| Jul 22, 2026 | -14.52% | Q2 FY2026 | Read transcript briefingQuarter in one view- Record Q2 deliveries with broad sequential growth: Americas +60%, APAC +27%, EMEA +12% QoQ; Model Y set records in Netherlands, Australia, New Zealand; largest order backlog since 2023. Production growth now limited by supply chain (batteries and electronic components), not demand.
- Auto gross margin ex-credits fell 19.2% → 16.3%, but Taneja framed it as ~flat ex the $230M Q1 one-timers (warranty true-downs, tariff relief). Rising interest rates increased upfront subvention costs, a structural margin drag.
- Energy flipped the script: 13.5 GWh deployed (+53% QoQ, second-largest quarter ever), but margin collapsed 39.5% → 20.4% on a ~$240M warranty true-up (vendor cell issues in legacy deployments), lapsed Q1 tariff benefits (>$200M), and guided ASP declines. Long-term margin now explicitly normalized to "mid- to low 20%."
- FCF turned negative as guided; CapEx more than doubled sequentially and will rise further in H2. The >$25B 2026 guide held, but the funding posture escalated: Tesla is securing debt facilities for up to $30B of borrowing capacity — new this quarter.
- Robotaxi disclosure improved materially: >380,000 unsupervised miles across six cities/two states (seven markets per Taneja), zero notable incidents, >10% weekly miles growth, fleet already running early v15 builds (40% of seven improvement tracks merged). Still no fleet vehicle count.
- Net income flattered by a $1B mark-to-market gain on SpaceX holdings (a new disclosure — Tesla holds SpaceX equity), offset by ~$300M FX and ~$100M Bitcoin losses.
What management is focused on- Speed over capital efficiency, explicitly. Musk told the team to spend "as fast as we can without it being too wasteful," accepting lower capital efficiency for faster completion as the higher-NPV outcome. Taneja disclosed Tesla acts as general contractor on nearly all its construction. Musk called it "the fastest industrial scale-up since World War II."
- FSD as the demand engine, now with data: ~55% of North American Q2 deliveries had an FSD subscription enabled at delivery; ~1.5M paid FSD customers globally (55% upfront / 45% subscription). Musk: customers are "buying Tesla full self-driving with a car attached."
- Robotaxi safety as the binding constraint. Musk reiterated the asymmetry: 30–40K annual US auto deaths get no press, but one Tesla injury is "worldwide headline news" triggering regulatory clampdown. Elluswamy framed scaling as the "March of 9s" toward 99.999% reliability.
- Optimus realism (of a sort): longest, most caveated ramp language yet — "hardest product to scale manufacturing that we've ever made," no supply chain exists, flat and long initial S-curve. But also the first aspirational volume numbers: ~1M units/year for Optimus 3, ~10M/year for Optimus 4 (Austin, more vertically integrated).
- Power as the AI bottleneck and Tesla's answer: Megapack demand framed around data-center training-run power smoothing (70% power drops in 100ms); Musk claimed US energy output could ~double with batteries alone (1.2–1.3 TW generation vs. 0.5 TW average use). SpaceX named as a major Megapack buyer for its data centers.
- Deepening SpaceX entanglement as strategy, not accident: Digital Optimus partnership (SpaceX's "big model" as task manager), Starlink integrated into Cybercab and all vehicles, Grok in the car, Terafab collaboration, and a formal "investment and framework agreement" earlier this year.
Key numbers and quarter mechanics- Deliveries: record Q2; sequential growth Americas +60%, APAC +27%, EMEA +12%. Largest order backlog since 2023.
- Auto gross margin ex-credits: 16.3% (from 19.2%); ~flat ex the $230M Q1 one-timers. Interest-rate subvention (expensed upfront as revenue offset) rose with rates — a recurring headwind.
- FSD: ~1.5M paid customers (from ~1.3M), 55% upfront / 45% subscription; ~55% NA delivery attach rate; purchase option removed in most markets.
- Energy: 13.5 GWh (+53% QoQ); margin 20.4% (from 39.5%) after ~$240M warranty true-up (vendor cell issues), lapsed >$200M tariff benefit, and ASP declines. Long-term margin guide: mid-to-low 20s.
- Services & Other margin: 14.1%, all-time high (from 9.2%) on volume and fleet cost management; includes deliberate robotaxi infrastructure investment.
- OpEx up sequentially: preproduction ramp costs (Semi, Optimus, Cybercab), AI initiatives, new compute depreciation, and litigation charges. Guided to keep growing through 2026 and beyond.
- FCF negative as guided; CapEx more than doubled QoQ, rising further in H2; FY2026 guide held at >$25B; CapEx to grow for 2–3 more years. New: debt facilities for up to $30B borrowing capacity.
- Net income mechanics: +$1B SpaceX mark-to-market gain, −$300M FX, −$100M Bitcoin.
- Robotaxi: >380,000 unsupervised miles, zero notable incidents, six cities/two states (Elluswamy) vs. seven US markets (Taneja) — a minor internal inconsistency; >10% weekly miles growth claimed and projected to sustain.
Product and launch scorecard- Robotaxi: strongest evidence quarter to date. 380K unsupervised miles, zero notable incidents, multi-city footprint (Florida, Texas, Bay Area), early v15 in fleet, city-launch effort "trending towards zero." Still no vehicle count, rides, revenue, or cost/mile; Langan's "dozens, not hundreds" fleet characterization was not disputed — management answered with miles-per-vehicle utilization logic instead.
- Cybercab: in production; factory rides running in Austin; manufacturing targets aligned to projected unsupervised-miles growth. New disclosure: Cybercab needs chassis-specific driving data via steering-wheel/pedal-retrofitted units before volume deployment — a real gating item. Starlink integration confirmed.
- Semi: production started (Musk, opening remarks — a delivered milestone vs. last quarter's "soon"). Autonomy for Semi guided to end of 2026/early 2027, deliberately deprioritized behind 3/Y/Cybercab.
- Optimus: Fremont line "looks incredible," production starting "soon" (S/X replacement). First volume aspirations: Optimus 3 ~1M/yr; Optimus 4 (Austin, more vertical integration) ~10M/yr — aspirational, heavily caveated. Training strategy detailed: human observation, dedicated data team, internet video, "Optimus Academy" RL flywheel. No production rate, no unveil date reconfirmed.
- FSD v15: seven improvement tracks, ~40% merged in robotaxi fleet builds; consumer timing not re-stated this quarter (prior: end of year/early next). HW3→HW4 upgrade intent reaffirmed and broadened: "financially sensible at some point to upgrade" all camera-equipped cars to a next-gen board (AI4 upgrade ~mid-2027; AI5 volume ~mid-2027, Optimus-first).
- Terafab: location announcement deferred to a standalone event — a slip from prior-quarter momentum; Austin development fab equipment orders placed (mask, logic, memory, packaging, testing under one roof). Samsung (Texas) and TSMC (Arizona) fabs cited as building AI compute for Optimus/robotaxi; Micron thanked for memory allocation.
- Megapack 3: production starting "soon." Lithium refinery and cathode refinery started production; LFP factory operational earlier this year.
- Solar: massive US manufacturing build-out planned (silicon refinement through cells/panels); Taneja claims Tesla will multiply US solar manufacturing "by an order of magnitude."
- Megapod (new product concept): AI4+x86 compute boxes in Megapack-style packaging, deployable at Superchargers (~7 GW of power cited) for distributed AI compute. Early-stage, no economics given.
Sell-side read-through- Five analysts through (Percoco, Perry/BofA, Langan/Wells Fargo, Piecyk/LightShed, Stein/Truist, Levy/Barclays); RBC's Tom lost to audio failure. Say.com questions were absorbed into prepared remarks — a pattern that reduces spontaneous disclosure.
- Langan asked the merger question directly ("SpaceX and Tesla combining"). Musk didn't deny — "we can't talk about combining companies… on an earnings call" — and the GC disclosed a formal investment + framework agreement. The non-denial plus the $1B SpaceX stake mark-to-market will fuel the combination narrative.
- Langan's fleet-size challenge (media reports: dozens of vehicles, not hundreds) drew no number — management pivoted to utilization (continuous operation → high miles per vehicle) and operational-kinks rationale. The fleet count remains deliberately undisclosed for a second straight quarter.
- Percoco extracted supplier co-investment detail: Samsung and TSMC "tens of billions" on US fabs partly dedicated to Tesla AI compute; Panasonic billions on cells; Micron memory allocation "on reasonable terms." Useful evidence that the CapEx burden is partially shared.
- Stein got the first Optimus volume aspirations (1M/yr Gen 3, 10M/yr Gen 4) and confirmation that HW3 upgrades remain planned — but no cost, capacity, or timeline for the retrofit program.
- Levy's CapEx question produced the quarter's most quotable admission: efficiency is deliberately subordinated to speed. No one challenged the $30B debt capacity, the SpaceX related-party economics, or the energy margin collapse beyond prepared remarks.
- Notable absences/evasions: no robotaxi revenue or cost-per-mile; no consumer unsupervised FSD timing update; no Optimus unveil date; no Terafab location/terms (deferred to a future event); no question on the $240M energy warranty true-up's recurrence risk; no EU/China FSD approval update (prior-quarter catalysts, unmentioned this call).
Management credibility- Delivered: Semi production started (vs. "soon" last quarter); Cybercab in production with factory rides; negative FCF and CapEx doubling occurred exactly as guided; energy lumpiness and margin compression occurred as pre-flagged; >$25B CapEx guide held.
- Proactive disclosure credit: Taneja again itemized one-timers in both directions ($230M Q1 benefit not repeating; $240M energy warranty true-up; $200M lapsed tariff benefit) and gave a long-term energy margin range — unusually candid normalization.
- New disclosure regression/inconsistency: Elluswamy said six cities/two states; Taneja said seven US markets — minor, but the fleet vehicle count remains withheld while miles are emphasized, and Langan's "dozens" framing went uncorrected.
- Terafab slipped rhetorically: last quarter brought structure (site, partner, ~$3B research fab); this quarter the location announcement was deferred to a standalone event. Optimus unveil timing went unmentioned — prior "mid-year" commitment now past due with no update.
- The SpaceX non-denial is a credibility fork: refusing to discuss a combination "on an earnings call" while disclosing an investment, framework agreement, Digital Optimus partnership, Starlink integration, and Megapack sales to SpaceX data centers is either appropriate process discipline or deliberate narrative management. Related-party exposure is now financially material ($1B mark-to-market swing).
- Musk's health noted on-call ("a little under the weather") — not material, but he was less dominant than usual.
- Consistency maintained: safety-first robotaxi pacing, S-curve caveats, supply-chain constraints, energy lumpiness — all pre-flagged and recurring.
What changed versus the prior quarter- Demand momentum confirmed and extended: record Q2 deliveries, backlog largest since 2023 (prior: highest Q1 backlog in 2+ years); growth now supply-constrained, not demand-constrained.
- Margin story inverted in both segments: auto ex-credits down to 16.3% (~flat ex-one-timers); energy collapsed to 20.4% with a new long-term mid-to-low-20s normalization guide — the >39% print is now officially an artifact.
- Funding posture escalated: from "negative FCF accepted" to actively securing up to $30B in debt facilities — leverage is now part of the plan.
- Robotaxi disclosure partially restored: miles (380K), incident record, and growth rate disclosed after last quarter's regression — but fleet count still withheld, and the geographic framing shifted from "dozen states" to city-by-city with launch effort "trending towards zero."
- SpaceX relationship formalized and financialized: investment + framework agreement, $1B mark-to-market gain, Digital Optimus partnership, Starlink-in-all-vehicles, SpaceX as Megapack customer — and an explicit non-denial of combination chatter.
- Optimus got volume aspirations for the first time (1M/yr Gen 3, 10M/yr Gen 4) but lost its unveil timeline (mid-year passed, unmentioned).
- Semi moved from "soon" to "in production"; autonomy for Semi explicitly deprioritized ~6 months.
- Terafab went from structured disclosure to deferred announcement; supplier co-investment (Samsung, TSMC, Micron, Panasonic) newly quantified in the tens of billions.
- New products appeared: Megapod distributed AI compute; lithium/cathode refineries in production; solar build-out scaled to "order of magnitude" of US capacity.
Bull case- Demand is supply-constrained, not demand-constrained: record deliveries, +60% Americas sequential growth, largest backlog since 2023, and a ~55% NA FSD attach rate suggesting the FSD-led demand thesis is working with verifiable numbers.
- Robotaxi finally has hard metrics: 380K unsupervised miles, zero notable incidents, >10% weekly compounding, early v15 in fleet, and falling per-city launch effort — the strongest evidence yet that the camera-only approach scales.
- Semi and Cybercab both in production — two S-curves now running simultaneously, with Cybercab manufacturing targets explicitly tied to the autonomy growth curve.
- Energy volume inflected (+53% QoQ, 13.5 GWh) with robust backlog and a structural data-center demand driver (training-run power smoothing) that hyperscalers — including SpaceX — are already paying for.
- Supplier co-investment de-risks the CapEx supercycle: Samsung, TSMC, Panasonic, and Micron are committing tens of billions aligned to Tesla's roadmap, and $30B of debt capacity removes near-term funding risk.
- Services margin at an all-time high (14.1%) shows the non-auto P&L can scale profitably even while absorbing robotaxi infrastructure spend.
Bear case- Earnings quality is thin: auto margin ex-credits at 16.3% with rising rate-subvention drag; energy margin halved with a $240M warranty true-up (vendor cell issues — recurrence risk unaddressed); net income propped by a $1B SpaceX mark-to-market gain.
- The CapEx cycle is now levered: negative FCF, CapEx doubling sequentially, >$25B this year growing for 2–3 more, and up to $30B of debt — with management explicitly deprioritizing capital efficiency. Any demand or autonomy stumble meets a much heavier balance sheet.
- Robotaxi fleet is still "dozens, not hundreds" per unchallenged media characterization; miles-based disclosure substitutes for vehicle counts, revenue, and unit economics, all still absent.
- Optimus volume aspirations (1M/10M units) are unfalsifiable against zero disclosed production rate, a slipped unveil, and Musk's own "hardest scale-up ever" framing.
- Related-party gravity is compounding: SpaceX is now simultaneously an equity holding, framework partner, Megapack customer, AI collaborator, and Terafab counterpart — and management declined to rule out a combination. Governance complexity is rising faster than disclosure.
- Regulatory friction is real: New Jersey state-level sensor requirements called "disheartening"; the federal posture is friendly now but administration-dependent.
- Terafab momentum stalled rhetorically — location deferred to a future event while the chip-supply rationale for it ("without which we will be constrained") grows more urgent.
Next-quarter watchlist- CapEx/FCF: magnitude of H2 CapEx ramp and negative FCF; drawdowns or terms on the $30B debt facilities; any update to the >$25B 2026 guide.
