Carvana's Machine Hits Record Scale — But the Inventory Constraint Is the Real Stall Point

Q2 delivered a record 197k units and a $3B+ EBITDA run rate, yet management's own regional data shows the recon recovery now leans on a Step-3 inventory catch-up.

CVNA · Earnings Call · 2026-07-29

A machine at record throughput

Carvana ended Q2 2026 having just digested the reconditioning cost spike that marred Q4 2025 and early Q1. The comeback was emphatic: 197,325 retail units (+38% yoy), revenue of $7.376B (+52%), and adjusted EBITDA of $769M — a $3B+ annualized run rate for the first time, with GAAP operating income converting at 88% of adjusted EBITDA. The company's framing has shifted to the "machine" — a deliberate, slightly self-aware vocabulary that management is using to sell the scale story. The core pitch: at just 2% of the U.S. used-car market (1.5% of total auto retail), the opportunity is "exceedingly clear," and the path to the 3-million-unit / 13.5% EBITDA margin goal now needs "under 4x" current scale versus "6x" five quarters ago. The most startling evidence is the regional production split Ernie Garcia presented:

The 2 regions where we added the most production capacity, the Midwest and the Northeast, grew inventory by 57%. In those markets, sales grew in the second quarter by 54%. In the 2 regions where we added the least incremental production capacity over the last year, the West and the Southeast, we grew inventory by 17%. In those regions, sales grew by 30%.
Ernest Garcia, Chief Executive Officer · 2026-07-29

That correlation is the positive feedback loop the company has teased for years, now shown as clean regional data. More inventory growth begets better conversion, cheaper marketing, lower shipping fees, and faster delivery — each feeding the next. Mark Jenkins called the 54% growth in regions covering ~1/3 of the country "a notch more impressive" given an industry backdrop down ~4 points.

The recon recovery and the Step-3 pivot

The reconditioning story tracks a three-step plan: Step 1, drive costs down — done, with hours per unit back to all-time-best levels; Step 2, return to sales growth — done mid-Q2; Step 3, "shift to cars that are closer to our traditional mix in terms of age of cars and mileage," meaning a deliberate pullback from the near-new, higher-priced vehicles used to get recon volume up fast. The result: ASP crept above $28,000, but management insists this is not structural — "we very much do not view that as structural" — and expects metrics to move back "toward the average car buyer." That mix shift has a notable data point attached: customers earning over $100k grew 60%+ yoy in Q2, which management frames as evidence of untapped demand rather than a ceiling. “If the cars were there for that customer segment, the growth is there... if we can build the cars that customers want... the demand is there and the economics are there.” — Ernest Garcia, Chief Executive Officer · 2026-07-29 Prior quarters confirm the pattern — in April, Ernie noted the recon team "took it very personally when we did not have a perfect fourth quarter," framing the new tools as foundational. “The new tools that were discussed are net new tools, and those are tools that we hope will drive additional fundamental gains over time.” — Ernie Garcia, Chief Executive Officer (CEO) · 2026-04-29 He also flagged the inventory tension, saying “Last quarter, inventory was up approximately 40% year over year. This quarter, it was up a little over 30%... That directional change is correct and implies our turn times have gotten a bit faster.” — Ernie Garcia, Chief Executive Officer (CEO) · 2026-04-29 The Q2 call goes further, admitting those tools — Roll Call and Leader Hub — are "not fully rolled out everywhere" and will take "the coming quarters" to deploy. That is honest but also a tell: the recon gains have not fully scaled. One more environmental wrinkle: the FTC's mid-quarter guidance forcing dealers to include doc fees and required add-ons in advertised pricing. Carvana has no dealer fees, but its pricing algorithms had to adapt as competitor listings shifted — a temporary distortion that management believes becomes a tailwind once settled. “For most dealers, let's say that their prices on their website just went up $500 or $600 and ours didn't move. That should be a tailwind in the grand scheme of things once all that settles out.” — Ernest Garcia, Chief Executive Officer · 2026-07-29

Financing: passing back the rate, keeping the GPU math

A different kind of flexibility showed up in other GPU. Benchmark rates rose, and Carvana chose to hold customer rates steady — passing back over 100 bps of rate. The math, per Ernie: that should have sliced other GPU by ~$500/unit; the actual decline was ~$192, implying ~$300 of fundamental gains in the finance platform — better credit models, higher attach, longer duration. “Given all the rates that we pass back, all else constant, you probably would have expected something closer to a $500 reduction in other GPU, but we actually saw was $200... about $300 of fundamental gains.” — Ernest Garcia, Chief Executive Officer · 2026-07-29 This is the company's recurring playbook: unlock gains, pass them to customers, wait for growth to compound. It is also the logic behind the first full-year adjusted EBITDA guidance of $2.7–3.0B, up from $2.24B a year ago. Mark Jenkins was careful to frame this as the same cadence used in Q3 2024 and Q3 2025 — "the same playbook" — not a new disclosure philosophy, and he pushed back on the idea that second-half deceleration implies any change in trajectory.

Why it matters: the growth-quality question

Two things give pause. First, the free-cash-flow picture: trailing-12-month operating income is $2.2B on $7.5B of net operating assets (~30% return), but free cash flow (less SBC) collapsed to $30M in Q1 2026, down 84% yoy and 92% qoq. That is the cost of stepping up inventory into the recon fleet — and it ties directly to management's admission that inventory "undergrew sales" for several months. Second, the stock sits ~27% below its January 2026 peak of $95.69, with the recent 90-day tape essentially flat (+3.9%) — the market is not yet pricing a re-acceleration. The Step 3 pivot back toward the average car buyer, and the benchmark rates environment, are the swing factors. If inventory catches up and unit growth re-accelerates without giving back recon gains, the "machine" story gets its confirmation; if the catch-up stalls, the ~$300 fundamental gains in finance become the only thing holding EBITDA per unit up.

Verdict

This is a capability reset, not a broken quarter. The recon tooling is real, the finance platform is compounding, and the scale math is improving. But the inventory-vs-sales lag is the live constraint, and Q3 sequential unit growth is the test. Watch whether the regional data — Midwest/Northeast at +54% — converges company-wide as Step 3 normalizes the mix. That is the moment the "machine" stops being a metaphor and becomes a model.