Margins Held, Robotaxi Expanded, Cybercab Started — But the Capex Number Just Got 25% Bigger and Free Cash Flow Goes Negative

Results vs. Consensus

Quality of Beat/Miss

• Revenue: +16% YoY is the strongest growth quarter since Q3 2024 and reverses three consecutive quarters of YoY decline (FY 2025 was Tesla's first-ever annual revenue decline). The mix is healthier than the auto-only headline suggests: Services & Other +42% on FSD subscription scaling, insurance, and Robotaxi infrastructure capex amortization; auto +13% on volume + ASP recovery; energy down -12% on lumpy deployment timing (8.8 GWh vs 14.2 GWh Q4).
• Margins: 21.1% total GM is the highest in 9+ quarters and is the most consequential pull-forward of the Q4 thesis. The auto ex-credits 19.2% includes one-time benefits (~$230M warranty + tariff relief); on a clean basis the Q/Q step-up from 17.9% is modest but positive. Energy 39.5% GM was a record but also includes ~$250M of one-time tariff recognitions from prior quarters — Taneja explicitly said normalized energy margins compress from here on competition + tariffs.
• EPS: Non-GAAP $0.41 beat clean. GAAP $0.5B net income was distorted by Bitcoin mark-to-market (BTC -22% Q/Q) and FX losses on intercompany borrowings — the same two below-the-line items that distorted Q4's GAAP. Strip them out and the operating earnings picture is on plan.
• Operating expenses: Up sequentially on a full quarter of the 2025 CEO performance award (one milestone still deemed probable) and AI-related spend on AI5 chip development, Cybercab/Semi/Optimus/Megablock R&D. Taneja flagged this elevated cadence persists through FY 2026.
• Deliveries: 358K missed the ~365K consensus. Taneja framed it as growth: EMEA "over 150% Q/Q in France and Germany," APAC growth in South Korea and Japan, slight U.S. Q/Q improvement, "highest Q1 order backlog in over 2 years." The production-over-deliveries gap of ~50K units is a real inventory build — to be watched.

Segment Performance

Automotive — Margin Step-Up Continues, but the Headline Includes One-Times

Auto revenue $16.2B (+13% YoY) on 358K deliveries (+6.3% YoY) marks the first quarterly delivery growth since Q3 2024. Taneja's framing was that Q1 is the demand-resurgence print Tesla has been waiting for: France and Germany Q/Q deliveries +150%+, South Korea/Japan growth, slight U.S. Q/Q improvement, and the highest Q1 order backlog in 2+ years. Auto margin ex-credits stepped from 17.9% in Q4 to 19.2% in Q1 — but the CFO called out, on the call, that this includes ~$230M of warranty true-downs and unspecified tariff relief. Adjusting, the underlying margin is closer to ~17.5–18% — directionally consistent with the Q4 thesis, but not the clean +130bps the headline implies.

"Auto margins, excluding credits, improved sequentially from 17.9% to 19.2%. Note that we have had certain onetime benefits from warranty true-downs around $230 million and some relief on tariffs. We have not realized any benefit from the recent Supreme Court ruling on IEEPA tariffs as there is still a lot of uncertainty around the final outcome." — Vaibhav Taneja, CFO

The order book commentary is the more durable signal. Taneja attributed the demand recovery partly to gas prices but emphasized the upturn started before that — i.e., the Tesla team's affordable-trim work and the brand stabilization (post-DOGE backlash) are doing real work. Berlin hit a record 61K units in Q1 driven by the in-house 4680 cell pack ramp; Reno is being retooled for higher pack output; in-house LFP module production is expanding in China.

Assessment: Auto is a real growth segment again, with two distinct tailwinds — international demand resurgence and Berlin volume ramp — partially offset by a 50K-unit inventory build that needs to clear in Q2. The clean underlying margin trajectory of 17.9% → ~17.5–18% (adjusted) is positive but modest. The structural-decline thesis is dead; the next debate is whether 17.5–18% is the new floor or the new ceiling. Battery pack capacity is now the binding constraint on volume — not demand.