- Robotaxi: fleet vehicle count (does disclosure return?); miles growth vs. the >10%/week claim; v15 track-merger progress (40% → ?); consumer unsupervised FSD timing (prior guide: Q4); any revenue, pricing, or cost-per-mile disclosure; New Jersey-type state regulatory outcomes.
- Cybercab: retrofitted test-fleet miles accumulation; timing of steering-wheel-free volume deployment; Starlink integration rollout.
- FSD: EU-wide and China approvals (prior-quarter catalysts that went unmentioned — did they stall?); subscription net adds from 1.5M; HW3 retrofit program specifics (cost, capacity, timing) now that upgrades are reaffirmed.
- Margins: auto ex-credits trajectory with rate subvention; energy margin vs. the new mid-to-low-20s long-term guide; any recurrence of vendor-cell warranty charges.
- Optimus: production start evidence at Fremont; V3 unveil (now overdue); any actual rate data; Optimus 4/Austin timeline.
- Terafab: the standalone location/details announcement; Austin development fab construction; AI5 volume timing (~mid-2027); Samsung/TSMC dedication specifics.
- SpaceX: any combination-process movement; framework agreement terms; related-party transaction scale (Megapack sales, Digital Optimus economics).
- Semi: production ramp rate; autonomous Semi development start (guided end of 2026/early 2027).
|
| Apr 22, 2026 | -3.56% | Q1 FY2026 | Read transcript briefingQuarter in one view- Demand rebounded and margins expanded again. EMEA deliveries grew >150% QoQ in France and Germany; APAC grew (South Korea, Japan); US grew slightly. Highest Q1 order backlog in over 2 years. Auto gross margin ex-credits rose 17.9% → 19.2%, though ~$230M of warranty true-downs and tariff relief were one-time benefits. Giga Berlin hit a record >61,000 units.
- CapEx guidance jumped again: >$25B for 2026 (from >$20B last quarter), with Taneja explicitly guiding to negative free cash flow for the rest of the year. FCF was ~$1.4B again. Orders already placed for the Austin research fab and solar manufacturing equipment.
- Robotaxi expanded to Dallas and Houston (three cities now, all on v14.3 variants), with "a dozen states or more" targeted by year-end — a walk-back from last quarter's "quarter to half of the United States." Musk conceded robotaxi revenue "would not be super material this year," material "probably in a significant way next year."
- Hardware 3 is officially dead-ended for unsupervised FSD — 1/8 the memory bandwidth of HW4; remedy is discounted trade-ins or computer+camera retrofits via new "micro factories." This is a major stranded-asset and customer-goodwill event.
- TerraFab got structure: Tesla builds a ~$3B research fab at Giga Texas (construction starting this year, a few thousand wafer starts/month); SpaceX takes the initial phase of the scaled-up TerraFab; Intel's 14A process named as the manufacturing partner. Related-party board approvals acknowledged as complex.
- Energy was lumpy as warned: 8.8 GWh deployed, -38% QoQ, but record >39.5% gross margin aided by >$250M one-time tariff recognitions. Normalized margin compression still guided.
What management is focused on- The CapEx supercycle as identity. Musk opened by framing Tesla alongside "major technology companies substantially increasing capital investments." Taneja: "a very big capital investment phase… going to start now and would last a couple of years," with negative FCF accepted as "the right strategy."
- FSD as the product, the car as the delivery mechanism. Taneja formally stated the evolved sales strategy: "we now emphasize FSD as a product and vehicle as only the delivery mechanism." Paid FSD base reached ~1.3M (from 1.1M), growth mostly subscriptions; upfront purchases +7% as the buy option is removed in some markets.
- Regulatory unlocks as the demand lever. Netherlands FSD approval with EU-wide review in Brussels in May (EU-wide approval hoped for later in Q2); China partial approval with full approval hoped by Q3. Management explicitly ties these to fleet adoption and incremental vehicle demand.
- Safety-first robotaxi pacing. Repeated emphasis: zero injuries/fatalities to date, "we don't want to have a single accidental injury." Ashok Elluswamy disclosed the real scaling bottlenecks are convenience/edge-case behaviors (cars getting "stuck," infinite loops around construction, paralysis at railroad crossings) rather than direct safety failures.
- Optimus secrecy as strategy. V3 unveil pushed to "middle of this year" and possibly closer to production because competitors "do a frame-by-frame analysis and copy everything." Fremont production start now framed as late July/August; second Optimus factory at Giga Texas starting ~summer 2027.
- Chip self-sufficiency. AI5 taped out (early, via "every weekend for 6 months"); AI5 redirected to Optimus and data centers, not vehicles (AI4 deemed sufficient for unsupervised); AI4.1/AI4+ upgrade (32GB/SoC, ~10% more compute) planned with Samsung for mid-2027; Dojo 3 ideation begun.
Key numbers and quarter mechanics- Auto gross margin ex-credits: 19.2%, up from 17.9% — but flattered by ~$230M warranty true-downs and tariff relief. No benefit yet realized from the Supreme Court IEEPA tariff ruling (uncertainty cited). Tariffs and high interest rates remain cost headwinds; interest-rate subvention is expensed upfront and will keep pressuring margins if rates rise.
- Deliveries: growth in all regions — France/Germany >150% QoQ, South Korea/Japan growth, slight US growth. Gas prices helped order rate but improvement predated the uptrend. Model 3 starting price now below the inflation-adjusted $48K of the original $35K promise.
- FSD: ~1.3M paid customers (+~180K in the quarter per Ferragu's math, unchallenged); subscriptions driving growth; churn declining; usage duration rising. V14 for HW3 (distilled) coming end of June.
- Energy: 8.8 GWh deployed, -38% QoQ (lumpiness pre-flagged); gross margin >39.5% (record) with >$250M one-time tariff recognitions; FY2026 deployments still guided above 2025; Megapack 3 production starting later this year at the new Houston-area factory.
- Services & Other margin: 9.2%, up from 8.8% — includes robotaxi; deliberate infrastructure investment continues.
- OpEx up sequentially: first full quarter of stock comp for the 2025 CEO award (one milestone still deemed probable), plus AI5 development and Cybercab/Semi/Optimus/Megablock spend; elevated all year.
- FCF: just over $1.4B; guided negative for the rest of 2026. CapEx guide raised to >$25B (from >$20B), covering six factories, AI infrastructure, research fab, and solar equipment orders.
- Bitcoin: -22% mark-to-market hit to net income; FX losses on intercompany balances again.
- Battery packs remain the top production constraint (not cells): Berlin 4680 packs ramping, Reno retooling for more output, China in-house LFP module/pack growth.
Product and launch scorecard- Cybercab: production started — on time versus the April commitment. Musk immediately caveated with the stretched S-curve: very slow initial production, exponential ramp "towards the end of the year and certainly next year." Reiterated as the long-term volume leader (>90% of miles are ≤2 occupants).
- Semi: production "soon," same S-curve caveat. No date, no rate.
- Robotaxi: expanded to Dallas and Houston on v14.3 variants; fleet grew QoQ (no number given — a disclosure regression from last quarter's ">500"). Target now "a dozen states or more" by year-end. Zero incidents claimed; NHTSA filings addressed as showing zero incidents. Revenue explicitly not material in 2026.
- FSD: v14.3 called "the last piece of the puzzle" for unsupervised, but large-scale deployment deliberately held for v15 (full architectural overhaul, runs on AI4, "hopefully by end of this year, certainly by early next year"). Consumer unsupervised FSD guessed at Q4, geography-by-geography. Netherlands approved; EU review in May; China hoped by Q3.
- Hardware 3: cannot run unsupervised FSD — ever. Retrofit requires new computer and cameras, necessitating dedicated metro "micro factories." Discounted trade-ins offered to FSD purchasers. This is a multi-year cost and liability overhang now formally acknowledged.
- Optimus: V3 unveil pushed to mid-year (from "a few months" last quarter — a slip), possibly delayed further toward production for competitive secrecy. Fremont start ~late July/August; S/X final assembly ends early May, line conversion takes months. Musk refused any 2026 production-rate estimate ("impossible to predict… 10,000 unique items"). Second factory at Giga Texas ~summer 2027. External usefulness "sometime next year."
- AI5: taped out early; redirected away from vehicles to Optimus/data centers — a notable roadmap change flagged by the say.com question. AI4.1 upgrade with Samsung (newer RAM, 64GB total, +10% compute) targeted mid-2027.
- TerraFab: first concrete details. Tesla research fab ~$3B at Giga Texas, construction this year, few thousand wafer starts/month, mask+logic+memory+packaging under one roof. SpaceX takes initial scaled-up phase; Intel 14A process partnership. Governance: dual-board conflict-resolution process acknowledged as slow and complex.
- Roadster: debut "in a month or so" (from "hopefully April" — slight slip), admitted to be revenue-immaterial.
- Solar: residential market in correction post-tax-credit loss; Tesla lease product launched to capture the credit itself; own panel and mounting system debuted. H2 demand recovery expected.
Sell-side read-through- Five analysts through (Stein/Truist, Ferragu/New Street, Levy/Barclays, Delaney/Goldman, Langan/Wells Fargo). Say.com questions did heavy lifting on Optimus timing, HW3, and AI5.
- Stein extracted the TerraFab structure — the most consequential disclosure of the call: Tesla research fab (~$3B), SpaceX building the initial scaled fab, Intel 14A. The SpaceX role introduces a new related-party financing channel that went unchallenged on terms, pricing, or Tesla shareholder protections beyond "independent director reviews."
- Ferragu's FSD math went unrefuted: ~180K net adds vs. ~100K North American cars sold implies most new HW4 owners in NA are subscribing — the strongest evidence yet that the subscription pivot is working. Elluswamy added churn is falling and engagement rising.
- Levy tested the TerraFab economic motive and got a clean denial: it's supply security, not supplier leverage, plus "long shot" research upside. No yield/cost-parity timeline offered.
- Delaney asked whether new conventional models are dead. Answer: effectively yes — Cybercab is the compact car, the future lineup is "almost entirely autonomous," Roadster is the sole manual exception. The affordable-car question is now closed.
- Langan pressed safety metrics and the NHTSA camera-glare filing. Got process (QA fleet, simulation, intervention analysis) but no actual metrics — no miles-per-intervention or disengagement data. Moravy confirmed cameras were changed "some months ago" and NHTSA is asking for "quite a bit of information"; resolution promised "in short order." Stricter camera-visibility gating now disables FSD when occluded.
- Notable evasions/absences: no robotaxi fleet number, rides, revenue, or cost/mile (and now a revenue-materiality deferral to 2027); no Optimus production rate; no challenge to the $25B CapEx jump or negative-FCF guidance; no question on the HW3 retrofit program's cost or the ~1.1M upfront-FSD purchasers' legal exposure; no follow-up on xAI investment terms; the "dozen states" target replacing "quarter to half of the US" passed without comment.
Management credibility- Delivered: Cybercab production started in April as committed. Robotaxi expanded to two new cities with zero incidents claimed. AI5 taped out early. FSD approvals in Netherlands and (partial) China are verifiable regulatory facts.
- Delivered with asterisks: the 19.2% margin — real, but $230M warranty true-downs and tariff relief are one-timers; Taneja disclosed them proactively, which is to his credit, but the underlying run-rate is lower than the print.
- Guidance walked back without acknowledgment: last quarter's "fully autonomous vehicles in a quarter to half of the United States by year-end" became "a dozen states or more." Last quarter's fleet "doubling every month" claim disappeared; this quarter gave no fleet number at all. Robotaxi revenue went from implied-near-term to "not super material this year."
- Optimus timeline slipped again: V3 unveil "a few months" (Q4) → "middle of this year," possibly later; production start now late July/August with zero rate commitment. Musk's 4-month line-conversion defense ("insanely fast… no other company on earth") is pre-emptive excuse-making.
- HW3 admission is a credibility wound years in the making: customers were sold FSD on hardware now declared permanently incapable of unsupervised driving. The retrofit/trade-in remedy is vague on cost, capacity, and timing.
- CapEx discipline is the open question: guide raised ~25% in one quarter ($20B → $25B), negative FCF now explicit, and the scaled TerraFab is being routed through SpaceX — a structure where Tesla shareholder exposure is genuinely unclear.
- Consistency maintained on: energy lumpiness/margin compression (predicted, occurred), S-curve caveats on new products, battery-pack constraint (flagged for multiple quarters).
What changed versus the prior quarter- Demand narrative flipped: from post-credit air pocket (-16% QoQ) to broad-based growth and the highest Q1 backlog in 2+ years. The Q4 hangover is over, at least in EMEA/APAC.
- CapEx guide raised from >$20B to >$25B, and the funding posture hardened from "internal resources first" to explicit acceptance of negative FCF for the rest of 2026.
- Robotaxi scaled (2 → 3+ cities) but ambitions narrowed: "quarter to half the US" → "a dozen states"; revenue materiality pushed to 2027; fleet disclosure regressed from ">500" to none.
- HW3 end-of-life formally admitted — a new, permanent liability category that didn't exist in prior disclosures.
- TerraFab went from vague intent to a structured program with a price tag on phase one (~$3B), a partner (Intel 14A), a site (Giga Texas), and a surprising related-party builder (SpaceX).
- AI5 exited the vehicle roadmap — AI4 declared sufficient for unsupervised; AI5 goes to Optimus and data centers; AI4.1 bridge planned with Samsung.
- FSD subscription pivot is working: 1.1M → 1.3M paid, subscription-led, churn falling; Europe and China approvals moved from "none of the record countries have it" to concrete regulatory paths (EU May review, China Q3 hope).
- Energy printed the guided lumpiness: -38% QoQ deployments, record margin on one-time tariff items.
- Optimus unveil slipped and secrecy rationale introduced; second Optimus factory (Giga Texas) is new.
- Musk comp accrual is now a full-quarter recurring OpEx line — normalized, no longer news.
Bull case- Demand and margin simultaneously inflected up: >150% QoQ growth in France/Germany, record Berlin output, highest Q1 backlog in 2+ years, and 19.2% ex-credit margin — even ex-one-timers, the underlying margin is likely above Q4's 17.9%.
- The FSD flywheel is showing real economics: 1.3M paid subs, subscription-led growth, falling churn, rising engagement, and Ferragu's unchallenged math implying majority attach rates among new HW4 owners in North America. EU and China approvals would open the two largest untapped fleets.
- Cybercab started production on schedule — the first time a Tesla autonomy-critical launch date has been hit. If the S-curve executes into year-end, 2027 volume math changes entirely.
- Robotaxi is now multi-city with a zero-incident record and a stated willingness to sacrifice speed for safety — the regulatory goodwill this builds is a real asset.
- Energy margins >39% even if one-time-flattered, backlog robust, Megapack 3/Houston factory coming, FY deployments still guided up.
- AI5 taped out early and AI4 is declared sufficient for unsupervised — the chip roadmap is de-risked near-term, and the Intel 14A partnership gives TerraFab a credible process foundation rather than a from-scratch fantasy.
