose about 4.2% while revenue per car rose 2.4%. If that repeats through 2027, the gap is being spent on volume rather than defended by cost, and the advantage is an operating edge that is eroding as it scales.
Customer captivity: shoppers compare prices in minutes
Customer captivity: Weak. The used-car buyer is close to the opposite of a captive. Purchases come every five to seven years, the listing price is public, and aggregators put the same vehicle side by side within minutes. There is no retraining cost, no data lock-in and no contractual tie; the loan does not hold the customer either, because Carvana originates the paper and sells most of it, so the relationship ends at delivery.
Carvana's own evidence of improvement — a net promoter score up 9 points in two years, twice the selection, delivery two days faster — describes a better shopping experience. That wins the current transaction and creates no barrier to the next one, because the customer has to choose Carvana again years later. Satisfaction is not captivity, and the company does not publish repeat-purchase rates, so the strength of any habit cannot be measured directly. What would change the assessment is a service and trade-in cycle that keeps a buyer inside the network for the life of the car. So far the evidence only shows that people like buying from Carvana more than they did two years ago.
Scale: fixed costs spread only as far as the delivery radius
Economies of scale: Moderate. The fixed costs are heavy: reconditioning centers, trucks, the technology stack, national advertising and the funding program that places loans into the asset-backed market. Minimum efficient scale is local. A center needs enough cars to run two shifts and a route needs density, so the logistics cost that decides per-car profit is set region by region, while purchasing and advertising spread nationally across 217,385 quarterly units. Management's mid-term goal of 3 million units a year at a 12.5% operating margin between 2030 and 2035 would be about 7% of the used market on its own 43.5-million denominator, roughly 3.5 times the current run rate. Reaching it would push fixed costs per car down and widen the advantage. Missing it while gross profit per unit keeps sliding would mean scale is buying revenue, not cost.
Scale is a barrier only when it is reinforced by captivity, and here it is not. A competitor with capital — CarMax, or a franchise group with a trade-in funnel and cheaper access to inventory — can build the same density and buy the same customers. The two failures in 2023 show the execution is hard. They do not show the position is safe.
Durability: the edge breaks if cost per car keeps rising
Durability: Moderate, over a three-to-five-year horizon. Substitutes are plentiful and cheap. Private-party listings on Facebook Marketplace and Craigslist carry no intermediary margin, and franchised dealers' certified pre-owned programs bundle a warranty with a local service relationship. If Carvana pushes price to defend gross profit per unit, those alternatives take the sale.
The number to follow is the spread between cost per car and revenue per car. In Q3 2026 it moved the wrong way by roughly two percentage points. A second year of that would show the per-car advantage being consumed by price and mix, and the twice-the-industry margin claim would go with it. The open question is whether the current squeeze is the price of building selection and delivery density — a temporary cost of a wider network — or the permanent condition of selling a commodity that anyone can list.
Generated by DeepVal AI. Data and models may contain errors or be outdated. For research only; not personalized investment advice.
Continue reading the full report
- Comparison and summaryHow Carvana Co compares with its closest rivals, and our moat rating
Free to sign up · 20 research points · unlocking this report uses 10
20 points cover 2 full reports. No card needed.
Research reflects its generation date. Preview the first two modules.
More on Carvana Co
Carvana Co
Business, moat and key questions
Open company page