Energy — A Lumpy Q1 with a Record-Margin Asterisk

Energy revenue $2.4B (-12% YoY) on 8.8 GWh of deployments — well below the 12–14 GWh that some sell-side desks were modeling, and a 38% sequential decline from Q4's record 14.2 GWh. Taneja was emphatic that the business is "inherently lumpy" tied to customer deployment timelines and reaffirmed FY 2026 deployments above FY 2025's 46.7 GWh. The 39.5% gross margin is a new record but includes ~$250M of one-time tariff recognitions from prior quarters; on a normalized basis margins compress from here on Chinese ESS competition and tariffs (most cells procured from China).

Assessment: Q1 GWh is a clear miss vs. the implied Q4 trajectory, but the lumpiness explanation is consistent with Tesla's prior guidance pattern. The order backlog is described as "robust." Megapack 3 production at the new Houston factory begins later 2026 — that's the real H2 catalyst. The 39.5% record-margin print is largely a one-time signal; the 2026 narrative for Energy is volume growth at compressed (mid-20s%) margins.

Services & Other — The Quietly Important Story

$3.7B (+42% YoY), now ~17% of total revenue. Margin stepped up from 8.8% to 9.2%. The composition matters: this segment now houses FSD subscriptions, insurance, supercharging, parts, used cars, and — importantly — Robotaxi infrastructure investment. Taneja called out that Tesla is "making deliberate investments in the infrastructure to help the Robotaxi in the future."

Assessment: Services is now the second-fastest-growing segment behind energy and the structural margin expansion vehicle. As FSD subscription transitions complete, this segment will pick up the deferred-revenue tail and become the cleanest read on Tesla's recurring-revenue thesis. We model Services to ~$18–20B FY 2026 from $14B FY 2025.

Key Topics & Management Commentary

Overall Management Tone: Forward-leaning, capex-defensive. Musk opened by reframing 2026 as the year Tesla "substantially increases its investments in the future" — a clear pre-emptive defense of the $25B+ number that Taneja then quantified. Taneja was disciplined about flagging one-time benefits in margin lines (warranty true-down, tariff recognitions) — a noticeable contrast to Q4's more triumphant tone, and probably the right move given the capex creep. Musk's Optimus/Cybercab/Semi cadence remained consistent ("the biggest product ever") but he leaned harder into the "S-curve takes time" framing, conditioning the Street for slow ramp. The body language: confident on the long-term inflection, slightly more measured about near-term cash flow.

Capex Blow-Up — $20B → $25B+ and Negative FCF Rest-of-Year

The Q4 2025 guide of "in excess of $20B" 2026 capex moved to "over $25 billion" on this call — a $5B (25%) increase in 90 days. Taneja explicitly said this will produce "negative free cash flow for the rest of the year." For context, Tesla generated $1.4B FCF in Q1 alone, so the negative ROY guide implies cumulative FY FCF below break-even — a sharper compression than even our Q4 model anticipated ($1–3B FY FCF). The line items: six factories in flight (Cybercab — production started; Semi; Megapack 3; Optimus Fremont; Optimus Giga Texas; new Megafactory), AI5 chip development, AI compute infrastructure for Robotaxi/Optimus, the Terafab research fab in Austin (~$3B), and now solar manufacturing equipment as well.

"Our current expectation for 2026 is over $25 billion of CapEx. And just to remind you, we are paying for 6 factories which we're going to go into operation… While this may seem a lot and will have the impact of negative free cash flow for the rest of the year, we believe this is the right strategy to position the company for the next era." — Vaibhav Taneja, CFO

Assessment: The capex blow-up is the central new information in this print. $25B+ is a step-function past the $20B that was already a stretch. The strategic logic is coherent — every line item maps to a near-term product (Cybercab, Semi, Megapack 3, Optimus) or a 5+ year strategic capability (Terafab, AI5/AI6 chip pipeline, internal solar manufacturing). But the FCF compression is real: a year ago Tesla generated $5B+; this year it goes negative; the inflection to incremental revenue from these investments is most credibly H2 2027. The bull case requires patience on FCF and faith in execution across six simultaneous factory programs. We mark this as bounded-but-real near-term thesis pressure — the Q4 thesis assumed a $20B/2-year capital trough, and this print extends it.

