Recon Recovers, Records Across the P&L, Q2 Records Guided — Upgrading to Outperform from Hold

Initial Read. The Q4 reconditioning stumble was a one-quarter event — April labor efficiency is back near the company’s all-time best — and a clean revenue + EBITDA + EPS beat with a forward guide for record Q2 units and adj EBITDA validates the upgrade trigger we flagged at Q4.

Results vs. Consensus

A clean broad-based beat that addresses every concern from the Q4 print. Most importantly, the operational stumble that drove the Q4 sell-off has measurably reversed.

Quality of Beat

• Volume: 40% retail unit growth into a high-prices, mixed-macro environment is the single most thesis-relevant data point. The market grew sub-5% in units; Carvana’s share gain is the entire growth story.
• Margins: Adj EBITDA margin at 10.4% is down 110bps Y/Y but that comparison is distorted by gross-revenue accounting on marketplace inventory. EBITDA dollars +$184M Y/Y on units +40% Y/Y is the cleaner read — per-unit EBITDA is approximately holding. Q1 had $36/unit ops expense reduction and $226/unit overhead leverage offsetting $92/unit advertising step-up — the SG&A walk continues to function.
• EPS: Operating quality. $1.69 GAAP EPS is from genuine operating performance, not below-the-line items.
• Recon flow-through: The Q4 stumble shows up as elevated cost-of-goods on cars produced in Q4 but sold in Q1 (Garcia explicitly noted: “cars carry the cost of reconditioning at the time they were produced, not at the time they were sold”). That means Q2 GPU benefits from the April labor-efficiency recovery only partially; Q3 is the first quarter to fully reflect normalized recon costs.

Operational Performance

Key Topics & Management Commentary

Overall Management Tone: The most confident posture from the past four calls. Garcia opened with extended, specific commentary on the Q4 reconditioning stumble and the recovery work, which read as both accountability for the prior miss and validation of the operating culture. The narrative explicitly framed bumps as “a chance to reevaluate” rather than as setbacks to deflect. Notably, no defensive posture was needed on any Q&A topic this quarter.

Reconditioning: From Stumble to Recovery in One Quarter

Garcia’s prepared remarks framed Q4’s recon issue as a culture test that the team passed. The substantive list of fixes was the cleanest articulation yet of how Carvana’s operations machinery actually evolves under stress:

“The recon team is using that pressure to make us better. ... Over the last couple of months, they built additional data integrations, developed tools to help managers make faster, higher quality decisions, and how they staff their lines, and how they optimize flow through their paint lines and implemented a productivity tracker to ensure feedback reaches the right groups quickly. ... So far in April, we are operating just shy of our all-time best and labor efficiency throughout the network.”
— Ernest Garcia, CEO

Mark Jenkins added the timing nuance — recon costs hit P&L when cars are sold, not when they’re produced — meaning Q1’s -$58 retail GPU still embedded some Q4-elevated recon, with cleaner reads coming in Q2 and especially Q3.

Read-through: The recovery is real and faster than the 3-to-6-month timeline Garcia committed to at Q4. Q3 2026 should print clean recon costs, which is when retail GPU re-acceleration becomes most visible. Critically, the new tools are net-additive — they push the floor of facility performance up, not just back to prior levels. This is one of the most thesis-positive operational data points of the past 18 months.

Wholesale Franchise: ADESA Clear Now Carrying Real Weight

Jeff Lick (Stephens) probed the wholesale GPU dynamics. Garcia’s response highlighted ADESA Clear’s buyer-side advantage as a discrete contributor to wholesale vehicle GPU strength:

“We made a comment in the letter that we feel like ADESA Clear, which is our digital platform, is now a best-in-class platform. ... We’ve built a platform for the buy side that we think is highly differentiated, and where there’s room to differentiate it further from here. And we think that’s showing up in the results.”
— Ernest Garcia, CEO

The 4.8-day end-to-end cycle anecdote (sourced from a customer, reconditioned, listed, sold, delivered) was the most concrete articulation yet of the wholesale-to-retail flywheel.

Assessment: The wholesale franchise is graduating from a “side benefit of ADESA acquisition” to a “structurally advantaged buyer of cars” story. Worth noting in the bull case as a distinct pillar separate from retail unit growth.

Q2 Outlook: Records Guided, GPU Y/Y Pressure Lapping

Jenkins’ Q2 guide was the most specific in two quarters: “sequential increase in both retail units sold and adjusted EBITDA, leading to all-time company records on both metrics.” The GPU caveats are significant though — ~$100 of Y/Y headwind from the 2024 tariff-pull-forward lap, $100–200 of wholesale-retail spread compression, and continued (smaller) non-vehicle cost pressure.