Bear case- Negative FCF is now guided for the rest of 2026 on a >$25B CapEx year that was >$20B ninety days ago — with the scaled TerraFab, solar fab, and second Optimus factory still largely unpriced. The trajectory of the number matters more than the number.
- Robotaxi's financial disclosure went backwards: no fleet count, no rides, no revenue, and an explicit "not material this year" — while the valuation increasingly rests on it. The geographic target was quietly cut from ~half the US to ~a dozen states.
- HW3 is a stranded installed base: potentially hundreds of thousands of FSD purchasers own hardware that can never run unsupervised FSD. Retrofit requires cameras + computer via yet-to-be-built "micro factories" — unpriced, unscheduled, and a litigation/brand risk.
- The 19.2% margin contains ~$230M+ of one-timers, interest-rate subvention is a structural drag if rates stay high, and energy margin compression is still guided — quality of earnings is softer than the headline.
- Optimus slipped again (unveil mid-year, production late July/August, no rate commitment, "10,000 unique items"), and the competitive-secrecy rationale for hiding V3 is unfalsifiable.
- SpaceX building the scaled TerraFab inserts a Musk-controlled related party into Tesla's largest-ever capital program, with terms subject to a conflict process Musk himself described as burdensome — governance risk is compounding, not resolving.
- NHTSA overhang is real: multiple filings referenced (incident data, camera glare), "quite a bit of information" requested, and the zero-incident claim sits alongside a QA-fleet validation process that discloses no metrics.
Next-quarter watchlist- CapEx and FCF: does the >$25B guide hold or rise again; magnitude of negative FCF in Q2; any debt issuance, fleet financing, or SpaceX/Tesla TerraFab term sheet.
- Robotaxi: fleet count disclosure (does it return?); new state launches vs. the "dozen states" target; any revenue/cost-per-mile metric; incident record as the fleet scales; v15 timing ("end of this year, certainly early next").
- Cybercab ramp: line rate exiting Q2; whether "exponential towards end of year" shows up in production data; Semi production start.
- FSD: EU-wide approval after the May Brussels review; China approval by Q3; consumer unsupervised launch (guided Q4); HW3 V14 distilled release end of June; retrofit program pricing, capacity, and micro-factory locations; subscription net adds and churn.
- Margins: ex-credit auto margin ex-one-timers; tariff trajectory post-IEEPA ruling; interest-rate subvention drag; energy margin compression magnitude as the $250M tariff benefit lapses.
- Optimus: V3 unveil (mid-year — does it slip again?); Fremont line conversion progress after early-May S/X end; first production units and any rate data; Giga Texas factory construction.
- TerraFab: research fab construction start; Intel 14A agreement specifics; SpaceX phase structure, funding split, and related-party terms; AI4.1/Samsung progress.
- Roadster debut ("a month or so") — a sentiment catalyst, admitted revenue-immaterial.
- NHTSA: resolution of the camera-glare information request and any new filings as unsupervised miles grow.
|
| Jan 28, 2026 | -3.45% | Q4 FY2025 | Read transcript briefingQuarter in one view- The Q4 air pocket arrived — and margins absorbed it. Deliveries fell 16% sequentially (post-credit-expiry hangover Taneja now explicitly confirms: Q3 "pull[ed] in some demand from Q4"), yet auto gross margin ex-credits jumped 15.4% → 17.9% on regional mix (more APAC/EMEA), and total gross margin hit 20.1% — best in over two years. Auto gross profit was flat despite the volume drop.
- The call's real headline is the pivot to an industrial conglomerate. Model S/X are being discontinued (production winds down next quarter) to convert Fremont into a million-unit Optimus line; 2026 CapEx guided to >$20B (vs. ~$9B in 2025) across six factories — refinery, LFP, Cybercab, Semi, new Megafactory, Optimus — plus AI compute. Excluded from that number: the solar-cell fab (100 GW/yr ambition) and the "TerraFab" (integrated logic/memory/packaging fab) Musk says Tesla "has to" build.
- Robotaxi crossed the unsupervised threshold. Paid rides with no safety monitor, no chase car, in Austin began in the last couple of weeks; fleet "well over 500" paid-customer vehicles across Austin and Bay Area, claimed to "probably double every month." Guidance: fully autonomous vehicles in "a quarter to half of the United States" by year-end, pending regulation; "dozens of major cities" even without federal preemption.
- FCF fell to $1.4B (from ~$4B in Q3) and OpEx rose ~$500M sequentially, including the first charges under Musk's 2025 performance award (deemed probable) and elevated AI/new-product spend. Bitcoin mark-to-market was a headwind (holdings depreciated 23%).
What management is focused on- "Amazing abundance" as the new corporate mission — Musk formally updated Tesla's mission statement, framing AI/robotics as leading to "universal high income." This is narrative scaffolding for the S/X cancellation and the Optimus/Cybercab pivot.
- Autonomy as the entire future of the vehicle business. Musk: long-term "the only vehicles that we'll make will be autonomous vehicles with the exception of the next generation Roadster" (debut targeted April). Cybercab production starts April; expected to eventually exceed all other Tesla vehicles combined "several times" over, justified by the stat that >90% of miles are traveled with ≤2 occupants.
- Supply-chain sovereignty as existential strategy. TerraFab (logic + memory + packaging, domestic), lithium refinery (Corpus Christi, "most advanced in the world"), cathode refinery (Austin), LFP factories, 100 GW solar-cell ambition — all framed around geopolitical paranoia ("only the paranoid survive"). Musk claims zero advanced memory fabs at scale exist in the US and that Tesla will be chip-supply-limited in 3–4 years without its own fab.
- Optimus as the GDP-scale prize. Optimus 3 unveil "in a few months"; million-unit/year line in the Fremont S/X space; Musk concedes a "stretched-out S curve" because the supply chain is entirely new, and significant production volume not until "probably the end of this year."
- FSD monetization model change. Transitioning fully to subscription-only FSD starting this quarter — Taneja flags this will pressure auto margins near-term. Paid FSD base: nearly 1.1M globally, ~70% historical upfront purchases.
- xAI investment defense. Announced today; justified as "master plan four" furtherance, Grok already in vehicles, and — notably — "a lot of investors ask us to do this… we're just doing what shareholders would ask."
Key numbers and quarter mechanics- Deliveries: -16% QoQ (no absolute unit number given on the call). Auto gross profit flat sequentially despite the drop, on richer APAC/EMEA mix. Record deliveries cited in Malaysia, Norway, Poland, Saudi Arabia, Taiwan; year-end backlog "bigger than in recent years" — and Taneja notes none of those countries have the latest FSD yet.
- Auto gross margin ex-credits: 17.9%, up from 15.4% — a large sequential jump attributed to regional mix, not cost. Total gross margin 20.1%, first time above 20% in over two years.
- Energy: record deployments and record gross profit again; FY2025 energy revenue ~$12.8B, +26.6% YoY. 2026 guided to growing deployments (Megapack 3 and Megablock launches) but with margin compression from low-cost competition, policy uncertainty, and tariffs.
- Services & Other margin: 10.5% → 8.8% on service-center hiring ahead of fleet growth; Supercharging margin improved; robotaxi costs still immaterial and buried here — "revenue and cost per mile metrics are not meaningful to discuss at the moment."
- OpEx +$500M QoQ: stock-based comp (including first accrual for Musk's 2025 performance award, "deemed probable"), AI initiatives, Cybercab/Semi/Optimus/Megapack spend — guided to stay elevated all of 2026.
- FCF: $1.4B (vs. ~$4B Q3). FY2025 CapEx came in slightly below the $9B guide. 2026 CapEx: >$20B, excluding solar fab and TerraFab.
- Cash and investments: >$44B. Funding plan: internal resources first; bank financing against robotaxi fleet cash flows ("conversations with banks" already held); debt or "other means" contemplated for the longer-tail infrastructure plays.
- Bitcoin: -23% mark-to-market headwind to net income (vs. +$80M gain in Q3); FX losses on intercompany borrowings also hit.
- Tariffs: "in excess of" $500M impact in Q4 (transcript is garbled here — the figure appears attached to OpEx commentary; directionally, tariff costs remain elevated and are cited as a 2026 energy margin headwind).
- Battery packs remain the biggest global production constraint; workaround is 4680 cells in non-structural packs.
Product and launch scorecard- Robotaxi: the quarter's biggest de-risking event. Unsupervised paid rides in Austin (no monitor, no chase car) began in the final weeks of the quarter — delivering, late but delivered, on Q3's "large parts of Austin by year-end" commitment. Fleet "well over 500" paid vehicles (Austin + Bay Area) vs. ~200 per public tracking; Musk claims monthly doubling. Still undisclosed: rides, revenue, cost/mile (explicitly deferred as "not meaningful"), intervention rates. New-market cadence now framed as "dozens of major cities" by year-end.
- Cybercab: production start firmed to April 2026 (Q3 said "Q2 2026" — consistent). No steering wheel/pedals reiterated; designed for 50–60 hr/week duty cycles vs. ~10 for driven cars; larger autonomous variants teased (prototypes shown, rides given). Musk: long-term Cybercab volume "several times" all other vehicles combined.
- Model S/X: discontinued. Production winds down next quarter; Fremont space converts to the million-unit Optimus line. An "honorable discharge" — strategically coherent with the autonomy pivot but removes the flagship price umbrella.
- Optimus: still R&D-stage, timeline held (barely). No material factory deployment ("not in usage in a material way"); old versions deprecated as new ones iterate. Optimus 3 unveil "in a few months" (Q3 said Q1, Feb–Mar — roughly consistent); significant volume "probably the end of this year" (Q3 said "towards the end of next year" — now pulled back to end-2026, i.e., re-accelerated on paper). Musk concedes the stretched S-curve and single-part-supply-chain fragility. China named as "by far the biggest competition"; Tesla's claimed edge: hand dexterity, real-world AI, manufacturing scale — "the only company that has all three."
- FSD: subscription-only transition begins now — a structural revenue-recognition and margin change. 1.1M paid customers (~70% upfront historically). Musk claims FSD "is 100% unsupervised" (in Austin robotaxis) while consumer rollout remains deliberately cautious, city-by-city (Wilshire/Santa Monica intersection anecdote). Driver monitoring to be reduced "proportionate to the safety of the FSD build" per release.
- AI5/AI6 and TerraFab: AI5 design "quite confident," Musk spending Saturdays + part of Tuesdays on it; AI6 aspirationally <1 year behind. AI4 already used in Tesla data centers alongside NVIDIA. TerraFab announced as intent (logic + memory + packaging, domestic) with "a bigger announcement in the future" — no cost, timeline, or site.
- Roadster: next-gen debut "hopefully" April — the sole non-autonomous vehicle exception.
- Cybertruck: line to transition to fully autonomous (local cargo delivery use case); claimed segment leader among electric trucks.
- Energy products: Megapack 3 and Megablock launching into 2026 deployment growth; 100 GW/yr solar-cell manufacturing ambition (raw materials to panels) — early phase, no numbers.
Sell-side read-through- Five analysts got through (Rosner/Wolfe, Andrew/Morgan Stanley, Levi/Barclays, Gianarikas/Canaccord, Langan/Oppenheimer). One say.com question skipped as off-topic; Andrew's follow-up was lost to a mute error.
- Rosner asked the most financially consequential question: dimension the >$20B CapEx, one-time vs. ongoing, and funding. Taneja's answers: six factories + compute + existing-capacity expansion; infrastructure plays (solar/chip fabs) are longer-tail and *excluded* from the $20B; funding = $44B cash, bank debt against robotaxi fleet cash flows, and potentially "more debt or other means." This is the first time Tesla has floated external financing for the buildout — a meaningful signal that >$20B is a floor, not a ceiling.
- Andrew (MS) pressed the xAI investment — the governance-sensitive item. Got a strategic rationale (Grok in cars, fleet orchestration, Optimus "orchestra conductor") plus Musk's striking justification that shareholders asked for it. No terms, valuation, or size disclosed on the call.
- Levi probed memory constraints and extracted the clearest forward-risk statement of the call: Tesla has logic/memory solutions for "roughly three years," beyond which it is supplier-limited — the explicit premise for TerraFab. Musk's "intelligence per gigabyte… order of magnitude better" claim is unverifiable as stated.
- Gianarikas got the China-competition concession: humanoid competition will come "by far" from China; no significant competitors seen elsewhere.
- Langan's R&D synergy question produced philosophy, not numbers — vertical integration "out of desperation," geopolitical hedging, and a plea for others to build refineries. No R&D budget dimensioning.
- Notable absences/evasions: no analyst challenged the 17.9% ex-credit margin's mix-driven sustainability; no one asked for robotaxi revenue/rides after Taneja pre-emptively declared the metrics "not meaningful"; no question on S/X wind-down costs or demand; no follow-up on the Musk comp-award accrual now hitting OpEx; the "quarter to half of the United States" autonomy claim went unchallenged.
Management credibility- Delivered on the hard one: unsupervised paid robotaxi rides in Austin (no monitor, no chase car) began within days of Q3's year-end commitment — slightly late, but real and verifiable by customers. Fleet disclosure (>500 paid vehicles) is new specificity, though it contradicts the ~200 public-tracking estimate without reconciliation.
- Margin delivery was strong: ex-credit auto margin +250bps sequentially on a 16% delivery decline; 20.1% total gross margin; FY CapEx under guide. Taneja's Q3 pull-forward warning proved accurate and he owned it.
- Slippage and re-acceleration whiplash on Optimus: Q2 said production early 2026 → Q3 said "towards the end of next year" (late 2026) → Q4 says significant volume "probably the end of this year" (end-2026) while simultaneously warning of a stretched S-curve and admitting zero material factory deployment today. The Fremont million-unit line is a concrete commitment; the ramp math is not.
- Escalating scope with deferred specifics: TerraFab, 100 GW solar, six simultaneous factories — all announced with no cost, timeline, or site detail, and explicitly excluded from the >$20B guide. The pattern of announcing the next capital-intensive frontier before the current one shows revenue continues.
- Claim inflation persists: FSD "is 100% unsupervised" (true only for a small Austin fleet); autonomy in "a quarter to half of the United States" by year-end; fleet "doubling every month"; intelligence density "an order of magnitude" ahead; Optimus moving US GDP. These are aspirations stated as trajectories.
- Governance items now hit the P&L: the 2025 performance award is being accrued as probable — the comp-vote overhang from last quarter is now a recurring OpEx line. The xAI investment, defended partly as shareholder-requested, deserves scrutiny on terms not disclosed.
What changed versus the prior quarter- The Q4 air pocket question was answered: deliveries -16% QoQ, confirming the pull-forward hangover — but mix-driven margin expansion (17.9% ex-credits) meant the P&L barely noticed. The "rough quarters" framing is now obsolete; management's tone is maximal expansion.
- Robotaxi went from supervised-scale to unsupervised-first-rides, and from "8–10 metros by year-end" (missed — still two markets) to a far larger claim: dozens of cities and up to half the US by end-2026. Fleet size finally disclosed (>500).