Robotaxi — Expanded to 3 Cities, Still Zero Injuries, Path to "Dozen States" by Year-End

Robotaxi expanded from Austin to Austin + Dallas + Houston in Q1, all running essentially v14.3 architecture. Per Musk and Ashok Elluswamy, the safety record is intact: zero accidental injuries to date, zero fatalities, and Ashok flagged that the customer fleet (close to 10 billion cumulative FSD-Supervised miles in the next few weeks) is now feeding back metrics that compress safety validation timelines. Musk's year-end guide: "unsupervised FSD or Robotaxi operating in… a dozen states by the end of this year" — a meaningful expansion of the Q4 "7 metros H1 2026" target.

"We've expanded Robotaxi to Dallas in Houston using the same software source in the Bay Area. And the limiting factor for expansion is really rigorous validation, making sure things are completely safe. We don't want to have a single accidental injury with the expansion of Robotaxi. And we have, to the credit of the team, not had a single one to date." — Elon Musk, CEO

However, Musk was explicit that Robotaxi revenue "would not be super material this year" — the material year is 2027. The constraints he flagged are not safety but convenience-mode failures (cars getting "scared," infinite-loop routing around construction, getting stuck behind crashed Waymos in Austin). Ashok noted the QA fleet is being expanded across the U.S. and the customer fleet is feeding the metric flywheel.

Assessment: The Q4 promise (7 new metros H1) tracked: Dallas and Houston are live; the architecture (v14.3) is the same as Austin; safety record is intact. But the Q1 disclosure stops short of delivering rev/cost-per-mile metrics — Tesla is still reporting Robotaxi as fleet count + paid miles "nearly doubled sequentially," not as a P&L line. The "dozen states by year-end" is a real expansion of guidance, but it's also a reframing — the Q4 deck targeted "25–50% of US population by EOY 2026," and the Q1 framing shifted to states (a coarser metric). Net: modestly positive — execution on time, but financial materiality is now Tesla's own admission of late-2026 / 2027.

Cybercab — Production Has Started

Musk in opening remarks: "We have just started production of Cybercab." This is the most consequential single line in the Q1 call — the Q4 2025 commitment to April 2026 SOP held. Musk reiterated Cybercab will likely become "the vast majority" of long-term Tesla production given that 90% of vehicle-miles-traveled have only 1–2 occupants. Volume guidance remained S-curve language ("very slow… ramping up… exponential towards the end of the year and certainly next year").

"We have just started production of Cybercab, and we'll begin production about Semi truck soon. And I should say, whenever you have a new product with a completely new supply chain, new everything, it's always a stretched out S-curve. So you should expect that initial production of Cybercab and Semi will be very slow, but then ramping up and going kind of exponential towards the end of the year and certainly next year." — Elon Musk, CEO

Assessment: The Q4 SOP pledge tripped bullish. The ramp will be slow — likely sub-1K Cybercabs in 2026 — but the binary "did production start?" question now has a yes. Combined with the architectural commitment (no steering wheel, no pedals), this validates Tesla as a real autonomy-vehicle business unit, not just a software layer on conventional vehicles.

FSD — 1.28M Paid, EU and China Approval Sequence Underway

FSD active paid subscriptions hit 1.28M globally (+51% YoY, +180K Q/Q from 1.1M at Q4) — the bulk of growth from subscriptions, with upfront purchases up only 7% as Tesla removed the upfront purchase option in some markets. Approvals secured: Netherlands in April; EU-wide review in May at Brussels (Taneja: "gated by how the regulators go about it"); China local approval received but broader approval still gated, target Q3.

"On the FSD adoption front, we continue to see improvement, reaching nearly 1.3 million paid customers globally. The bulk of the growth came from subscriptions, while upfront purchases only increased 7% as we remove the purchase option in some markets in Q1." — Vaibhav Taneja, CFO

Pierre Ferragu (New Street) ran the math on the call: 180K new paid users in the quarter against ~100K U.S. cars sold = roughly 2x FSD adds vs. new sales, implying that within the U.S. Hardware-4 installed base, take rates are now in the 30–35% range — saturating.

Assessment: FSD paid base is now a real subscription business. At a ~$99/mo blended run rate, 1.28M subs implies ~$1.5–1.7B annualized FSD ARR exiting Q1, vs. our $2.5B FY26 exit-rate model from Q4. That's a beat of pace, and the EU/China approval sequence is the meaningful 2026 unlock.

