Locked-Up Capital

At the risk of attracting some hate coming my way for speaking positively about $CVNA, I’ll give this a shot. Value Proposition. They hit on multiple dimensions of value that can impact consumer preferences, but the big ones are: (a) greater selection with dependable standards, (b) an efficient, no-negotiation e-commerce experience with home pickup/delivery if desired, (c) great value (for buyers and sellers). Any of these alone could swing customer preferences and market share their way, but together they can offer a disruptive experience for buyers and sellers that are willing to work within the e-commerce modality. Of course, I’m aware that there are some unhappy customers that have had a bad experience with Carvana (as with any business that has done millions of transactions), but there are many more happy customers that generate significant referral volume, which generates nice positive feedback in the business. Economics. Like other dealers, $CVNA acquires inventory at or around prevailing wholesale prices and makes a spread on each unit. Let’s walk through how they make that spread. Rough numbers, let’s say they buy a car for $20,000 and it takes them 50 days to move it, recondition it, and sell it. It might cost them $500 of depreciation during their holding period and $1,000 to recondition the unit. Then they may sell that unit for $25,000, making $3,500 of retail gross profit (“metal margin”). They generate additional profit by originating loans and selling ancillary services (about $3,000/unit, these are attached to a unit sold), and by selling other cars in the wholesale market (about $900/unit – these are unattached to a unit sold included in their definition of GPU). In aggregate, last quarter they generated over $7,400 in gross profit per unit sold (GPU), including some exceptional benefits. Variable costs per unit are about $2,300. This suggests about $5,000 of normalized incremental EBITDA per unit, excluding exceptional benefits. I hear a lot of pushback around Carvana’s GPU because it’s so much higher than others’ in the industry. There are some definitional differences (e.g. logistics costs aren’t included in CVNA’s, it is in KMX’s; Carvana includes wholesale GPU, etc.), but the primary reason Carvana is more profitable is because it is vertically integrated and purpose-built for scale across the network. It takes some time to discern, but you can figure it out within a pretty reasonable range if you study it. Balance Sheet. There are a few areas on the balance sheet worth highlighting. (1) Like other dealers, they have a floor plan facility that can be used to acquire inventory. They have $1.5B of capacity but used only $76M as of Q3 2024, meaning they are funding volume primarily through their own cash flow. They could raise cash levels quickly if desired, but it would be sub-optimal to pay the ~8.5% interest rates if they don’t have to. (2) As mentioned above, they originate consumer loans on about 80% of their unit sales. They sell most of those loans to other institutions. Sometimes individual loans are sold to Ally or other counterparties, sometimes they are bundled together into prime and non-prime securitizations like the 2024-P4 prime securitization that was recently marketed. They generate income from a combination of interest income on the loans they hold to maturity and/or hold before selling to counterparties, and by selling loans that they hold at a premium. (3) They have <$6B of holdco net debt, the majority of which was restructured in 2H 2023 and is trading well above par. Now that they are generating positive FCF and growing >50%, their cost of capital should decline rapidly. Valuation. Apply the incremental economics above ($5,000 of incremental EBITDA per unit) and pick a growth rate to get a sense for valuation. Let’s walk through an illustrative example. Let’s say they’ll do about 420k units in 2024 and $1.4B of EBITDA. To make the math easy, if they add +200k units in 2025 at $5k of incremental EBITDA/unit, they’d do $2.4B of EBITDA excluding any additional improvements in “fundamental gains.” If they grow by another 200k again in 2026, that’s $3.4B of EBITDA. With the cash flow generation improving the balance sheet between here and there, it’s trading at about 15x EBITDA whenever they reach 820k units (in this illustrative example, in 2026), and maybe cheaper than that if they keep extracting efficiencies from fundamental gains. Recall that they have asset capacity >1.4M units which can be expanded to ~3M+ for relatively cheap within the ADESA real estate footprint. Let’s say it will cost $1.5B to expand by another 1.5M units. $5k of incremental EBITDA x 1.5m units = $7.5B of incremental EBITDA – a no brainer.