- S/X cancellation is the quarter's structural break — Tesla is explicitly becoming an autonomy/robotics company with a shrinking conventional car lineup; Cybercab is slated to outproduce everything else combined.
- CapEx went from "substantially higher" to a number: >$20B, more than double 2025, with solar fab and TerraFab still unpriced on top. Funding language shifted from pure internal cash to banks and potentially debt.
- FSD business model changed: subscription-only from this quarter, with an explicit near-term margin warning; paid base disclosed at ~1.1M (vs. ~12% penetration disclosed last quarter — now an absolute number).
- Optimus timeline re-accelerated on paper (end-2026 volume) while the factory commitment became concrete (Fremont conversion) and the competitive framing shifted to China.
- Energy guidance turned two-sided: record quarter, but 2026 margin compression explicitly guided — new caution absent from Q3's unqualified strength narrative.
- Musk's comp award moved from vote to accrual, and Tesla became an xAI investor — the governance/financial entanglement with Musk's other ventures deepened.
Bull case- Margin resilience through the demand air pocket: 17.9% ex-credit auto margin and 20.1% total gross margin on 16% lower deliveries — the strongest evidence yet that mix and cost control can defend profitability without credit-expiry volume.
- The core autonomy milestone is now fact, not promise: paid, unsupervised, chase-car-free rides in Austin, a >500-vehicle paid fleet, existing charging/service infrastructure as a scaling moat, and a consumer fleet of millions of AI4 cars as latent supply (Airbnb-style fleet opt-in could make ownership cash-flow-positive for customers).
- Cybercab has a date (April), a design logic (cost/mile, 5–6x duty cycle), and a demand thesis (>90% of miles are ≤2 occupants). If the S-curve executes, Tesla's volume ceiling resets upward by multiples.
- Energy is a proven second engine — $12.8B FY revenue, +26.6%, record gross profit — with Megapack 3/Megablock launching into a strong, diversified backlog.
- Balance sheet funds the pivot: >$44B cash, $1.4B FCF even in a transition quarter, and identified financing levers (fleet-backed bank debt) before any equity.
- Vertical integration (lithium, cathode, LFP, chips, solar) is a genuine moat if geopolitical disruption materializes — Tesla would be among the least supply-exposed Western manufacturers.
Bear case- The >$20B CapEx year (with fabs unpriced on top) is being committed before robotaxi shows a single dollar of disclosed revenue or unit economics — "not meaningful to discuss" remains the entire financial disclosure for the business the whole valuation now rests on. FCF already fell to $1.4B; the trajectory points toward cash burn and external financing.
- The 8–10-metro year-end commitment was missed (still two markets), and the response was to raise the target to "dozens of cities" and "a quarter to half of the US" — guidance inflation in the face of a miss.
- The 17.9% margin is mix-driven, not structural — Taneja attributed it to regional mix, and the FSD subscription shift plus energy margin compression are explicit 2026 headwinds. If mix normalizes, margins retrace.
- Optimus is a million-unit factory bet on a product with zero material deployment, an unfrozen design, an entirely new supply chain, and admitted S-curve fragility — against Chinese competitors Musk himself calls "next level."
- TerraFab is the largest scope expansion in company history — logic, memory, and packaging fabs are among the hardest industrial undertakings in existence, announced with no cost, partner, or timeline, justified by a 3–4-year-out supply constraint.
- Governance entanglement deepened: comp-award accruals now flowing through OpEx, plus an xAI investment with undisclosed terms defended as shareholder-requested.
- S/X cancellation and subscription-only FSD both pressure near-term auto economics, and the demand hole in the US post-credit was papered over by small-country records, not core-market strength.
Next-quarter watchlist- Cybercab April production start: does it happen on time; initial line rate; any cost-per-mile disclosure now that unsupervised rides are live.
- Robotaxi scaling evidence: fleet count vs. the "doubling every month" claim; new city launches and regulatory approvals (federal preemption progress); first revenue/rides/cost-per-mile disclosure — Taneja's "not meaningful" defense has a shelf life; any incident disclosure now that monitors are removed.
- Q1 2026 deliveries and US demand: does the backlog claim hold; incentive levels; whether the 17.9% ex-credit margin survives mix normalization and the FSD subscription transition's margin drag.
- S/X wind-down: exit costs, Fremont conversion timeline, and any demand signal from the final-order push.
- Optimus 3 unveil ("a few months"): hand dexterity evidence, task-learning demos, and whether "end of this year" volume holds; first units-per-week line data from Fremont.
- CapEx granularity and funding: breakdown of the >$20B; first numbers for solar fab and TerraFab ("bigger announcement in the future"); any debt issuance or fleet-financing facility; FCF trajectory.
- FSD subscription economics: net adds under the new model, churn, pricing; consumer unsupervised rollout cadence and driver-monitoring reductions; international FSD approvals (Taneja flagged none of the record-delivery countries have it).
- Energy margin compression: magnitude; Megapack 3/Megablock launch execution; tariff cost trajectory vs. Q4's elevated level.
- xAI investment terms: size, valuation, governance, related-party safeguards.
- AI5/AI6: tape-out timing at TSMC Arizona/Samsung Texas; TerraFab site, partners, and cost framework; memory supply agreements.
|
| Oct 22, 2025 | +2.28% | Q3 FY2025 | Read transcript briefingQuarter in one view- The predicted Q4 air pocket never showed up in Q3 — records everywhere. Q3 set records for deliveries, deployments, total revenue, energy gross profit and margins, and free cash flow of ~$4B. The IRA credit expiry clearly pulled demand forward: Greater China/APAC +33%, North America +28%, EMEA +25% sequentially. The "rough quarters" warning was *not* retracted but also not re-discussed for Q4.
- Robotaxi scaled meaningfully and got hard KPIs. Austin: >250,000 miles with no one in the driver's seat; Bay Area: >1M miles with safety driver; three Austin geofence expansions. Musk committed to removing safety drivers in "large parts of Austin by the end of this year" and operating in 8–10 metro areas (Nevada, Florida, Arizona named) by year-end. This is a material step up from Q2's 7,000 miles / "handful of vehicles" — but no fleet size, ride counts, intervention rates, or revenue were disclosed.
- Musk declared "clarity" on unsupervised FSD and pivoted to aggressive capacity expansion. "100% confident" unsupervised FSD can be solved at superhuman safety; aspirational 3M-unit annualized rate within ~24 months; Cybercab production starting Q2 2026 (first explicit start date). This is the largest strategic pivot of the call — from caution to expansion — with 2026 CapEx guided "substantially" higher.
- Governance dominated the close. Musk framed the Nov 6 compensation vote as about voting control ("mid-twenties approximately") to protect his direction of the "robot army," and attacked ISS/Glass Lewis as "corporate terrorists." Taneja made closing remarks urging votes for the comp proposals and three directors.
What management is focused on- Unsupervised FSD as the unlock for everything. Musk: "I feel confident in expanding Tesla's production… as fast as we reasonably can" now that he sees "clarity" on unsupervised FSD. The killer-app framing: texting while the car drives itself.
- Robotaxi expansion cadence. Safety-driver removal in Austin within "a few months," 8–10 metros by year-end, and a formalized new-market playbook (~3 months with a safety occupant before removal).
- Optimus as the "infinite money glitch." New vocabulary: Optimus = "sustainable abundance," working becomes optional, 5x human productivity. Production-intent V3 prototype to be shown in Q1 (February or March), million-unit line with production start "towards the end of next year" — note this is a slip from last quarter's "production start early 2026."
- AI5 as a strategic asset. 40x better than AI4 "by some metrics" (Musk later hedged to "10x improvement" to "say safely"); dual-sourced at TSMC Arizona and Samsung Texas; explicit goal of oversupply with excess chips going into Tesla data centers alongside (not replacing) NVIDIA.
- Energy growth narrative tied to AI/datacenter demand. Megapack/Megablock demand "really strong into next year"; hyperscaler interest; Megapack 4 integrating substation functionality (35kV direct output); new Buffalo-made residential solar panel shipping Q1.
- The Nov 6 shareholder vote. Repeated, explicit campaigning for the compensation proposals and director re-elections; Musk wants ~mid-20s voting influence, framing it as protection against being "ousted" after building the robot fleet.
Key numbers and quarter mechanics- Record deliveries and deployments (no unit numbers given on the call itself; regional splits: China/APAC +33%, NA +28%, EMEA +25% QoQ).
- Auto revenue +29% QoQ, described as in line with delivery growth — implying flat-ish ASPs vs. Q2's mix benefit.
- Auto gross margin ex-credits: 15.4%, up from 15.0% — first disclosure of this metric in two calls; attributed to material cost and fixed-cost absorption on higher volumes.
- Regulatory credits declined sequentially, though new contracts signed.
- Tariff impact: >$400M in Q3, split roughly evenly between auto and energy — up from $300M in Q2. Energy bears higher tariff exposure as % of COGS (cells sourced from China); Megafactory Shanghai ramp is the mitigation for non-US demand.
- Free cash flow: ~$4B, a record (vs. $146M in Q2); cash and investments >$41B.
- CapEx: ~$9B for 2025 (consistent with revised guide); 2026 guided to "increase substantially" for AI/Optimus and growth investments — a directional reversal.
- Bitcoin: +$80M mark-to-market gain in Q3 vs. +$284M in Q2; other income also hit by FX.
- OpEx up sequentially: restructuring charges tied partly to convergence of AI chip design efforts, legal expenses, shareholder-meeting costs in SG&A, and rising AI-related equity comp — guided to keep increasing.
- Robotaxi costs sit in Services & Other — disclosed as small; no standalone P&L.
- Paid FSD take rate disclosed: ~12% of the current fleet — "still small," first explicit fleet-level penetration number.
Product and launch scorecard- Robotaxi: genuinely scaled, KPIs still partial. >250K driverless-seat Austin miles, >1M Bay Area miles, three Austin expansions, two operating markets. Committed: no safety drivers in large parts of Austin by year-end; 8–10 metros (NV, FL, AZ named) by year-end. Not disclosed: fleet size, rides, intervention rates, pricing, revenue, utilization.
- FSD V14: broadly available in the US via opt-in advanced software. 14.1 shipped; 14.2 targeted to fix comfort ("safe but jerky"); reasoning (parking-spot logic, drop-off-and-park) in 14.3/14.4 or "by end of this year for sure"; order-of-magnitude parameter increase planned beyond 14.1. 6B cumulative supervised FSD miles. The Q2 admission that the production build lagged the robotaxi build by months was effectively resolved: Ashok said the two are now "mostly the same."
- HW3: partial concession. "V14 Lite" for HW3 planned ~Q2 2026 — first concrete deliverable for HW3 owners — while full retrofit/replacement remains deferred behind solving autonomy. Taneja acknowledged these customers but gave no cost/scope.
- Cybercab: first production date — Q2 2026 start, called "the single biggest expansion in production." No steering wheel/pedals reiterated.
- Optimus: slipped within the quarter. Q2 said V3 prototypes in ~3 months and production early 2026; Q3 says production-intent prototype shown Q1 (Feb–Mar) and production start "towards the end of next year" on a million-unit line, with rolling design changes even after production start. Musk conceded the hand/forearm "more difficult than the entire rest of the robot" and no supply chain exists. Long-range: Optimus 4 at 10M units, Optimus 5 at 50–100M.
- Semi: on schedule. Building complete, equipment installing, validation fleet on road; larger build late 2025, first line builds early 2026, ramp ~Q2 2026, real volume H2 2026.
- Energy: Megablock ships next year from Houston; Megapack 4 (integrated substation, ~35kV output) is the engineering priority; Buffalo residential solar panel ships Q1; solar demand surging on policy changes into 2026.
- AI5: dual-sourced TSMC/Samsung (both US fabs); half-reticle design via deletions (legacy GPU, ISP); claimed 2–3x best performance/watt and ~10x performance/dollar vs. alternatives; excess supply earmarked for Tesla data centers.
Sell-side read-through- Only four analysts got through (Rosner, Levy, Walt/LightShed, Langan); Jonas lost to audio issues — again. Two retail future-product questions were skipped outright, cutting forum content.
- Rosner (Wolfe) asked the sharpest question — how the production-expansion declaration maps to the ~3M capacity, timeline, and whether it means prioritizing volume over margin. Got the ~24-month/3M annualized aspiration and Musk's "demand will be pretty nutty" denial of margin sacrifice.
- Levy (Barclays) asked about TAM boundaries and Tesla/xAI complementarity. Musk: Grok is 10–20x larger than Tesla's models, targets AGI against Gemini/ChatGPT; Tesla = real-world AI; Grok supplies voice for Optimus/car. More structured than prior quarters but still no financials.
- Walt (LightShed) got the most operationally useful answers: the ~3-month safety-occupant playbook per new market and confirmation that consumer FSD and robotaxi builds are now converged.
- Langan (Oppenheimer) pressed the Optimus timeline vs. dexterity/supply-chain readiness — and extracted the significant concession that hardware design won't be frozen even through production start, plus the million-unit line "towards the end of next year."
- Notable absences/evasions: no analyst asked about Q4 demand after the credit expiry (the biggest open question given Q3's pull-forward); no follow-up on the "rough quarters" guidance; robotaxi fleet size/rides/revenue again undisclosed and again unchallenged; no auto margin bridge beyond ex-credits; the $16.5B Samsung deal question was answered with a chip-design monologue rather than deal terms or timeline/yield evidence.
Management credibility- Delivered: robotaxi scaling metrics improved ~35x on Austin miles (7K → >250K) in one quarter with continued zero notable incidents; the "very strong" energy performance materialized as record deployments/gross profit/margins; ~$9B CapEx guide held; FCF rebounded to a record ~$4B from $146M; record deliveries confirmed the pre-credit pull-forward Taneja predicted.
- Positive new disclosures: auto margin ex-credits (15.4%), FSD paid penetration (~12%), robotaxi miles by market, Cybercab production date, Semi ramp schedule, HW3 V14 Lite date.
- Slipped within one quarter: Optimus production moved from "early 2026" (Q2) to "towards the end of next year" (Q3); V3 reveal moved from ~3 months (i.e., roughly now) to "Q1, probably February or March." The Q2 commitment to port the robotaxi build back to production cars "in the next several weeks" was quietly resolved as convergence — without acknowledging the prior lag.
- Escalating rhetoric with hedged numbers: "100% confident" on unsupervised FSD; AI5 "40x better… but just to say safely, 10x"; Optimus as "infinite money glitch," 50–100M units for V5. The pattern of maximalist claims alongside softer footnoted numbers continues.
- Governance answers under pressure were candid but escalatory: Musk named a specific desired stake (mid-20s%) and attacked proxy advisors; Taneja's closing pitch blurred the call's purpose into a proxy campaign.
- Q4 silence is itself notable: after pre-announcing "a few rough quarters" last call, management offered no Q4 demand or margin framing despite record pull-forward deliveries — the most consequential omission of the call.
What changed versus the prior quarter- The "rough quarters" warning went unmentioned despite Q4 being the first quarter after credit expiry; Q3's record results confirmed the pull-forward mechanism but left the hangover unquantified.