Hardware-3 Disenfranchisement — A Real Customer Friction Point

Musk confirmed on the call that Hardware 3 vehicles (sold through ~Aug 2023, ~3M cars) cannot achieve unsupervised FSD due to insufficient memory bandwidth (1/8th of HW4). For customers who paid for FSD on HW3 cars, Tesla is offering: (a) a "discounted trade-in" for HW4 cars or (b) a hardware retrofit (computer + cameras) executed via "small factories or micro-factories in major metropolitan areas" because service-center retrofits are "extremely slow and inefficient." Vaibhav added that a distilled v14 build for HW3 is being released end of June (drive-from-park-state with v14 features, but not unsupervised).

"Unfortunately, hardware 3 — I wish it were otherwise, but hardware 3 simply does not have the capability to achieve unsupervised FSD. We did think at one point, it would have that, but relative to hardware 4, it has only 1/8 of the memory bandwidth of hardware 4." — Elon Musk, CEO

Assessment: This is a real liability. Tesla sold FSD as a feature that would deliver autonomy on a future date, and HW3 customers — who could number in the high six figures of paid FSD — are now told they need a retrofit or trade-in. The financial exposure (warranty/goodwill) is bounded because Tesla has flexibility to price the upgrade, but the brand and class-action risk is real. The fact that Tesla is willing to build "micro-factories" to do retrofits at scale is the tell on how much they consider this a strategic liability. Mark this as a yellow flag — bounded but not zero.

AI5 Chip Taped Out Early; Terafab Research Fab Confirmed

Musk announced AI5 has taped out (2 quarters earlier than the original H2 26 plan), driven by team working "every weekend for 6 months straight, including every holiday." AI5 will go into Optimus and the data center first; Tesla has decided AI4 is sufficient for unsupervised FSD in vehicles. AI4+ (32GB RAM, +10% compute) is set for mid-2027 production at Samsung. AI6 design is underway; Dojo 3 ideas being discussed.

The Terafab research chip fab on the Giga Texas campus was confirmed: ~$3B initial spend, "a few thousand wafers per month" capacity, partnered with Intel on Intel's 14A process. SpaceX will fund "the initial part of the large-scale Terafab" — a related-party arrangement that requires both boards' independent director approval. (Note: the Intel 14A anchor-customer commitment was the key signpost in our INTC Q1 2026 thesis check, validated here from the Tesla side as well.)

"Terafab is not some sort of mechanism to generate leverage over our chip suppliers. It's just literally, we don't see a path to having enough efficient quantity of AI chips down the road… we just anticipate hitting a wall if we don't make chips ourselves." — Elon Musk, CEO

Assessment: AI5 ahead of schedule is the cleanest "AI execution is on plan" signal in this print. Terafab is more speculative — it's a 5+ year capability commitment with related-party governance complexity (SpaceX involvement, conflict-of-interest review by independent directors). Strategic logic is sound (AI silicon scarcity is a real long-term constraint); execution risk is real.

Optimus — V3 Reveal Pushed; Production "Late July, August"

The Optimus story decompressed slightly from Q4. Musk pushed the V3 reveal "closer to production" (mid-2026) and admitted V3 "works functionally but there's some aesthetic elements that need to be finalized" — and explicitly that Tesla doesn't want to give competitors more frame-by-frame copying material. The S/X line dismantling at Fremont starts in May (final S/X build); start of Optimus production targeted "somewhere around the late July, August time frame." A second Optimus factory is under construction at Giga Texas (production summer 2027). Musk reiterated Optimus is "our biggest product, not just Tesla's biggest product ever, but probably the biggest product ever" — but flagged "I don't know what the production rate of Optimus will be this year. It is impossible to predict these things."

Assessment: Optimus tracks the Q4 1M-unit Fremont conversion framework. The honest "I don't know the production rate" framing is appropriate given supply chain de novo, but it tempers near-term unit modeling. We hold our $20–100K FY 2027 Optimus units range from the Q4 model.

Guidance & Outlook

Implied Q-over-Q ramp: Q2 2026 deliveries should be flat-to-up modestly on the 50K Q1 inventory build clearing into the strong order backlog. We model Q2 deliveries 380–410K. Q1's 19.2% reported auto GM ex-credits will not repeat clean — we expect Q2 ex-one-times to print around 17.5–18.5%.