On wholesale-retail spreads, Jenkins’ framing was that Q1 saw a hot wholesale market that was not fully passed through to retail prices, creating temporary spread compression. Garcia followed up that this was “a transitory impact” following normal seasonal patterns with a longer-than-usual lag.

Assessment: Q2 is set up for a clean operational beat (records on units and EBITDA) but headline retail GPU will still print down Y/Y. This is a quarter where investors will need to read past the GPU optic to the underlying operational delivery. The setup increases the value of the post-Q2 print as a confirmation event.

SG&A Walk: Operating Leverage Compounding

Daniela Haigian (Morgan Stanley) probed SG&A line items. Jenkins’ breakdown was the cleanest articulation of the operating-leverage flywheel: ops expense down slightly Y/Y, overhead substantially levered, advertising marching up by design. The Q1 sequential overhead step-up (Rajat Gupta, JPMorgan) was attributed to Q1 share-based comp seasonality, weather-related expenses, and ongoing technology investments.

Jenkins explicitly characterized Q1 overhead as a “new level” rather than a one-off — so investors should not expect overhead to step back down in Q2.

Macro & Affordability

Garcia’s response on car prices being up 35–40% vs. pre-pandemic (vs. consumer goods +25%) was matter-of-fact rather than alarmist. His framing — that the impact on aggregate sales is small at the level of a normal cycle — matches our prior. The early Q2 commentary on Iran-related gas-price spikes and tax-refund timing was that effects are visible but smaller than expected and should normalize.

Read-through: Macro is a watch item but not a thesis-driver. Carvana’s 40% growth into sub-5% market growth means even substantial macro deterioration would still leave significant share-gain runway.

Gas Prices, EVs, Mix

Brian Nagel (Oppenheimer) probed gas-price impacts. Garcia confirmed mix shift (lower large SUV demand, higher EV demand), but framed it as already partly normalized over recent weeks. The system’s ability to adapt buying patterns within days — given fast inventory turn — was the underlying point: macro and product-mix shifts get absorbed at the inventory layer rather than the P&L layer.

Capital Allocation: Production Build-Out Plus Continued Deleveraging

Daniela Haigian also probed capex outlook for the build-out program. Jenkins’ clearest articulation:

“We’re going to start doing some of those full build-outs, which we think make a lot of sense. ... Greenfield IRCs — that’s not a priority at this time.”
— Mark Jenkins, CFO

Read-through: Capex steps up in 2026 but stays disciplined — ADESA build-outs (existing land) preferred over greenfield. Combined with 1.1x leverage and growing TTM EBITDA, the FCF profile remains constructive.

Guidance & Outlook

Implied math: Q2 records on both units and EBITDA imply ~200K units and ~$700M+ adj EBITDA, both at the high end of plausible Street modeling. Annualized run-rate based on Q1 2026 + the implied Q2 trajectory points to FY26 EBITDA of $2.8–3.0B, which is consistent with the 30–35% growth framework.

Guidance style: The Q2 records-on-records guide is the closest CVNA has come to a quantified near-term framework all year. Consistent with management’s preference for directional rather than precise guidance, but specific enough to anchor expectations. Tactically, the Q2 guide is set up to be beatable (Q2 is seasonally strong; the recon recovery is just starting to flow through; tax-refund tailwind already at least partly captured).

Analyst Q&A — Notable Exchanges

Q&A skewed toward recon recovery validation, GPU dynamics, capex/build-out plan, and macro absorption. Notable threads:

• Chris Pierce (Needham) opened on the new recon tools (centralized planning, productivity trackers); Garcia framed the tooling as a floor-lift across all sites rather than just a fix for underperformers. Pierce’s follow-up on broader market dynamics drew Garcia’s clearest framing yet on used-car-market scale (low elasticity, structural).
• Daniela Haigian (Morgan Stanley) probed SG&A leverage trajectory and longer-term capex plans for going beyond the 3M-unit goal. Jenkins’ ADESA-build-out vs. greenfield framing was the cleanest articulation of the capital-deployment philosophy.
• Rajat Gupta (JPMorgan) drew attention to the Q1 sequential overhead step-up; Jenkins attributed to Q1 share-based comp seasonality, weather, and technology investments, and characterized the new level as “more like a new level” vs. seasonally one-off.
• Sharon Zackfia (William Blair) got the cleanest framing on Q2 retail GPU: sequential improvement but Y/Y down, with the path to 13.5% adj EBITDA margin still “clear visibility.” Her tax-refund-and-Iran-impact follow-up drew Garcia’s read that the tax-refund tailwind was smaller than expected and partly offset by gas-price effects.
• Brian Nagel (Oppenheimer) probed gas-price impacts on consumer behavior; Garcia’s response on aggregate-vs.-mix effects was consistent with prior calls. The wholesale-retail-spread question got clearest articulation on the lag dynamics.
• Jeff Lick (Stephens) raised the wholesale dynamics — lower wholesale-to-retail ratio (44.6% from 47.4%) plus very strong wholesale GPU per unit (~$1,327). Garcia’s ADESA Clear best-in-class framing + the 4.8-day cycle anecdote was the call’s most thesis-positive disclosure.
• John Colantuoni (Jefferies) asked about Other GPU runway given continued rate cuts to customers; Garcia’s 13.5% margin walk reaffirmation was the cleanest path-to-target framing of the call. His advertising-channels follow-up was the most useful CFO-level disclosure on the marketing strategy.
• John Healy (Northcoast) pressed twice on the franchise-dealership acquisitions; Garcia explicitly declined twice. Pattern continues from Q3 and Q4 — the franchise-dealership story is being kept off the model.
• Marvin Fong (BTIG) probed inventory dynamics (+30% Y/Y vs. units +40% Y/Y, implying faster turn). Garcia confirmed inventory build is constrained by recon throughput and connected to Phoenix-test conversion-rate data.
• Andrew Boone (Citizens) pressed on centralized-planning tools and ADESA Clear long-term opportunity. The longest single substantive answer of the call from Garcia on the wholesale platform’s strategic positioning.
• John Babcock (Barclays) probed retail GPU mechanics and the centralization risk. Jenkins’ balance — quantitative software-driven planning paired with on-the-ground judgement — was the operationally substantive framing.
• Michael McGovern (Bank of America) closed on labor-hours-per-unit recon improvement — back to Q2 2025 all-time best level. Garcia framed continued improvement as multi-quarter rather than near-term.
• Michael Montani (Evercore ISI) got the diesel/fuel impact framing (small Q2 headwind, not material) and Garcia’s reinvestment-philosophy framing.

What They’re NOT Saying

1. Still no quantified FY26 unit or EBITDA range. Carvana continues to prefer qualitative full-year framing. The Q2 records-guide partially substitutes for a full-year number but leaves the H2 trajectory open. We don’t treat this as a red flag given the recon recovery is fresh and the wholesale-retail spread compression is acknowledged transitory, but it remains a notable choice.
2. No quantified FY26 capex number. ADESA full build-outs are confirmed for 2026 but no dollar number provided. Modeling: $400–500M of incremental capex over the historical baseline, but management hasn’t anchored a specific figure.
3. No update on franchise-dealership strategy. Third consecutive quarter of explicit decline. The pattern is now clear: management is keeping a strategic option live without committing to it. Watch for an Investor Day or framework update at some point in 2026.
4. No commentary on share buybacks or dividends. With 1.1x leverage, the next capital-allocation question is whether Carvana ever introduces a buyback. Not yet.
5. No further commentary on the related-party allegations from Q4. Substantive defense was offered at Q4 and not reiterated; the absence is appropriate (not raising the issue when it was not raised by analysts is the right tactical move) and should not be over-read.
6. No updated cohort-level disclosure for legacy markets. Same conspicuous non-disclosure as the prior three quarters. The continued silence on mature-market unit economics remains the single most-watched disclosure gap.

Market Reaction

• Pre-print setup: CVNA had been re-rating moderately positively into the print on the loan-platform expansion + balance-sheet improvement narrative, partially offsetting the Q4 sell-off. Sentiment was cautiously optimistic with elevated short interest as a partial offset; the implied move was sized for both directions.
• Initial reaction: Print landed after the close on Apr 29. Shares rose as much as ~10% in extended trading on the broad-based beat, recon recovery, and forward Q2 records guide.

The reaction is the cleanest positive print-tape in three quarters and validates that the Q4 sell-off was about the operational stumble rather than a deeper thesis change. Combined with the loan-platform expansion + balance-sheet inflection, the market has resumed pricing the structural-compounder thesis. We expect the Q2 records-guide to be the next near-term confidence-builder; if Q2 prints with normalized recon costs and any quantified FY26 framework, the multiple has further room to expand.

Street Perspective

Debate: Is the recon recovery durable or seasonal?

Bull view: The new tools (centralized planning, productivity trackers, paint-line optimization) are floor-lifters across all sites, not just fixers for the laggards. April labor efficiency near all-time best confirms the recovery is genuine. With Q3 being the first quarter to fully reflect normalized recon costs, retail GPU re-acceleration is mechanically set up.