- Robotaxi went from pilot-scale to early-scale: >250K Austin driverless miles and a two-market operation, with dated commitments (safety-driver removal, 8–10 metros) replacing vague expansion talk. Still no fleet/rides/revenue.
- Posture flipped from caution to expansion: Musk now claims "clarity" on unsupervised FSD and commits to maximizing production (~3M annualized aspiration within ~24 months); Cybercab anchored to Q2 2026.
- CapEx trajectory reversed again: 2025 held at ~$9B, but 2026 guided "substantially" higher — after the >$10B → >$9B whipsaw of the prior two quarters.
- Tariff cost grew ($300M → >$400M) as the latency effect Taneja flagged played out; energy's China-cell exposure now explicitly acknowledged with Shanghai as mitigation.
- Optimus slipped and was re-scoped again: production start pushed within-90-days from early 2026 to late 2026; hand dexterity elevated to the central engineering risk; design explicitly unfrozen through production start.
- FSD narrative converged: production and robotaxi builds now "mostly the same"; V14 shipped broadly; reasoning features dated to year-end; HW3 got its first concrete deliverable (V14 Lite, ~Q2 2026) instead of pure deferral.
- FSD economics got a baseline metric: ~12% paid penetration of the fleet — small, and now measurable quarter to quarter.
- Governance intensified: from Q2's "major concern" about his 13% stake to a specific mid-20s% target, a named vote date, and open hostility toward ISS/Glass Lewis.
Bull case- Record quarter on almost every axis — deliveries, revenue, energy gross profit/margins, ~$4B FCF, >$41B cash — achieved while absorbing >$400M of tariffs and declining credit revenue; margin ex-credits improved to 15.4%.
- Robotaxi evidence is finally compounding: 35x mile growth in one quarter, incident-free on disclosure, a repeatable market-entry playbook (~3 months supervised), and dated commitments (Austin unsupervised, 8–10 metros by year-end) that are testable within 90 days.
- FSD flywheel has measurable levers: 6B supervised miles, only ~12% paid penetration (large headroom), V14 broadly available, reasoning features dated, and HW3 owners given a path rather than abandonment.
- Energy is a second engine with structural tailwinds: record deployments/margins, AI-datacenter demand, Megablock shipping next year, Megapack 4 removing the substation bottleneck, Shanghai factory neutralizing tariffs for non-US demand, solar demand surge into 2026.
- Capacity expansion is now an explicit, funded plan — ~3M annualized aspiration plus Cybercab from Q2 2026 — backed by record cash generation rather than balance-sheet strain.
- AI5 dual-sourcing and oversupply strategy de-risks compute supply and creates an option on a distributed inference/data-center asset.
Bear case- Q4 is a policy-created air pocket that management declined to discuss. Q3's record was manufactured by credit-expiry pull-forward (+25–33% regional sequential jumps); the prior quarter's own "rough quarters" guidance was neither reiterated nor withdrawn — the single biggest gap in the call.
- Robotaxi remains financially invisible: no fleet size, rides, revenue, or unit economics; costs buried in Services & Other. Commitments (8–10 metros, driverless Austin) remain unproven until year-end.
- Optimus timelines slipped again within 90 days, the design is explicitly unfrozen, the hand is admitted to be harder than the rest of the robot, and there is no supply chain — yet a million-unit line is being built. Execution risk is extreme relative to rhetoric ("infinite money glitch").
- Auto margin ex-credits of 15.4% leaves thin cushion if Q4 requires incentive spending to fill the post-credit demand hole; Musk's "demand will be pretty nutty" is an assertion, not evidence.
- Tariff costs are still rising ($300M → >$400M) with energy structurally exposed until localization matures.
- Governance overhang escalated: the comp vote is now framed by Musk himself as a condition of his continued direction, with a specific mid-20s% control ask and public attacks on proxy advisors — a binary Nov 6 event with tail risk either way.
- Claim inflation continues: "100% confident," "40x better (safely 10x)," 50–100M Optimus units — the gap between stated certainty and disclosed operating evidence remains the core credibility tension.
Next-quarter watchlist- Q4 deliveries and demand shape: size of the post-credit air pocket; incentive levels; whether management re-acknowledges "rough quarters"; any order-rate disclosure now that pull-forward is spent.
- Robotaxi year-end commitments: safety-driver removal in Austin (did it happen, and where?); actual metro count vs. the 8–10 claim (NV/FL/AZ approvals); fleet size, rides, and any incident disclosure; first revenue/pricing data.
- FSD V14.x execution: 14.2 comfort fixes; reasoning in 14.3/14.4 "by end of this year for sure"; the order-of-magnitude parameter increase; paid penetration vs. the 12% baseline; China/EMEA approval progress.
- Nov 6 shareholder vote outcome and Musk's response if the comp proposals fail; any board/governance fallout.
- Optimus V3 reveal (promised Q1, Feb–Mar): production-intent prototype evidence, hand functionality, and whether "end of next year" production start holds or slips again.
- Cybercab: confirmation of Q2 2026 production start; any cost-per-mile update vs. the $0.25–0.30 target.
- Semi: larger build by year-end, first line builds early 2026, Q2 ramp on schedule.
- Auto margin ex-credits and tariff trajectory: does >$400M stabilize; does margin hold without credit-expiry volume; credit-revenue decline rate.
- 2026 CapEx specificity: "substantially higher" needs a number; FCF path against AI/Optimus/robotaxi spend.
- AI5: tape-out/production timing at TSMC Arizona and Samsung Texas; evidence for the performance claims; data-center deployment of surplus chips.
- Energy: Megablock shipments from Houston; Megapack 4 unveiling; Buffalo solar panel shipments; whether record deployments repeat without policy pull-forward.
|
| Jul 23, 2025 | -8.20% | Q2 FY2025 | Read transcript briefingQuarter in one view- Robotaxi launched — the Q1 promise was kept, at pilot scale. Paid, driverless rides in Austin began in June; service area already expanded once, with a >10x expansion of the operating region planned and Bay Area, Nevada, Arizona, Florida in regulatory process. But the disclosed KPIs are tiny: "a handful of vehicles" and ~7,000 cumulative autonomous miles. Musk's new framing: autonomous ride-hailing addressing "half the population of the U.S. by end of year," subject to approvals.
- The policy environment turned sharply negative. The "One Big Bill" repeals the $7,500 IRA EV credit at end of Q3 (pull-forward demand now, air pocket after) and zeroes out emissions penalties, structurally shrinking regulatory credit revenue. Tariffs cost ~$300M in Q2 (~2/3 auto, 1/3 energy) with the full impact still to come.
- Sequentially better auto mechanics, weak cash generation. Auto revenue +19% QoQ on +14% deliveries (new Model Y ASPs), margins improved sequentially despite tariffs; but free cash flow was only $146M as CapEx rose. 2025 CapEx guide trimmed to "in excess of $9B" from Q1's ">$10B."
- Musk pre-announced "a few rough quarters" (Q4, Q1, maybe Q2) — unusually direct negative guidance — while reiterating autonomy materially impacts financials "around the end of next year."
- Affordable model slipped again: production started June as planned, but the ramp was deferred to Q4 to maximize pre-credit-expiry output of current models.
What management is focused on- Autonomy as the entire equity story. Musk: "autonomy is the story… amplifies the value to stratospheric levels." Nearly all opening remarks covered robotaxi expansion, FSD parameter count (~10x increase claimed possible on existing hardware), and regulatory approvals (Netherlands/EU, China).
- Managing the IRA credit expiry as a demand event. Taneja urged US buyers to order now; deliveries can't be guaranteed for orders placed late August onward; incentives already fully rolled out and being pared back. This is an explicit pull-forward that sets up the Q4 air pocket Musk acknowledged.
- Regulatory credit decline reframed as planned. Taneja: "we now plan our business around such sales" — but conceded lower revenue going forward from the penalty repeal.
- FSD adoption as a near-term lever: 25% increase in penetration since V12/V13 in North America; ~half of eligible Tesla owners have never tried FSD; new in-car prompts and service-visit education planned. Safety report cited: FSD "10x safer."
- Optimus redesign reset: Optimus 3 is now "the right design" with "no significant flaws"; prototypes in ~3 months, production start early next year, 100K/month by month 60. The Q1 "thousands of units by end of 2025" target was not restated.
- Musk's control stake as a live governance issue: he told Jonas his 13% stake is "a major concern" he wants addressed at the shareholder meeting — he wants enough control to steer "an army of humanoid robots" but not so much he can't be removed.
Key numbers and quarter mechanics- Auto revenue +19% QoQ vs. deliveries +14% — gap driven by new Model Y ASPs; margins also helped by mix and fixed-cost absorption, partially offset by tariffs.
- Tariff cost: ~$300M in Q2 (~$200M auto, ~$100M energy); full impact lands in coming quarters due to manufacturing/sales latency.
- Free cash flow: $146M — operating cash flow up sequentially but CapEx up more.
- CapEx guide: "in excess of $9B" for 2025 — down from Q1's ">$10B," reversing the tariff-driven increase of one quarter ago.
- Bitcoin: +$284M mark-to-market gain in Q2 vs. -$125M loss in Q1 — the other-income volatility continues in both directions.
- Energy: highest gross profit ever, again, on lower deployments (mix shift to higher-margin Megapack); record Powerwall deployments; margins improved sequentially.
- Services margin improved on Supercharging profits and insurance/service-center profitability.
- OpEx up sequentially on AI spend, stock comp, and AI compute depreciation; R&D guided to keep growing.
- Regulatory credits: revenue already down QoQ; structural decline ahead from penalty repeal.
- No delivery volume guide given or requested — third consecutive call.
Product and launch scorecard- Robotaxi (Austin): launched on time, metrics minimal. Paid driverless rides live; one geofence expansion done; >10x area expansion planned; Bay Area launch planned *with* a safety driver initially to expedite while awaiting approval. Disclosed: ~7,000 miles, "handful of vehicles," zero notable safety-critical incidents, self-imposed 40 mph cap. No fleet size, ride count, or revenue disclosed. Financial materiality: "around the end of next year" — consistent with Q1's deferral.
- Autonomous factory-to-customer delivery: one car delivered autonomously in Q2; Musk said default self-delivery in Austin/Bay Area by end of 2025.
- Unsupervised FSD for personal use: "by end of this year in certain geographies" — reaffirmed, still hedged ("being very careful"). Notably, the production FSD build is "several months behind" the Austin robotaxi build; elements being ported back "in the next several weeks."
- Customer cars joining the robotaxi network: "confidently next year," with an Uber/Airbnb-style vehicle inspection checklist. First concrete timeline for this.
- Affordable model: June production start confirmed; ramp pushed to Q4 (credit-expiry prioritization + ramp complexity); design details refused ("what happens in the studio stays in the studio"); goal stated as not diluting revenue/gross margin.
- Cybercab: cost target refined — sub-$0.30/mile potential, maybe $0.25, vs. "over $0.50" for the existing fleet; design trade-offs explained (gentle ride, efficient tires, lower top speed). No production timeline update this call.
- Optimus: effectively re-baselined. V3 design locked-ish; prototypes ~3 months; production early 2026; 100K/month by mid-2030. The Q1 "thousands by end of 2025" and the rare-earth magnet issue went unmentioned.
- Dojo/AI5: Dojo 2 at scale (~100K H100-equivalents) next year; AI5 volume production around end of 2026; AI5 so powerful it must be "nerfed" for export markets. Dojo 3/AI6 convergence hinted.
- Energy: first US LFP cell facility online by end of 2025; third Megafactory near Houston in 2026; "very strong second half" deployment forecast.
- HW3 retrofit: deferred again — unsupervised on HW4 first, "then we will go back and look at" HW3. No scope, cost, or timing.
Sell-side read-through- Five analysts through (Rosner, Jonas, Levy, Delaney, Stein); Edison Yu lost to audio issues. Again one question plus follow-up each after a long retail Q&A; limited sustained pressure.
- Rosner (Wolfe) got the only hard robotaxi KPIs: ~7,000 miles, handful of vehicles, no safety-critical incidents, 40 mph cap — and the Cybercab vs. fleet cost-per-mile split ($0.25–0.30 vs. >$0.50). The numbers confirm a very small pilot.
- Jonas (Morgan Stanley) spent both questions on governance and AI Day — extracting Musk's on-record concern about his 13% stake and control. No financials.
- Levy (Barclays) got the customer-car network timeline ("confidently next year") and the funding answer: balance sheet now, debt financing once recurring cash flow exists.
- Delaney (Goldman) got the FSD adoption datapoint (25% penetration increase since V12/V13) and, on pricing, Musk's "few rough quarters" admission — the most consequential guidance of the call.
- Stein (Truist) tried twice on affordable-model specifics and xAI/Tesla boundaries; got a refusal on design details and a talent-segmentation answer (xAI = ASI/terabyte models, Tesla = real-world AI, 100x smaller models).
- Notable evasions/absences: the xAI investment question was flatly refused ("not the forum"); no robotaxi fleet size, ride counts, pricing, or revenue; no auto gross margin ex-credits; no Q2 delivery miss discussion; no Optimus magnet update; no HW3 cost disclosure; no one pressed the "half the US population by year-end" claim against the 7,000-mile reality.
Management credibility- Delivered: the June Austin paid driverless launch — the central Q1 commitment — happened on schedule, with geofence expansion already executed. Affordable model production did start in June as promised. Energy gross profit record repeated.
- New candor worth crediting: Musk's "we probably could have a few rough quarters" is rare explicit negative guidance; the 7,000-miles/handful-of-vehicles disclosure tempers the launch hype; Taneja quantified the tariff hit ($300M) and the credit-revenue headwind plainly.
- Slipped/quietly dropped: the affordable-model ramp moved from Q3 to Q4 within one quarter; the Q1 "thousands of Optimus units by end of 2025" target was not restated and the program was re-baselined to Optimus 3 prototypes in ~3 months and production in early 2026 — a de facto slip; the Q1 magnet export-license problem vanished from discussion; HW3 retrofit deferred behind HW4 unsupervised with no timeline.
- CapEx guide whipsawed again: flat (Q4) → >$10B (Q1) → >$9B (Q2) in three quarters.
- Under pressure: the xAI investment question was refused outright; the affordable-model design question was deflected with jokes; Musk's "half the US population by end of year" robotaxi claim sits against a handful of vehicles and 7,000 miles, and no analyst challenged it.
- Grandiosity maintained: "most valuable company in the world by far," "biggest product ever" (Optimus), "hyper-exponential" scaling, "100,000 Optimus a month in 60 months — I'd be shocked" otherwise. The gap between rhetoric and disclosed operating metrics remains the central credibility tension.
What changed versus the prior quarter- Robotaxi went from promise to operating service — but with disclosed scale (handful of vehicles, 7,000 miles) at the low end of even Q1's tempered 10–20 car framing.