Street at: Pre-print FY26 EPS consensus ~$2.30–$2.60. After this print, the bifurcation continues: bulls mark up FY revenue and FSD ARR pace; bears mark down on the $5B capex creep and explicit negative FCF guide. Net: estimates probably hold, with the FCF line visibly down vs. Q4 marks.

Guidance style: Tesla again gave no formal vehicle volume guide but the operating cadence (Cybercab in production, Optimus July/August start, AI5 taped, Megapack 3 H2, Roadster demo "in a month or so") is sufficiently dense to model. The capex framing is now harder than Q4 — "over $25B" with explicit FCF coloring is a clearer commitment to the spend trajectory.

Analyst Q&A Highlights

Terafab & Intel 14A Partnership

• Will Stein, Truist: Asked who funds, designs, builds, operates Terafab. Musk: Tesla builds the ~$3B research fab on Giga Texas; SpaceX takes "the initial phase of the scaled-up Terafab"; intercompany governance must clear both Boards via conflict resolution. Intel partners on 14A process technology — Musk: "we plan to use Intel's 14A process, which is state-of-the-art… given that by the time Terafab scales up, 14A will be probably fairly mature… We have a great relationship with Intel."
  Assessment: This validates the Tesla-as-Intel-14A-anchor-customer commitment from the Intel side that we flagged in our INTC Q1 thesis. Cross-coverage signal is consistent. The SpaceX involvement adds related-party governance complexity but isolates Tesla shareholders from the heaviest scaling spend.

FSD Adoption Velocity

• Pierre Ferragu, New Street: Calculated 180K new paid FSD users vs. ~100K U.S. cars sold = 2x adds-to-sales ratio, suggesting near-saturation among HW4 owners in North America. Ashok confirmed framing: "Yes, I think you're thinking about it the right way." Also flagged that subscriber churn is dropping and customers are driving longer distances under FSD — both leading indicators for ARR durability.
  Assessment: Substantively the most useful Q&A in the call. The math implies FSD penetration in HW4 North America is now 30–35%. That sets up two things: (a) the EU/China approval sequence is the next leg of TAM expansion; (b) HW3 retrofit/upgrade economics matter — that base hasn't been monetized yet but is now the lowest-hanging incremental ARR.

Robotaxi Safety & Scaling

• Colin Langan, Wells Fargo: Asked which safety metrics Tesla tracks (miles per intervention, miles per accident, miles per fatality). Ashok: "We track basically all the metrics you mentioned" — large QA fleet across the U.S., simulator validation of every intervention, neural-net-based scenario modeling. Musk added that the limiter is not safety but convenience — cars getting "scared," infinite-loop routing — and mentioned a Waymo crashing into a bus blocking Tesla Robotaxis in Austin as illustrative.
  Assessment: The "we track everything but won't disclose specifics" stance is now consistent across two quarters. Investors deserve at least one quantified safety metric (e.g., interventions per 1,000 miles) by Q3 26.
• Mark Delaney, Goldman Sachs: Asked about new vehicle programs (compact/family vehicle). Musk pivoted to Cybercab as the answer ("Cybercab is compact… it's a 2-person vehicle… most of our production long term will be Cybercab"). Also flagged the new Tesla Roadster demo "in a month or so" — "spectacular demo" but "I don't think it moves the needle massively from a revenue standpoint."
  Assessment: The "new vehicles" debate is being reframed as "Cybercab is the new vehicle" — which is internally consistent but means there is no compact-affordable mass-market product separate from the autonomy stack. Roadster is a halo unveil, not a revenue program.

Battery Pack Capacity

• Mark Delaney, Goldman Sachs (follow-up): Battery cell vs. pack constraint, in-house production roadmap. Taneja: "the limiter is not the cells itself. It's the battery pack capacity." Lars Moravy: Berlin in-house 4680 packs ramping; Reno being retooled for higher pack output; China LFP module/pack production growing.
  Assessment: The pack constraint is consistent with Q4 commentary (battery was flagged as global constraint then too). Net: 2026 volume is supply-constrained, not demand-constrained — bullish for pricing power, bearish for delivered units near-term.

