April 24, 2026
The Q1 2026 Tesla earnings cycle produced two numbers that do not usually show up in the same week. The first was a gross margin print of 21.1 percent, the best sequential improvement since mid-2024 and the data point that validated every bullish spreadsheet on the company since the cost-cutting campaign began. The second was a CEO admission that the hardware sitting in hundreds of thousands of customer vehicles is not capable of the feature those vehicles were sold against for eight years. Both numbers belong in every Q2 model. Neither one fully offsets the other.
CNBC’s earnings recap logged the top line on April 22. Revenue of $22.39 billion against $22.5 to $22.64 billion consensus. Adjusted EPS of $0.41 against $0.37. Auto revenue of $16.2 billion, up 16 percent year over year. Gross margin up 478 basis points year over year and 100 basis points sequentially from the Q4 20.1 percent base. Deliveries of 358,023 against a 372,160 consensus. The post-close stock move was plus 4 percent on the print and reversed through the call as the capex reset was disclosed. Electrek’s live coverage carried the 2026 capex guide of $25 billion, up $5 billion from the prior range, with six new factories plus unspecified AI, Optimus, and robotaxi spend. Musk described the rest of 2026 free cash flow as “negative,” which is the direct consequence.
The gross-margin expansion is real and sequentially compounding
The data point that deserves independent attention before anything else is the gross-margin line. Automotive gross margin at 21.1 percent is the best Tesla print since mid-2024 and reflects a combination of Model Y Juniper cost reductions, reduced incentive spending at the Model 3 trim level, and better-than-expected operating leverage on the Shanghai Gigafactory Model Y L production ramp. A 100 basis point sequential improvement is not the floor number most models had coming into the print. That is the one line on the P&L where Wall Street was wrong into Q1 earnings, and it is the single most durable reason to believe Tesla’s near-term profitability story independent of the FSD narrative.
Whether the gross-margin trajectory holds through Q2 depends on the 358,023 deliveries number. Below-consensus volume with above-consensus margin is the classic “quality of the beat” debate, and it is running live on the sell side this week.
| Metric | Q1 2026 | Consensus | Delta |
|---|---|---|---|
| Adjusted EPS | $0.41 | $0.37 | Beat |
| Revenue | $22.39B | $22.50B-$22.64B | Miss |
| Deliveries | 358,023 | 372,160 | Miss |
| Automotive gross margin | 21.1% | ~20.1% Q4 | +100 bps QoQ |
| 2026 CapEx guide | $25B+ | ~$20B | +$5B |
Source: Tesla Q1 2026 release, CNBC, Electrek, MarketBeat — April 22-23, 2026.
The capex raise is the growth bet that cost the multiple
The $5 billion capex increase is the more consequential structural move. Tesla is now committed to $25 billion of 2026 capital spending against an auto business that is guiding to negative free cash flow for the remaining three quarters. Six new factories plus AI infrastructure plus Optimus plus the robotaxi fleet buildout are the line items Musk named. The specific dollar split was not disclosed.
The sell-side response was immediate. TD Cowen cut its price target from $519 to $490 while keeping the Buy rating. The cut is the first major price-target reduction on Tesla in April and sits inside a consensus average across TD Cowen, Cantor Fitzgerald, and Wedbush of $533.33, still 43 percent above Thursday’s close. The Buy rating survives the capex reset. The multiple does not.
The Hardware 3 admission is the consumer story
The revenue and capex conversation dominated Wednesday after-hours. Thursday’s trading was driven by something else. On the Q1 earnings call, Musk told analysts that Hardware 3 “simply does not have the capability to achieve unsupervised FSD.” He also pulled the end-of-2026 robotaxi scope back from the “vast majority of the US” framing of prior quarters to “a dozen or so states” and described FSD and robotaxi revenue contribution in 2026 as “not super material.”
Hardware 3 shipped on Tesla vehicles from 2019 through early 2023. It is the autonomy compute platform in the Model 3, Model Y, Model S, and Model X vehicles from that period. The number of HW3 vehicles on the road is in the low-to-mid hundreds of thousands, not counting global volume on the same platform. Every one of those vehicles was sold against a consumer marketing promise that the onboard hardware was sufficient for future unsupervised full self-driving, subject to software updates. Musk has now publicly set that aside.
The consumer-expectation question is the part of the story that does not show up in a sell-side model. Tesla has not committed to an HW3-to-HW4 or HW3-to-AI5 retrofit program. The cost of such a retrofit, the throughput at service centers that are already constrained, and the legal mechanism under which Tesla would offer an upgrade path for vehicles sold on the FSD promise are all open questions. Tesla’s prior FSD transfers between vehicles were voluntary consumer programs, not hardware-retrofit commitments. The 100,000 jailbroken Teslas story from April 13 is one of several markers that Tesla’s FSD ecosystem in 2026 is larger and more fragmented than the direct subscription business alone.
The Dallas and Houston geofences are the ground-truth check
Two days before Tesla reported, the company activated unsupervised robotaxi in Dallas and Houston. Gizmodo documented the first-24-hour availability at 0 to 2 percent with tiny geofences, roughly 25 square miles in Houston and 30 to 35 square miles in Dallas. One vehicle ran in Dallas. Two ran in Houston.
That availability profile is the ground-truth read against which the “a dozen or so states by end of 2026” revised scope has to be evaluated. A state-level robotaxi rollout requires a multi-city operating capability rather than a single 30-square-mile downtown geofence. The Austin robotaxi business is still operating under the NHTSA PE25012 investigation into 14 incidents and 80 traffic violations from the June 2025 launch window, with the second NHTSA deadline already extended. Adding states to the map does not solve the per-city availability problem. The Dallas and Houston launches on April 19 and April 20 are the data points that say so in public.
The repriced multiple
Tesla’s Q1 print contained three directional signals that move in three different directions. The gross-margin expansion is the genuine positive and belongs in every Q2 and full-year model. The capex raise is the growth commitment that converts near-term free cash flow from positive to negative and is the reason the TD Cowen price target came down. The Hardware 3 admission and the narrowed robotaxi scope are the narrative reset that affects what Tesla can credibly tell customers, dealers, and competitors about unsupervised autonomy through the end of 2026. For buyers of 2019 to 2023 Teslas sold on the FSD-capable hardware promise, the week provided the clearest statement in eight years that the hardware in their vehicles is not the hardware Tesla now says the product requires. For Tesla investors, the story is that Musk just spent $5 billion of 2026 free cash flow to build a robotaxi future that now comes with a smaller state map and a narrower hardware-compatible installed base than it did on April 21. The numbers that go into the Q2 model are clear. The broader narrative that used to carry the multiple is the part that has to be repriced.