- Autonomy materiality timeline held ("end of next year") but the geographic ambition escalated sharply: from "many other cities by end of 2025" to "half the US population."
- Policy became the dominant near-term headwind: Q1 was tariffs; Q2 adds IRA credit repeal (end of Q3) and emissions-penalty repeal (structural credit-revenue decline). Tariffs moved from modeled (~$2K/unit) to actual ($300M in the P&L).
- CapEx guide cut back to >$9B one quarter after being raised to >$10B on tariffs.
- Affordable model ramp deferred to Q4 — the second consecutive "slower than hoped" signal, now with an explicit cause (credit-expiry pull-forward).
- Optimus re-baselined around V3 with production in early 2026; the 2025 unit target and magnet constraint disappeared.
- Customer-car network participation got a date: "confidently next year," with an inspection regime — new specificity.
- FSD adoption quantified for the first time: 25% penetration increase since V12/V13; ~50% of eligible owners never tried FSD.
- Musk's control/ownership became an explicit call topic — new governance overhang tied to the shareholder meeting.
- Bitcoin swung positive again (+$284M vs. -$125M), confirming the recurring other-income volatility flagged in Q1.
Bull case- The falsifiable June commitment was met: paid driverless robotaxi is live, expanding, incident-free on disclosed metrics, with regulatory processes underway in four-plus states and a >10x Austin area expansion weeks away.
- Auto mechanics improved sequentially (revenue +19% on +14% deliveries, ASP/mix/margin gains) even while absorbing $300M of tariffs — the Model Y changeover headwind is behind them.
- Energy is a durable profit engine: record gross profit in consecutive quarters on lower deployments, record Powerwall, customers accepting tariff costs, US LFP online by year-end, third Megafactory in 2026, strong H2 deployment forecast.
- FSD monetization staircase is forming: 25% adoption increase, huge untapped owner base, attention-monitoring relaxation coming, unsupervised personal use by year-end, customer-car fleet participation next year.
- The Q4 air pocket is pre-announced and policy-driven, not demand-driven; management is pulling demand forward deliberately and guiding through it.
- Cybercab economics quantified ($0.25–0.30/mile potential) with a coherent design rationale.
Bear case- Management itself guided to "a few rough quarters" — credit expiry, incentive loss, tariff latency, and credit-revenue decline all hit simultaneously starting Q4.
- Robotaxi is operationally tiny: a handful of vehicles and 7,000 miles after a month, a 40 mph cap, safety drivers planned for the Bay Area — versus "half the US population by year-end." The gap between claim and disclosed reality is the widest yet.
- Free cash flow of $146M against rising AI CapEx and a balance-sheet-funded robotaxi buildout; earnings quality still noisy (Bitcoin +$284M this quarter).
- Guidance instability continues: CapEx whipsawed three straight quarters; Optimus re-baselined; affordable ramp slipped; HW3 retrofit indefinitely deferred; no volume guide for a third call.
- Regulatory credit revenue faces structural decline from the penalty repeal — a high-margin revenue stream with no replacement disclosed.
- Governance risk elevated: Musk publicly tying his continued direction of "an army of humanoid robots" to increased control, plus a refused question on Tesla investing in xAI, adds an unresolved overhang.
- The production FSD build lags the robotaxi build by months — customer cars are not the robotaxi, and the year-end unsupervised promise depends on porting that gap.
Next-quarter watchlist- Robotaxi scaling evidence: fleet size, miles, ride counts vs. the 7,000-mile baseline; the promised >10x Austin expansion; Bay Area launch (with or without safety driver); any incident disclosure; progress toward the "half the US population" claim.
- Q3 deliveries and the pull-forward: magnitude of pre-credit-expiry demand; incentive levels; order-inflow commentary for Q4; whether "rough quarters" guidance hardens.
- Auto margin ex-credits and tariff trajectory: does the $300M quarterly tariff cost grow as latency washes through; credit-revenue decline rate.
- Affordable model: Q4 ramp evidence, pricing/monthly payment, form factor, and whether it dilutes margin despite the stated goal.
- Unsupervised FSD for personal use: dated commitment by year-end; whether the production build actually catches up to the Austin build "in the next several weeks."
- Optimus: V3 prototypes in ~3 months (i.e., by the Q3 call); any restatement of 2025 unit targets; magnet supply status.
- Energy: H2 deployment ramp vs. the "very strong second half" forecast; US LFP facility online by year-end; residential-storage demand shift from consumer-credit expiry.
- CapEx and FCF: does >$9B hold; does FCF recover from $146M as robotaxi spending scales.
- Governance: shareholder-meeting outcome on Musk's control stake; any Tesla–xAI investment proposal; AI Day/Optimus disclosure plans.
- HW3: any retrofit scope/cost commitment once HW4 unsupervised ships — or continued deferral.
|
| Apr 22, 2025 | +5.37% | Q1 FY2025 | Read transcript briefingQuarter in one view- A weak auto quarter with a reaffirmed autonomy timeline. Q1 deliveries declined (Model Y changeover plus brand damage), auto margins fell sequentially on lower volumes, poor fixed-cost absorption, and lower regulatory credit revenue — partially offset by slightly higher pricing on the new Model Y. Management's answer to everything remains the June Austin paid robotaxi launch, reaffirmed again.
- The June launch is now scoped very small: Musk said "maybe 10 or 20 vehicles on day one," Model Ys (not Cybercab), Tesla-operated, scaling "rapidly after that." Financial materiality from autonomy is now explicitly pushed to "around the middle of next year, second half of next year" — a new, important admission that 2025 autonomy revenue is immaterial.
- Tariffs are now a quantified, multi-front problem: Section 232 auto tariffs effective May will hit profitability (Taneja cited external modeling of "a couple of thousand" dollars per unit, in line with Tesla's own forecast); the energy business faces an "outsized" hit from China-sourced LFP cells; and tariffs on imported factory equipment pushed the 2025 CapEx guide back above $10B — reversing Q4's "flat YoY" guide.
- Brand damage acknowledged for the first time as a demand factor. Taneja cited "vandalism and unwanted hostility towards our brand" as impacting certain markets; Musk opened the call with an extended defense of his DOGE role and said his government time drops "significantly" starting in May, to a day or two per week.
- Energy was the bright spot: record gross profit for energy storage despite a sequential deployment decline; Powerwall 3 supply-constrained; Shanghai Megafactory started operations in Q1 (on the Q4-promised timeline).
- Bitcoin mark-to-market swung the other way: a Q1 loss vs. Q4's gain drove a $472M sequential drop in other income — the earnings-quality wrinkle flagged last quarter is now live in both directions.
What management is focused on- Musk's time allocation as a governance issue. He led the call with DOGE, framed protests as "paid for" by recipients of waste/fraud, and committed to materially more Tesla time from May. This is management choosing to address the brand/attention overhang head-on rather than being asked.
- June Austin robotaxi as the falsifiable anchor — with heavy expectation management. "Laser focused" on June, but: 10–20 cars at launch, an S-curve ramp that is "very difficult to predict," and financial materiality only in mid/late 2026. The acid test Musk offered: "can you go to sleep in your car and wake at your destination" in many US cities by end of 2025.
- Tariff triage across three fronts: vehicle content (85% USMCA-compliant on a weighted-average basis; >95% local content in Shanghai), energy LFP cells (commissioning US LFP equipment that covers only "a fraction" of installed capacity; seeking non-China suppliers), and CapEx equipment imports (China has the most capacity; costs rising).
- Vertical integration as the tariff defense. Extended discussion of the South Texas lithium refinery ("biggest outside of China"), Austin cathode refinery, in-house 4680 cells claimed as lowest cost/kWh, multi-sourced cell components from at least two countries of origin.
- Affordable models and factory utilization. Production start still planned for June; Lars Moravy admitted the ramp "might be a little slower than we had hoped" and confirmed the new models are constrained to existing lines and "will resemble, in form and shape, the cars we currently make" — effectively confirming the "cheaper Model Y variant" reporting management declined to confirm directly.
- Optimus as a development program, with a new supply-chain problem. Thousands of units by end of 2025, back-weighted to year-end; the ramp is gated by ~10,000 unique parts and, newly disclosed, China's export-license requirement on rare-earth magnets used in Optimus arm actuators — Tesla is seeking a license with assurances the magnets aren't military.
Key numbers and quarter mechanics- Auto margin declined sequentially — drivers: lower total deliveries, lower fixed-cost absorption from the changeover, lower regulatory credit revenue; offset partially by slightly higher pricing from new Model Y despite legacy-model incentives.
- Other income down $472M sequentially, primarily Bitcoin mark-to-market (Q1 loss vs. Q4 gain), remainder FX. Management pre-announced continued volatility in other income from the new standard.
- CapEx guide: "in excess of $10 billion" for 2025, inclusive of tariff impacts — a reversal of Q4's flat-YoY guide (2024 CapEx was >$11B, so this is technically a decline, but the direction of guidance language flipped from "flat" back to ">$10B" with tariffs cited as the driver and "still evaluating what more to do").
- OpEx up sequentially on AI initiatives including Optimus and development of Cybercab, Semi, and cheaper models (R&D), partially offset by SG&A declines.
- Energy: record gross profit in Q1 despite sequential deployment decline; Powerwall 3 supply-constrained; Shanghai Megafactory operational in Q1, serving non-US markets (which also mitigates the US tariff hit on energy).
- Services/other margin slightly down sequentially on used-car and insurance pressure; labor productivity improvements in service/collision continuing.
- Model Y changeover executed as pre-announced: legacy Model Y produced until mid-to-late February, sold out in US/China and other markets; new Model Y reached prior production rates in under eight weeks across all four factories. New Model Y availability in most markets only in the last few weeks of the quarter — management's primary explanation for the delivery miss, alongside brand hostility.
- Battery constraint lifted for vehicles: Karn Budhiraj stated Tesla is "not constrained on battery cell supply for vehicles" — a direct change from Q4's "battery packs are the 2025 limiter." Energy remains tariff-exposed on LFP.
- No volume growth guide was given or asked about. No 2025 delivery target was restated.
Product and launch scorecard- Robotaxi (Austin, June): reaffirmed, narrowed, and de-risked rhetorically. Model Ys, current production hardware, Tesla-operated, ~10–20 vehicles day one, paid rides, public access "end of June or July." Ashok disclosed the validation problem plainly: critical interventions are so rare in the Austin QA fleet that "you can go many days without getting any single intervention," forcing reliance on simulation and neural-network video generation to measure safety. The Austin build includes audio input for emergency vehicles — a feature the customer fleet lacks. Remote operators exist for stuck vehicles ("availability," not safety, per management). "Many other cities in the US by end of this year" reaffirmed.
- Unsupervised FSD for personal cars: "before the end of this year" in several/many US cities — pulled forward from Q4's "probably next year" for customer-car network participation, though the framing here is personal use, not fleet monetization.
- Affordable models: June production start reaffirmed; ramp "a little slower than hoped"; built on existing lines, resembling current Model 3/Y form factors; focus is monthly payment/lowest price. The "new model" is now clearly a derivative, not a new platform.
- Cybercab: B-sample validation now; first big builds end of Q2; equipment installation at Giga Texas in coming months; production "next year" (2026) — consistent with prior timelines. Built in the existing Austin factory, not a new building.
- Unboxed manufacturing: de-risking claims (subassembly marrying, aluminum corrosion, seam sealing, early crash testing); target cycle time of one unit per 5 seconds or less eventually, vs. 33 seconds at Shanghai currently. First real full-scale test is the Q2 Cybercab builds; test vehicles on roads "in a couple of months."
- Optimus: thousands of units by end of 2025, back-half weighted; still a development program; China rare-earth magnet export license is a live constraint on arm actuators; 1M units/yr by 2029–2030 claimed. No design-lock or external-sales update this quarter.
- FSD pricing: multi-tier pricing under consideration (supervised vs. unsupervised); management hinted current $99/month is "too cheap" and attention-monitoring will be relaxed "every few weeks or every month" — the eyes-off progression continues without a hard date this quarter.
- India: actively being worked, but 70% tariff plus 30% luxury tax (~100% price inflation) is the blocker; no launch commitment.
Sell-side read-through- Five analysts got through (Ferragu, Rosner, Yu, Gianarikas, Langan, plus Jonas after audio issues) — one question plus one follow-up each, after a long retail Q&A. The format again limited sustained pressure; no one asked about the missing volume guide, ex-credit margins, or the magnitude of the delivery decline.
- Rosner (Wolfe) got the most substantive autonomy disclosure: the public FSD build still has interventions; the Austin solution involves burning down a city-specific issue list, system-level redundancy (no "red hands" fallback), and audio input that customer cars lack. The admission that the launch build differs from the customer fleet build is new and important — it undercuts the "your car today is the robotaxi" claim in the near term.
- Yu (Deutsche Bank) extracted the launch scale: 10–20 vehicles day one — a materially smaller start than the "millions next year" framing implies, and the first concrete fleet number given.
- Langan (Wells Fargo) challenged vision-only on glare/fog; Musk answered with "direct photon counting" bypassing the image signal processor — a specific technical rebuttal rather than the usual philosophy. Langan also surfaced the affordable-model delay reports; management's answer (existing lines, resembles current cars) effectively confirmed the substance while denying a delay.
- Gianarikas (Canaccord) got the multi-tier FSD pricing signal and the India tariff detail.
- Jonas (Morgan Stanley) spent his questions on tariff policy advocacy and US-vs-China physical AI — no financials. Musk's answers: he's "one of many advisors," advocates lower/predictable tariffs, and conceded China dominates drones (~70% of production, ~100% supply-chain dependency) while claiming Tesla is rank one in humanoids with ranks 2–10 likely Chinese.
- Ferragu (New Street) asked why Model 3/Y hold only ~15% of their addressable market; Musk's answer — "in the future, most people are not going to buy cars" — declined to engage with the near-term demand/share question at all.
- Notable absences: no questions on Q1 delivery shortfall magnitude, auto gross margin ex-credits, the CapEx guide reversal, energy deployment decline, HW3 retrofit status (dropped entirely from the agenda), or the Bitcoin loss's earnings-quality impact.
Management credibility- Delivered/consistent: Model Y changeover executed on the pre-announced timeline (legacy sold out, new model at rate in <8 weeks); Shanghai Megafactory operational in Q1 as promised; June Austin reaffirmed with concrete detail (10–20 cars, Model Y, paid, public access by end of June/July); Cybercab B-sample and Q2 builds on schedule; affordable model June start reaffirmed.
- New candor worth crediting: Ashok's description of the validation measurement problem (interventions too rare to detect regression) is unusually honest for a pre-launch autonomy program; the 10–20 car launch scale tempers the hype; the magnet export-license problem was disclosed unprompted.
- Slipped/reversed: CapEx guide flipped from "flat YoY" (Q4) to ">$10B" in one quarter, with tariffs as the stated reason — the Q4 "CapEx efficiency" framing lasted one quarter. Autonomy financial materiality is now explicitly mid/late 2026, which quietly re-frames the June launch as commercially immaterial for ~12+ months. The Q4 battery constraint was declared resolved for vehicles without quantification.