Optimus & xAI / Grok Architecture

• Pierre Ferragu, New Street (follow-up): Asked whether Optimus intelligence runs on-board or via xAI/Grok in cloud. Musk: most intelligence runs locally on-robot for connectivity-loss resilience; Grok handles "orchestration" (manager-level commands) and low-latency voice. Optimus can "work for several hours without management oversight."
  Assessment: This is the cleanest articulation yet of how Tesla's $2B xAI investment from Q4 2025 connects to Optimus economics — Grok is the orchestration layer. Strategic logic is sound; the on-robot AI5 inference (per Musk's earlier remarks) handles the heavy compute.

What They're NOT Saying

1. Robotaxi unit economics, again: Two consecutive quarters of "we'll disclose later." 500+ vehicles in Q4, expanded fleet across three cities in Q1, paid Robotaxi miles "nearly doubled sequentially" — but no rev/cost-per-mile. The "later" framing is wearing thin.
2. FY26 vehicle volume target: Still no number. Taneja's "highest Q1 backlog in 2 years" plus Musk's "ramping production at all our factories" is the closest to guidance, but no quantified target.
3. FY26 FCF range: Negative ROY guided but no quantification. We model FY26 FCF in the $(2)B–$0B range.
4. HW3 retrofit cost / customer impact size: The retrofit program was announced but not sized. How many HW3 cars carry paid FSD? How many will trade in vs. retrofit? What's the warranty exposure? All deferred.
5. Q1 IEEPA tariff benefit explicit size: Taneja flagged "some relief on tariffs" but did not quantify (the warranty true-down was sized at $230M; tariff relief was not). Likely a deliberate omission.
6. Cybercab consumer pricing: Production has started but no announced price. Musk's old "$30K" framing predates inflation; current target unstated.
7. Energy 2026 GWh range: "Higher than 2025" is the only commitment. The Q1 8.8 GWh print is a notable downshift; ambient sell-side modeling for 75–100 GWh FY 2026 needs the next two quarters to validate.
8. SpaceX-Tesla intercompany governance: The Terafab structure (SpaceX takes "scaled-up" portion) requires conflict resolution between independent directors of both Boards. Process and economic terms not disclosed.

Market Reaction

• YTD context: TSLA entered the Q1 print down ~20% YTD per pre-print sell-side coverage, on growth-narrative skepticism after Q4's $20B capex disclosure and a slower-than-hoped January–March news flow.
• After-hours move: Modestly positive on the headline beats; the negative FCF guide and capex creep capped enthusiasm.
• Volume: Elevated (typical for TSLA earnings).
• Analyst reactions (within 48 hours):
  • Mixed: bulls highlight FSD 1.28M, Robotaxi expansion, Cybercab SOP; bears highlight $25B+ capex creep and 50K-unit inventory build.

The reaction is rational. Tesla beat top and bottom lines, hit every major Q4-promised autonomy signpost, and disclosed FSD ARR-positive metrics — but the $5B capex creep and the explicit negative-FCF guide convert the narrative from "Q4 thesis intact" to "Q4 thesis intact with a longer trough." The market is doing the right math.

Street Perspective

Debate: Is the $25B+ Capex Number a Buying Opportunity or a Red Flag?

Bull view: Six factories simultaneously, AI5 ahead of schedule, Cybercab in production, Terafab confirmed — this is exactly the "spend hard now, harvest 2027–2028" capital allocation that creates platform companies. The $5B creep vs. Q4 reflects accelerating opportunity, not slipping discipline. Negative FCF for 9 months is the price of admission to a multi-trillion-dollar AI/autonomy/robotics complex.

Bear view: Capex was supposed to be $20B; it's already $25B+; Taneja excluded TerraFab and solar from the prior $20B but those are now in the new number; this is exactly the bloat track Tesla has run before. The FCF compression is not bounded — the mid-case is no better than break-even FCF for 12+ months, and the upside cases all rely on Robotaxi and Optimus ramping faster than the production-line S-curve framing implies.

Our take: The bear has the better short-term math (we model FY26 FCF at $(2)B to $0B, materially worse than our prior $1–3B), but the bull has the better strategic math (every line item maps to a real product or capability). The right framing is that the trough got deeper but the thesis trajectory is intact. Net: the $5B creep is a real near-term headwind that warrants haircutting our 2026 EPS by ~5–10% and our 12-month price target by ~10%, but does not reverse the thesis.

