Investment Memorandum · Preview
For informational purposes only. Not investment advice.
Carvana Co.
CVNA
May 27, 2026
Carvana Co. (NYSE: CVNA) is a fully online used vehicle marketplace and retailer, founded in 2012 and headquartered in Tempe, Arizona. The company sources, inspects, reconditions, finances, and delivers used vehicles entirely through its digital platform — eliminating the physical dealership layer and enabling a ~20-minute end-to-end purchase. Carvana monetizes via three streams: used vehicle retail sales (~70% of revenue), wholesale vehicle sales (~20%), and financial/insurance income — gain-on-sale of originated auto loans, warranty commissions, and GAP waivers (~10%). The $2.2B acquisition of ADESA's U.S. physical auction network in 2022 gave Carvana 56 reconditioning sites (4,000+ acres, 2M+ unit annual capacity) that now form the physical backbone of its cost advantage. FY2025 revenue was $20.3B (+49% YoY); retail units were 596,641 (+43% YoY); adjusted EBITDA was ~$2.24B (11.0% margin). Q1 2026 set company records across all operational metrics.
▲ Bull Case
- ◆ADESA reconditioning leverage creates nonlinear EBITDA margin acceleration. The 56-site network is running at ~30% of its 2M+ unit capacity. As units scale from 600K toward 1.5M on the same fixed infrastructure, reconditioning cost per unit falls nonlinearly. Bull-case projection has adj. EBITDA margins reaching 16–17% by FY2030 — above management's 13.5% target — driven by this operating leverage.
- ◆CarMax is in structural decline and no credible digital competitor remains. CarMax lost ~8% of unit volume in FY2025 while Carvana grew 43%. Every purely digital competitor (Vroom, Shift) has failed, validating that Carvana's ADESA-backed model is not easily replicable. Winner-takes-most dynamics in online auto retail are playing out.
- ◆Tariff-driven demand shift from new to used vehicles provides a multi-year structural tailwind. 2025–2026 auto tariffs (25% on imported vehicles, 25% on parts) have elevated new car prices and pushed consumers toward used vehicles. If tariff policy persists through 2027+, Carvana's addressable demand is structurally elevated — a TAM expansion not fully embedded in consensus estimates.
▼ Bear Case
- ◆The SEC investigation is an unresolved binary risk with catastrophic downside. Hindenburg Research's January 2025 report alleges ~$800M in auto loan sales to an undisclosed related party (suspected DriveTime affiliate), inflating gain-on-sale income and reported profitability. If true, restatements could reverse FY2023–2025 net income, trigger debt covenants, force management change, and generate securities fraud litigation — destroying 70–80% of equity value.
- ◆GPU appears to be approaching a structural ceiling. Q4 2025 sequential GPU decline of ~$255/unit (non-GAAP retail) and Q1 2026 GPU below prior-year levels suggest reconditioning cost inflation — rising labor costs, complexity of EV and ADAS-equipped vehicles, and logistics cost normalization — may limit the unit economics expansion that has driven the turnaround.
- ◆Extreme valuation leaves no margin of error. At ~82x trailing FCF and 27x forward EV/EBITDA, even modest operational disappointments — one quarter of sub-target GPU, a rate-driven demand slowdown, an SEC subpoena disclosure — would trigger severe multiple compression. Reverse DCF implies the market is pricing ~22–25% annual EBITDA growth for 10 years.
“The bull-bear debate in May 2026 centers on three specific issues: (1) GPU sustainability — Is the Q4 2025 / Q1 2026 GPU softness a one-quarter anomaly or the beginning of a structural ceiling? Bulls argue reconditioning scale savings offset rising complexity costs; bears argue cost inflation is structural. (2) SEC investigation probability weighting — The 78% Buy consensus implies analysts assign ~5–10% probability to adverse outcome; we assign 10–15%, a $0.70 expected value difference per share. (3) Multiple expansion vs. compression — At ~27x FY2026E EV/EBITDA, CVNA is priced closer to SaaS than retail. Bulls require sustained or expanded multiple as EBITDA grows to $5–6B; bears see compression to ~15–18x as growth decelerates. Consensus: 78% Buy, 22% Hold, 0% Sell with mean target ~$87–$96, implying 23–37% upside, likely established before Q4 2025 / Q1 2026 GPU deceleration signals.”
- ◆SEC investigation closure — no action (removes ~$14/share risk discount)
- ◆Q2/Q3 2026 adjusted EBITDA margin >12% (confirms nonlinear margin expansion thesis)
- ◆Q2 2026 GPU recovery ≥$6,900/unit (validates GPU expansion thesis not structurally broken)
- ◆FY2026 unit guidance raised mid-year (signals sustained volume momentum)
- ◆Capital return announcement via buyback or debt prepayment (confirms financial health and discipline)
- ◆SEC adverse finding or enforcement action (trigger immediate exit protocol)
- ◆GPU sustained below $6,500/unit for two consecutive quarters (signals reconditioning cost inflation is structural)
- ◆SEC adverse finding / enforcement (10–15% probability, -70 to -90% equity impact, thesis kill-switch)
- ◆Consumer affordability squeeze from elevated rates (40% probability, -15–25% GPU/volume impact)
- ◆Used car price correction per Manheim indices (25% probability, -$1–2K GPU impact)
- ◆Recession / demand destruction for high-ticket deferrable purchase (20% probability, severe impact)
- ◆GPU ceiling structural due to reconditioning cost inflation (25% probability, moderate impact)
- ◆Amazon or well-funded entrant into automotive e-commerce (10% probability, severe long-term impact)
- ◆Tariff policy reversal eliminating new-to-used demand shift (30% probability, low-moderate impact)
Full Memo Continues
5 more sections, locked
- ●Valuation Range & DCFBase/bull/bear fair-value range, WACC, terminal growth, sensitivity to revenue + margin assumptions.
- ●Risk/Reward AssessmentPosition-sizing framework with explicit upside/downside skew and entry conditions.
- ●Management & Capital AllocationMulti-year capital-allocation track record, incentive alignment, and management readout.
- ●Monitoring FrameworkWhat to watch each quarter — leading indicators and inflection signals tracked by the analyst.
- ●Unresolved QuestionsOpen analyst questions and follow-up research items — the depth signal.
For Agents — $2 per memo
Call the JSON API with a Stripe Shared Payment Token. No account, no signup — just pay and call.
GET /api/v1/research/CVNA/memo Authorization: Bearer spt_...
Fund managers — coverage subscriptions launching soon. See marginofinsight.com.