- Demand messaging is inconsistent: Taneja cited brand hostility as a delivery headwind; minutes later an executive said "we don't see any reduction in demand" absent macro, citing record Q1 test drives and California best-seller status. Both cannot fully explain the quarter; the order-inflow question was answered with anecdotes, not data.
- Under pressure: Musk's answer to the market-share question was a horizon deflection; the tariff-policy answer to Jonas was carefully lawyered ("I am not the President"); the affordable-model answer confirmed the derivative nature only after a media report forced it.
- Grandiosity maintained: "most valuable company in the world by far… as valuable as the next five companies combined," "99% market share or something ridiculous" in autonomy, "10 million autonomous cars in a few years," terawatts of storage. The gap between these claims and a 10-car June launch is the central credibility tension.
What changed versus the prior quarter- Autonomy materiality was explicitly deferred: Q4 framed 2025 as the year autonomy arrives; Q1 says it "moves the financial needle" only around mid/late 2026. June is now a 10–20 car pilot, not a business.
- Customer-car autonomy pulled forward in language: personal-use unsupervised FSD "before end of this year" in several cities vs. Q4's "probably next year" for customer fleet participation — though the Austin launch build having features (audio input) the customer fleet lacks complicates this.
- CapEx guide reversed: flat YoY (Q4) → "in excess of $10B" (Q1), tariffs on imported equipment cited.
- Battery constraint lifted for vehicles — Q4's stated 2025 production limiter is gone; the constraint migrated to energy (China LFP cells) and to Optimus (rare-earth magnets).
- Tariffs went from "very likely" risk to quantified P&L impact: ~$2K/unit external modeling endorsed, May Section 232 effective date, energy "outsized" impact, CapEx impact — plus a new China export-license problem on magnets.
- Brand damage formally acknowledged as a demand factor for the first time, alongside Musk's commitment to reduce DOGE time starting May.
- Bitcoin mark-to-market went from a $600M Q4 gain to a Q1 loss, a $472M sequential other-income swing — the recurring volatility is now demonstrated, not theoretical.
- HW3 retrofit disappeared from the narrative — not mentioned once, despite being a confirmed program last quarter.
- Energy narrative shifted from deployments to profitability: record gross profit despite sequential deployment decline; no restatement of the ≥50% 2025 deployment growth guide.
- Optimus gained a specific near-term blocker (magnets) and kept the "thousands by year-end" target, back-weighted.
Bull case- June is weeks away and management added specificity, not hedges: 10–20 Model Ys, paid, driverless, public access by end of June/July, current production hardware, QA fleet already running in Austin, factory autonomy routine and publicly visible at Fremont.
- The demand explanation is checkable and partly exculpatory: the changeover genuinely removed the world's best-selling car from most markets for much of the quarter; legacy inventory sold out; new Model Y ramped to rate in under eight weeks; record test drives suggest interest survived the brand damage.
- Energy is becoming a profit engine: record gross profit on lower deployments implies strong unit economics; Powerwall 3 supply-constrained; Shanghai online for non-US markets; gigawatt-class orders described as common.
- Tariff positioning is relatively advantaged: 85% USMCA content, >95% Shanghai localization, in-house cells at claimed lowest cost/kWh, lithium and cathode refining in-house — Tesla framed itself as the least-exposed OEM, and the ~$2K/unit modeled impact is absorbable relative to peers.
- Affordable models in June plus robotaxi halo could re-rate demand in H2; multi-tier FSD pricing and attention-monitoring relaxation create a monetization staircase.
- Musk's time returning to Tesla from May addresses the most-cited governance/brand overhang directly.
Bear case- The quarter was weak and the explanations are partly unfalsifiable: deliveries fell, margins fell, credits fell, other income swung negative — and management attributed it to a planned changeover plus "hostility," while simultaneously claiming no demand problem. Order-inflow data was requested and not given.
- Autonomy was quietly de-materialized for 2025: a 10–20 car pilot in one city with remote-operator backup, a launch software build that differs from customer cars, and financial impact deferred to mid/late 2026 — while the stock narrative remains autonomy-now.
- Guidance credibility is eroding: CapEx flipped in one quarter; the battery constraint vanished without quantification; the volume guide remains absent for a second straight call; the affordable "model" is now confirmed as a derivative of existing cars on existing lines with a slower ramp.
- Tariff exposure is real and growing: ~$2K/unit on vehicles, an "outsized" energy hit with US LFP capacity covering only a fraction of needs, higher CapEx, and a China export-license dependency for Optimus magnets — all while Musk's policy influence demonstrably did not prevent any of it.
- Brand damage is now a management-admitted demand factor with no quantification and no clear remediation plan beyond Musk spending less time in government.
- Earnings quality: Bitcoin swings now demonstrated in both directions ($472M sequential swing), regulatory credit revenue declining, and no ex-credit margin disclosure.
- HW3 retrofit — a confirmed liability last quarter — went entirely unmentioned, and the Austin build requiring audio input the customer fleet lacks raises fresh questions about the installed fleet's robotaxi readiness.
Next-quarter watchlist- June Austin launch: on time? Fleet size vs. the 10–20 stated, pricing, operating domain, intervention/incident disclosure, remote-operator usage rate, and whether the public can actually hail rides by end of June/July.
- Post-launch expansion: evidence of the "many other US cities by end of 2025" claim; any regulatory friction; whether personal-car unsupervised FSD ("before end of this year") gets a dated commitment.
- Q2 deliveries and margins: new Model Y ramp sustaining at rate; auto gross margin ex-credits; regulatory credit trajectory; whether brand-damage commentary continues or disappears.
- Tariff quantification: actual Section 232 impact vs. the ~$2K/unit model; US LFP equipment commissioning progress; non-China cell supply contracts; any resolution of the Optimus magnet export license.
- CapEx: whether ">$10B" holds or climbs; tariff share of the increase; AI vs. factory split.
- Affordable model: June production start evidence, pricing/monthly payment, form factor confirmation, and ramp pace vs. the "slower than hoped" caveat.
- Cybercab: Q2 first builds completed, test vehicles on public roads "in a couple of months," equipment installation at Giga Texas, 2026 production reaffirmed.
- Energy: deployment growth resuming after two sequential declines; record gross profit sustainability; Shanghai Megafactory ramp; any restatement of the ≥50% 2025 growth guide.
- Optimus: evidence of builds ramping in H2 toward "thousands"; magnet supply resolution; any design-lock or external-customer news.
- HW3 retrofit: any re-emergence of scope, cost, or timing disclosure after a silent quarter.
- Musk time allocation: evidence the May DOGE step-back actually happened; brand-perception data (order inflow, test drives, cancellation rates) in Q2.
|
| Jan 29, 2025 | +2.87% | Q4 FY2024 | Read transcript briefingQuarter in one view- Record Q4 deliveries at a ~2M annualized rate; Model Y named best-selling vehicle of any kind globally for 2024. But the quarter's mechanics were soft: auto margins declined QoQ on lower ASPs (discounts/financing to clear inventory) and the absence of Q3's one-time FSD revenue recognition; Q4 FCF was $2B, FY2024 FCF $3.6B.
- The headline commitment: paid unsupervised FSD (no one in the car) launching in Austin in June 2025 — Tesla's own fleet first, customer cars joining the network "probably next year." Musk also committed to unsupervised FSD in California and "many regions" of the US by end of 2025, nationwide in 2026.
- The 20–30% 2025 volume guide from Q3 was not repeated. Instead, Musk disclosed a new constraint: battery pack supply is the 2025 production limiter, and Taneja flagged "several weeks of lost production" in Q1 from the simultaneous global new Model Y changeover — with margin impact from idle capacity.
- CapEx guide cut to flat YoY for 2025 (vs. >$11B in 2024, itself raised from >$10B in Q3). Cumulative AI-related CapEx to date: ~$5B.
- HW3 retrofit is now a commitment, not a contingency: Musk's "honest answer" — Tesla will have to upgrade HW3 computers for FSD buyers; "painful and difficult," and he's "kind of glad that not that many people bought the FSD package."
- Optimus ambitions escalated dramatically: ~10,000 units internal plan for 2025 ("several thousand" more likely), <$20K cost at 1M units/yr, external sales maybe H2 2026, and a "north of $10 trillion revenue" long-term claim. Design is explicitly not locked.
- Energy: record Q4 deployments but sequential decline; 2025 guided to at least 50% growth (vs. >100% delivered in 2024); Shanghai Megafactory starting "this quarter"; a third Megafactory announced.
What management is focused on- Autonomy as the entire equity story, with a dated proof point. Musk framed 2025 as "maybe the most important year in Tesla's history" and a path to Tesla being worth "more than the next top five companies combined" — overwhelmingly from autonomous vehicles and Optimus. The June Austin launch is the near-term falsifiable anchor.
- Factory-floor autonomy as the demo. Thousands of cars daily already drive unsupervised from end-of-line to delivery parking at Fremont, expanding to Austin — used as evidence the system works without geofenced maps.
- Battery supply as the binding constraint. Musk: the executive team was discussing "just before this call" how to increase total GWh of battery output in 2025; cell allocation between stationary and mobile is an explicit trade-off.
- New Model Y changeover management. All factories switch simultaneously starting next month — Taneja called it unprecedented for the world's best-selling car and pre-flagged Q1 margin damage.
- Optimus industrialization narrative. Extended monologue on manufacturing being "100 times harder" than prototypes; current line designed for ~1,000 units/month, then 10K, then 100K/month lines; everything designed from first principles because no off-the-shelf actuators/sensors worked.
- Tariff risk acknowledged plainly. Taneja: tariffs are "very likely," Tesla remains reliant on global parts, and profitability will be impacted — notable given Musk's proximity to the administration driving the tariff policy.
Key numbers and quarter mechanics- Q4 FCF: $2B; FY2024 FCF: $3.6B despite CapEx increasing >$2.4B in 2024. 2025 CapEx guided flat YoY.
- Cost per car below $35,000 (materials-driven), achieved despite higher depreciation from new Model Y prep.
- Auto margin declined QoQ — lower ASPs (financing, discounts, inventory-clearing programs) plus the Q3 FSD revenue recognition not repeating. Lowest finished-goods inventory in two years was the goal and was achieved.
- Q4 net income included a $600M Bitcoin mark-to-market benefit under a new digital-asset accounting standard; mark-to-market now flows through other income every quarter — a new recurring earnings-quality wrinkle.
- Energy: record Q4 deployments, but down sequentially; Megapack and Powerwall both supply-constrained; 2025 deployments guided to ≥50% growth.
- OpEx up YoY and QoQ (AI R&D, sales/marketing, referral program); 2025 OpEx guided to increase further.
- Safety data updated: Q4 Vehicle Safety Report — one crash per 5.9M miles with Autopilot/FSD engaged vs. ~700K without; Ashok framed it as "8.5x safer." Note: Q3's figure was 7M miles, so the headline metric moved the wrong direction QoQ, and it remains Autopilot-based, not unsupervised-FSD intervention data.
- Service/other gross profit up YoY, down sequentially (service center costs, used-car profit decline).
Product and launch scorecard- Unsupervised FSD — dated and scoped. June 2025, Austin, paid, Tesla-owned fleet only, "toe in the water." Customer fleet add/subtract (Airbnb model) "probably next year." California and "many regions" by end of 2025; nationwide 2026. Europe supervised FSD pending Dutch/EU approval expected May; China constrained by data-transfer rules (workaround: internet video + simulation for bus lanes). This is more specific than Q3's "Texas and California next year" — and the CA piece is now a 2025 commitment.
- New Model Y — orders open in all markets; all factories start production next month; several weeks of lost Q1 production pre-announced.
- Affordable model — reaffirmed H1 2025, no new detail on price, form, or volume.
- FSD software — V13 shipped and improving; V14 will scale model size, context length (minutes of driving memory), audio/emergency-vehicle handling, and fleet-intervention data. HW3 got 12.6 ("baby V13"); HW3 releases will trail HW4; retrofit now confirmed as necessary for FSD buyers — timeline and cost not given.
- Optimus — not design-locked. 2025: internal factory use only, several thousand units hoped (10K internal plan). V2 design launches "sometime next year" (Musk wants early 2026, admits maybe mid-2026); external deliveries "very rough guess" H2 2026. Cost <$20K at 1M units/yr; price "set by the market." The Q2-era "V1 early 2025" framing has effectively slipped into an internal-use-only year.
- Semi — Reno factory roof/walls closed last week; first high-volume-design builds late 2025, ramp early 2026 (consistent with Q3's pilot H2 2025 / production H1 2026). Musk sized it at "probably like a $10B a year thing… $1B a month at some point."
- Solar roof — not cancelled; pivoting to certified-installer distribution rather than direct install; positioned as a premium product.
- FSD licensing — "significant interest from a number of major car companies," but Tesla won't engage until unsupervised FSD works across the US and volumes would be very high. No deals, no economics.
Sell-side read-through- Four analysts got questions (Roeska, Jonas, Ferragu, Levy) after a long retail Q&A — again a managed format, though the retail questions this time forced the HW3 retrofit admission and the Optimus design-lock denial.
- Roeska (Bernstein) asked for KPIs to track the autonomy thesis; Musk offered none — instead a defense of his prediction record ("media reports when I'm late, never when I'm early") and a restatement of the city-by-city 2025/2026 rollout. The evidentiary gap is now the analyst's problem to fill.
- Jonas (Morgan Stanley) spent both questions on LiDAR philosophy and US manufacturing policy — no financial questions. Musk confirmed pure-vision stance (radar in cars is turned off) and used the SpaceX Dragon LiDAR program to rebut the "doesn't understand LiDAR" critique.
- Ferragu (New Street) extracted the two most useful clarifications of the call: (1) June Austin is Tesla's fleet, not customer cars; (2) eyes-off supervised FSD (check-your-email mode) is "low single-digit months" away, with Musk admitting the current perverse incentive of drivers disengaging Autopilot to text.
- Levy (Barclays) probed EV-policy risk under Trump and whether Tesla would sell cars at no margin to seed the autonomy fleet. Musk deflected policy ("sustainable transport is inevitable") and answered the volume question with the battery-constraint disclosure — demand is not the issue, cells are.
- Notable absences: no analyst asked about the missing 20–30% growth guide, the flat-CapEx reversal, ex-credit margins, tariff quantification, or the HW3 retrofit cost. The hardest items on the sheet went unchallenged.
Management credibility- Delivered/consistent: record Q4 deliveries and energy deployments; cost per car below $35K; Shanghai Megafactory starting on the Q3-promised Q1 timeline; Semi Reno construction on schedule; affordable model H1 2025 reaffirmed; V13 shipped with claimed large improvements.
- Hardened commitments: June 2025 paid unsupervised FSD in Austin (from "sometime next year" in Q3); CA unsupervised in 2025; HW3 retrofit now stated as fact rather than "some chance."