Debate: Is the 19.2% Auto GM Ex-Credits Sustainable?

Bull view: Even adjusting for ~$230M warranty and tariff one-times, the underlying ~17.5–18% is a step up from FY 2025's ~14–15% and Q4's 17.9%. International mix (EMEA +150% Q/Q in France/Germany) is structural; FSD attach (1.28M) is increasingly margin-accretive; Berlin's in-house 4680 pack ramp lowers BOM cost. 2026 trends to 18%+.

Bear view: The headline reads great because of one-times. Strip them out and the Q/Q step-up from 17.9% to ~17.5–18% is almost flat. The FSD subscription transition adds 100–200bps of near-term margin headwind via deferred revenue. Tariff regime is uncertain (IEEPA Supreme Court ruling unincorporated). 2026 lands closer to 16–17%.

Our take: Underlying 17.5–18% is the right base case for Q1. We model FY26 auto ex-credits at 17–18% — modest improvement from FY25 but below the bullish 19%+ scenario. The path to 19%+ requires Cybercab volume contribution and EU/China FSD revenue recognition.

Debate: Robotaxi Expansion Pace — On Track or Already Slipping?

Bull view: Three live cities by Q1 vs. seven targeted for H1 — Tesla is ahead of pace. "Dozen states by year-end" is a meaningful expansion. Zero injuries, zero fatalities, 10B cumulative FSD-Supervised miles imminent — the safety flywheel is the real moat.

Bear view: The Q4 deck targeted "25–50% of US population by EOY 2026"; Q1 framing shifted to "states" which is coarser. Actual rev/cost-per-mile is still undisclosed. "Dozen states" can mean 12 cities at sub-100 vehicles each — geographic count is a vanity metric without economic disclosure.

Our take: Tesla is on pace operationally and modestly behind on financial-disclosure pace. We hold our Q4 model: ~2,000-vehicle Robotaxi fleet by mid-2026 and material revenue contribution late-2026 / early-2027. The Q3 26 print should bring rev/cost-per-mile disclosure.

Debate: Hardware-3 Disenfranchisement — Bounded or Material?

Bull view: Tesla has flexibility on retrofit pricing; the "micro-factories" approach is creative; HW3 customers retain v14 distilled features (released June). This is a sub-$1B bounded warranty event.

Bear view: Tesla sold FSD as autonomy-on-future-date to ~3M HW3 cars. Class-action exposure is not bounded by Tesla's pricing flexibility — it's bounded by what plaintiffs argue Tesla owes. Reputational damage extends beyond financial.

Our take: Likely bounded financially ($500M–$1.5B over 24 months), but reputationally the "FSD as autonomy promise" muscle memory is now visibly tested. Mark this as a yellow flag — not yet pricing into our model meaningfully but tracking.

Model Update Needed

Valuation impact: Pre-print TSLA in the high $300s; trailing EPS ~$1.45. At our revised FY26 $2.30 EPS midpoint that's ~150x — still expensive on near-term earnings but defensible if the 2027–2028 inflection (Cybercab volume, FSD ARR EU/China, Optimus pilot) lands. We modestly trim our 12-month price target range to $385–$455 (from the prior $400–$475 post-Q4). The reduction is driven entirely by FCF compression and EPS haircut, not thesis change.

Thesis Scorecard Post-Earnings

Overall: Thesis tilts still positive but at lower conviction. Six of seven Q4 bull pillars confirmed (margin recovery softer than headline but directionally right; Cybercab SOP, FSD ARR, AI5, Robotaxi expansion, auto growth all confirmed). The bear concerns from Q4 (capex compression) confirmed at a worse magnitude than expected, and a new bear concern (HW3 disenfranchisement) emerged. The strategic execution is on plan; the financial cost is higher than priced.

Action: Maintain Outperform at lower conviction. The Q4 inflection thesis remains intact — every operational signpost tripped favorably — but the FCF trough is deeper and the disclosure pace on Robotaxi economics is below where it should be by Q3 26. We would downgrade to Hold if Q2 capex slips further (toward $30B) without offsetting Robotaxi revenue disclosure, or if Q2 auto margin ex-credits prints below 17% on a clean basis.