Bear view: Q4 was the first sign that ADESA-site ramp curves are non-linear. The recovery happened fast, but the underlying issue (newer managers + single-line vs. multi-line capacity + scaling complexity) is persistent. The next stumble is one or two quarters away.

Our take: Closer to bull. The specific tooling improvements described are durable structural fixes, not one-time re-prioritization. We accept that operational complexity at 40%+ growth is bound to produce occasional stumbles, but the demonstrated 1-quarter recovery + the new floor-lifting tools materially de-risk the medium-term margin trajectory.

Debate: Does the Q2 GPU lapping create downside revision risk?

Bull view: Management telegraphed the GPU mechanics specifically to manage expectations. The +$100 tariff lap and +$100–200 wholesale-retail spread are clearly transitory; both should normalize by Q3. Q2 will print records on units and EBITDA dollars regardless.

Bear view: Sell-side estimates haven’t fully absorbed the Y/Y headwinds for Q2 GPU; even with EBITDA-dollar growth, Y/Y margin compression is likely to drive negative model revisions. Stock could re-rate lower into the Q2 print.

Our take: Closer to bull. The transparent management framing of the GPU walk reduces the risk of a Q2 expectations gap. Q2 is more likely to be a “record beat with Y/Y margin noise” print than a meaningful disappointment.

Debate: Is the path to 13.5% adj EBITDA margin credible at this point?

Bull view: The walk is straightforward — fixed-cost leverage + advertising eventually settling at mature-cohort levels + continued fundamental gains across the GPU stack. With Q1 at 10.4% margin, the gap to 13.5% is ~310bps over 4–9 years — an annual cadence well within historical CVNA delivery patterns.

Bear view: Each year that customer pass-throughs (shipping, rates) re-absorb fundamental gains, the path to 13.5% extends. Q1 revealed that advertising is being framed as a multi-year “new level” rather than a step that pays back. The 13.5% goal is achievable but the timeline is more 8–10 years than 5–7.

Our take: Bull on the achievability, neutral on the timeline. We model 13.5% as a 2030–2032 event rather than 2030. The thesis still works at that timeline because the unit-growth trajectory compounds the EBITDA dollar profile.

Model Implications

• FY26 retail units: 775–810K (33–35% growth); we anchor at 800K.
• FY26 adj EBITDA: $2.85–3.05B; we anchor at $2.95B (32% growth) on continued unit expansion + a Q3/Q4 recon-cost normalization tailwind partially offset by capex-related opex and continued advertising spend.
• Q2 2026: Records on units and adj EBITDA. We model ~200K units (+35% Y/Y) and ~$720M adj EBITDA. Retail GPU down Y/Y on lapping but up sequentially.
• Q3 2026: First quarter to fully reflect normalized recon costs. We model retail GPU swinging back to flat-to-slightly-up Y/Y on the tooling improvements + lapping the Q3 2025 strong-Other-GPU baseline.
• FY26 capex: $400–500M of incremental ADESA build-out spend on top of historical baseline.
• Net debt: 1.1x and declining. Plausibly sub-1.0x by mid-2026 even with capex step-up. Investment-grade ratios is now a 2026–2027 event.
• FCF: Inflecting; FY26 FCF likely positive even with the capex step-up given EBITDA growth. Cash tax shield from the Q4 DTA realization is a structural FCF tailwind.

Thesis Scorecard

Overall: Thesis materially strengthens. Six bull pillars confirmed (one new); three of six bear pillars receding/pushed out; only the customer-reinvestment margin-cap pillar remains active — and even that is by management design, not a structural problem. The composition of the print — broad-based beat + recon recovery + Q2 records guide + balance-sheet inflection + ADESA Clear contribution — is the cleanest demonstration of the long-term compounder thesis since we initiated coverage.

Action: Upgrading to Outperform from Hold. We initiated at Hold at Q2 2025 and maintained at Q3 and Q4 2025 specifically on a valuation gate that required (a) the recon stumble to clear and (b) the operational story to re-anchor. Q1 2026 cleared both. The franchise is delivering, the recon recovery is concrete, the balance sheet has inflected to a career-best 1.1x leverage, the loan-platform / DTA / capital-allocation story is intact, and ADESA Clear has graduated to a discrete structural-advantage pillar. Even at the post-print level, the underlying compounding profile underwrites a 12-month total return above the S&P 500. Multiple expansion is potential upside, not the rating gate. We continue to monitor for FY26 framework articulation and Q3 retail GPU normalization as the next thesis-confirming events.

Net: Carvana spent the past three quarters proving it could absorb a stumble; this print proves it can recover from one. Outperform.