Carvana Co.

The business quality improvement here is earned and real — nobody hands you a fifty-percent ROIC improvement by accident, and the reconditioning center network represents a decade of capital and organizational learning that cannot be replicated with a website and a press release. The problem is that the market has already capitalized every dollar of that improvement and then kept going. You are paying for a business that has proven it can execute through a near-death experience, and then paying a second premium on top of that for the next ten years of compounding, leaving zero margin for error in a business where the two most critical variables — used-car prices and consumer credit appetite — are entirely outside management's control. The trajectory is genuinely compelling. A market share in the low single digits inside a fragmented, massive market, with fixed-cost infrastructure already built and beginning to leverage, is a rare setup. The thirty-percent of customers completing purchases without speaking to anyone and the NPS improvement signal that the product is getting better faster than competitors can catch up. If volume compounds toward the three-million-unit target, the EBITDA margin expansion math is credible, and the operating returns would justify a substantial business. The single biggest specific risk is gross profit per unit compression triggered by two simultaneous forces: a normalization of used-car prices from post-pandemic elevated levels, and a widening of credit spreads on the loans Carvana originates and sells. Both are cyclical phenomena that tend to arrive together in a downturn. In a high-fixed-cost model where the reconditioning centers and logistics network run regardless of volume, a GPU compression of even modest magnitude flows almost entirely to the operating line — and today's valuation is built on assumptions that current GPU is not just sustainable but improvable. That's the thesis-killer hiding inside every bull case.