Mercedes inside Nissan inside Carvana: the 27bp prime auto stack tells you everything about where risk is priced.

Three prime auto AAAs cleared this week inside a 27bp band — Mercedes-Benz 2026-1 A-3 at I+41, Nissan 2026-A A-3 at I+44, and Carvana 2026-P2 A-3 at I+68 — all at roughly identical 2.3-year WALs. The captive-versus-Carvana basis is now 24-27bp, which is tight by any post-2022 standard and suggests the market is no longer demanding a meaningful 'Carvana-the-issuer' premium so much as a 'Carvana-the-collateral' premium. With the 48-month auto loan rate down 35bp on the week to 7.36%, the consumer carrying cost is finally moving in originators' favor for the first sustained stretch this cycle.

Card paper showed up in size: Chase printed $1.25B and FNB another $500M, both AAA at 2.97 WAL, against a backdrop where credit-card charge-offs collapsed 62bp to 3.84% and 30+ DQs ticked down 14bp to 2.92%. That charge-off print is the cleanest tape we've seen in roughly a year, and it lines up with what the loan-level data is whispering: the prime/non-prime FICO gap has compressed to 132 points, 5.2σ below the 24-month mean of 137. Non-prime pools are not deteriorating — prime pools are letting them catch up, because originators have been adding cleaner non-prime borrowers at the margin.

Position: the relative-value trade is no longer captive-vs-deep-subprime in autos — it's mezz card vs. mezz auto, where card fundamentals are improving faster than spreads are tightening. If charge-offs hold sub-4% through the June prints, the BBB card stack looks mispriced against a BBB Carvana tranche carrying materially more loss volatility. Watch next week's subprime auto shelf — if DRIVE or AMCAR clear inside +120 at the 2-year, the FICO-gap compression story stops being a curiosity and starts being a spread thesis.