- Softened/missing: the 20–30% 2025 vehicle growth guide was not restated — replaced by "battery packs are the constraint" and a Q1 production-loss warning; energy growth guide halved (≥50% for 2025 vs. >100% in 2024); CapEx flipped from raised (> $11B) to flat; Optimus V1 external timeline effectively pushed to internal-use-only 2025 with V2 in 2026.
- Under pressure: Musk's response to the KPI question was rhetorical, not analytical; the HW3 answer ("kind of glad not that many people bought FSD") is candid but confirms a real, unquantified retrofit program; the safety statistic cited (5.9M miles) is worse than Q3's (7M) yet was presented as improvement — "year-over-year improvement" was the claim, and the QoQ comparison was not addressed.
- Grandiosity escalated: "worth more than the next top five companies combined," Optimus "north of $10 trillion in revenue," "largest asset value increase in human history." These are unfalsifiable horizon claims that sit alongside very specific near-term promises — the June Austin date is where credibility will actually be tested.
What changed versus the prior quarter- The volume guide disappeared. Q3's explicit 20–30% 2025 growth was replaced by a battery-supply constraint and a warned-down Q1 (Model Y changeover). No new growth number was given.
- Autonomy got a date and a business model: June 2025 paid unsupervised rides in Austin with Tesla's fleet — versus Q3's "driverless paid rides sometime next year, TX definitely, CA probably." Customer-car participation slipped to 2026.
- HW3 went from disclosed risk to admitted liability: "some chance" (Q3) → "we're going to have to upgrade people's Hardware 3 computer" (Q4). Still no cost or unit quantification.
- CapEx trajectory reversed: raised to >$11B in Q3; now flat YoY for 2025, with "CapEx efficiency" emphasis and ~$5B cumulative AI spend disclosed.
- Energy growth guide halved: ≥50% for 2025 vs. "more than double" delivered in 2024; a third Megafactory was announced but undated/unlocated.
- Optimus timeline realism: Q3 reiterated V1 early-2025; Q4 says 2025 is internal-use only, design not locked, external sales H2 2026 at the earliest — while the long-term claim inflated to $10T+ revenue.
- New earnings-quality item: $600M Bitcoin mark-to-market gain in Q4 net income, with recurring mark-to-market going forward.
- Tariffs moved from background to named risk — "very likely," with profitability impact, from the CFO.
- FSD licensing interest claimed for the first time at this specificity ("significant interest from a number of major car companies"), though gated behind US-wide unsupervised FSD.
Bull case- A dated, funded, near-term autonomy catalyst: paid unsupervised FSD in Austin in ~5 months, factory fleets already running unsupervised daily at Fremont, V13 shipped, V14 scaling on every axis, Cortex cluster operational, and eyes-off mode "low single-digit months" away.
- Demand is not the problem — supply is. Management frames 2025 as battery-constrained, with record Greater China deliveries in a hyper-competitive BEV market and Model Y the world's best-selling vehicle.
- Cost leadership continues: sub-$35K cost per car, flat 2025 CapEx after the AI buildout, $3.6B FY FCF, and the new Model Y launching globally with orders already open.
- Energy remains a compounding second business: record Q4 deployments, supply-constrained on both Megapack and Powerwall, Shanghai coming online now, a third factory announced, ≥50% growth guided.
- Optionality stack is widening: Semi ramping early 2026 (~$10B/yr potential per Musk), FSD licensing interest from major OEMs, Optimus external sales from H2 2026 at claimed <$20K cost.
- If June happens, the licensing and fleet-monetization narratives become underwritable — and management explicitly set that test up.
Bear case- The growth story quietly de-rated. No 20–30% restatement, a battery constraint, a warned-down Q1, and energy growth guided to half of 2024's pace — while OpEx rises and tariffs loom. The near-term P&L setup (margin decline, ASP pressure, changeover costs) is weak.
- June is a high-wire act with a narrow definition: Tesla-owned fleet, one city, geofenced-in-practice caution — and customer-car robotaxi slipped to 2026. Any incident or delay hits a stock priced on autonomy.
- HW3 retrofit is now a confirmed, unquantified cost affecting FSD buyers, with software already trailing HW4 — and Musk's relief that "not that many people bought FSD" undercuts the FSD attach-rate narrative.
- Earnings quality questions: $600M Bitcoin gain in Q4 net income, recurring crypto mark-to-market ahead, no ex-credit margin disclosure, and ASPs still falling on incentives.
- Optimus claims are escalating faster than the product: design not locked, 2025 reduced to internal use, external sales a "very rough guess" for H2 2026 — paired with a $10T revenue assertion.
- Regulatory and political exposure is two-sided: Europe approval not expected before May, China data rules unresolved, and US tariff policy — from the administration Musk advises — is a named profitability risk.
- Safety evidence still doesn't match the claim: the 5.9M-mile figure is supervised/Autopilot data, worsened QoQ from 7M, and no unsupervised intervention data was published ahead of a June driverless launch.
Next-quarter watchlist- June Austin launch: does it happen on time, paid, and driverless? Fleet size, operating domain, pricing, incident rate, and whether any regulator intervenes.
- Eyes-off supervised FSD: the "low single-digit months" promise — shipped or slipped?
- 2025 volume framing: is the 20–30% guide formally reinstated, revised, or abandoned? Battery GWh additions and any supplier/4680 capacity detail.
- Q1 deliveries and margins: magnitude of the Model Y changeover hit vs. the "several weeks" warning; ramp speed in March; auto gross margin ex-credits.
- HW3 retrofit: program scope, unit count, cost, timing, and whether it hits CapEx or service margins; HW3 software parity with V13/V14.
- Affordable model: H1 2025 launch evidence — pricing, form factor, line allocation — with half the window now elapsed.
- Energy: Shanghai Megafactory ramp, third-factory location/timing, and whether ≥50% 2025 growth tracks after the sequential Q4 decline.
- Optimus: evidence of several thousand units built and doing useful factory work in 2025; V2 design-lock timing; any external customer commitments.
- Tariffs and policy: quantified tariff exposure, supply-chain mitigation, and any federal AV framework movement tied to Musk's government role.
- FSD licensing: any named OEM, pilot, or term sheet — or confirmation it stays gated behind US-wide unsupervised FSD.
- Bitcoin/accounting: size of recurring mark-to-market swings in other income and their distortion of operating trends.
|
| Oct 23, 2024 | +21.92% | Q3 FY2024 | Read transcript briefingQuarter in one view- Record Q3 with a new volume guide. Record deliveries, record operating cash flow of $6.3B, 7-millionth vehicle produced, auto margins improved QoQ (helped by FSD revenue recognition, volume, localized deliveries), and Musk gave his first explicit 2025 volume guide: 20–30% vehicle growth, caveated for "force majeure" events.
- The $25K car is dead; Cybercab is the $25K car. Musk called a regular $25K model "pointless… completely at odds with what we believe." The affordable model (H1 2025, sub-$30K with incentives) and Cybercab (~$25K cost, volume production 2026, at least 2M units/year, maybe 4M) are now the entire affordability strategy.
- Autonomy timelines got more specific and more falsifiable: internal estimate for FSD safer-than-human is Q2 2025 ("may end up being Q3"); paid driverless ride-hailing in Texas and California next year (CA contingent on approval); v13 expected to deliver 5–6x improvement in miles between interventions vs. 12.5, and ~1,000x vs. January 2024.
- Hardware 3 risk disclosed for the first time: "some chance that Hardware 3 does not achieve the safety level that allows for unsupervised FSD" — with a promise to upgrade HW3 FSD buyers for free if so. A potentially large contingent cost now on the record.
- Energy margins hit a record >30% despite a Q3 deployment decline; Lathrop at 200 Megapacks/week (40 GWh/yr run rate); Shanghai Megafactory starts at 20 GWh/yr in Q1 2025; full-year deployments still guided to more than double 2023.
- CapEx guide raised to >$11B (from >$10B) on AI compute; 50K GPU cluster in Texas on track for end of this month, ahead of schedule.
What management is focused on- Autonomy as identity, now with a manufacturing moat narrative. Musk claims Cybercab's line is "half an order of magnitude better" than other car lines — "the hottest Tesla product to copy will be the factory." The 10/10 event is cited as proof: 20 Cybercabs + 30 Model Ys running autonomously all night, "thousands" of rides, no incidents, no steering wheels/pedals.
- FSD metrics and the validation bottleneck. Musk: "not currently training compute constrained" — the constraint is now validation (at ~10,000 miles between mistakes, distinguishing software versions takes a long time). Ashok Elluswamy confirmed 100x improvement already in 2024, targeting 1,000x by v13, crossing human miles-per-critical-intervention by Q2 2025.
- Demand via financing and awareness. ASPs fell on financing incentives (recognized as upfront revenue reduction via third-party banks); management plans repeated 30-day FSD trials with each major software release and claims FSD take rate "improved substantially especially after the 10/10 event" — again no quantification.
- Cost per vehicle as the core auto lever. Q3 was the lowest cost per vehicle ever; Musk gave an extended "game of pennies" monologue (10,000 items × $4 = $40K car) framing how hard the affordable model is.
- Energy scale-up: Lathrop 40 GWh run rate, Shanghai 20 GWh starting Q1, path to 100 GWh/yr "not long" away, "multiple terawatt hours" ultimately; lithium refinery starting production in 2025.
- Regulatory strategy: state-by-state AV deployment (Texas easy, California slow), plus a new push for a federal AV approval process — Musk explicitly tied this to his "department of government efficiency" role: "I'll try to help make that happen."
Key numbers and quarter mechanics- Operating cash flow: record $6.3B. CapEx ~$3.5B in Q3, sequential increase on AI compute; FY2024 CapEx now >$11B (raised from >$10B).
- Auto revenues grew QoQ and YoY despite ASP reduction from financing incentives; auto margins improved QoQ on FSD/feature revenue, volume, and localized deliveries (lower freight/duties). Taneja: "Sustaining these margins in Q4 will be challenging."
- FSD (Cybertruck release + Actually Smart Summon) contributed $326M of revenue in the quarter — the first time FSD revenue has been quantified at this scale.
- Regulatory credits: "elevated," >$2B of revenue YTD — still a major margin contributor with no ex-credit margin disclosed.
- Energy: Q3 deployments declined (lumpiness reiterated), but record energy margin >30% (project mix); Q4 sequential growth expected; full-year deployments to more than double 2023.
- OpEx declined QoQ and YoY (Q2 restructuring savings partially offset by AI spend).
- Safety data published: Q3 vehicle safety report shows one crash per 7M miles on Autopilot vs. ~700K US average (10x) — the first quantitative safety figure offered in this autonomy push, though it is Autopilot, not FSD, and not intervention data.
- Smart Summon: >1M uses already; ride-hailing app live for Bay Area employees with safety drivers, tested "the better part of the year."
- Production rate claim: ~35,000 autonomy-capable vehicles/week vs. Waymo's fleet of <1,000.
Product and launch scorecard- Affordable model — on track, H1 2025, sub-$30K with incentives. Moravy confirms factory prep underway. Still no product detail; the $25K non-robotaxi variant was explicitly killed on this call.
- Cybercab — dated and sized. Volume production 2026 ("I do feel confident"), ≥2M units/yr, maybe 4M, multiple factories, ~$25K cost, purchasable individually. Manufacturing revolution claims (5x factory, part deletion) are unverified assertions.
- FSD — v13 "going out soon," 5–6x better miles-between-interventions than 12.5; safer-than-human internal estimate Q2 2025 (Musk: "not sandbagging"). HW3 may not reach unsupervised safety — free upgrades promised if so; HW4-first development with backporting "takes longer."
- Ride-hailing — real but supervised. Employee service live in Bay Area with safety drivers; public rollout Texas (definitely) and California (probably, regulatory-contingent) in 2025, other states possible; driverless paid rides "sometime next year."
- Semi — back on the agenda with dates. Reno factory capex released; pilot builds H2 2025, production H1 2026, ramping through 2026; "ridiculous demand" (Pepsi drivers cited); all Semis FSD-hardware-equipped, FSD once fleet training matures. This resolves last quarter's silence but pushes external volume to 2026.
- 4680 — "tracking to be the most competitive cell in North America" fully landed net of incentives/duties; "not quite there yet, but close." Supplier cells remain the majority. No dry-cathode Q4-launch confirmation given on this call.
- Optimus — new hardware detail: next-gen hand/forearm with 22 degrees of freedom (double prior), better tactile sensing; "most advanced humanoid robot by a long shot." No update to V1 early-2025 / end-2025 several-thousand-units timeline.
- Roadster — design "close to finalizing," explicitly deprioritized ("cherry on the icing on the cake"); flying-car tease. No production date reiterated (Q2 said 2025).
Sell-side read-through- Only two analysts got questions (retail/investor questions dominated) — itself a notable format choice on a quarter with a raised CapEx guide and a new volume target.
- Pierre Ferragu (New Street) pressed compute strategy (bigger models vs. faster training) and ride-hail rollout sequencing. Answers: real-world AI is a different problem than LLMs (gigabytes of video context distilled to few control outputs; heavy training compensates for ~200W inference); rollout will follow each state's safety-driver requirements, with driverless paid rides targeted next year. No fleet-size or economics detail given.
- Adam Jonas (Morgan Stanley) asked the governance question directly: is xAI competitive with Tesla for talent, tech, and Musk's time? Musk's answer shifted from last quarter's defensive framing to a benefit narrative — xAI "has been helpful to Tesla AI quite a few times" (training-run recovery cited as recent example) — plus "different problems" (AGI vs. real-world autonomy). No mention this quarter of the shareholder vote on a Tesla–xAI investment; the conflict question remains open.
- Notable absences: no analyst challenge on the 20–30% growth guide, the HW3 retrofit liability, ex-credit margins, or the killed $25K car — the Q&A format insulated management from the hardest follow-ups.
Management credibility- Delivered: 50K GPU cluster tracking to deploy by end of October "ahead of schedule" (Q2 promise); affordable model H1 2025 reaffirmed; 10/10 robotaxi event happened on the rescheduled date with a working demo; record Q3 deliveries and OCF; energy on track for >2x annual deployments.
- New falsifiable stack: 20–30% 2025 vehicle growth; FSD safer-than-human Q2 2025 ("extremely likely" within 2025); v13 5–6x improvement; driverless paid rides in TX/CA 2025; Cybercab volume production 2026 at ≥2M/yr; Shanghai Megafactory 20 GWh in Q1 2025; Semi pilot H2 2025 / production H1 2026; free HW3 upgrades if needed; FY CapEx >$11B.
- Slipped/softened: the $25K standalone car is cancelled (a multi-year promise, reframed as "pointless"); Roadster's 2025 production language from Q2 was not repeated; dry-cathode Q4 launch and year-end 4680 cost-parity targets were not reaffirmed on this call (replaced by "close to most competitive"); Europe/China supervised-FSD approval "before end of the year" (Q2) was not mentioned.
- Under pressure: Musk's safer-than-human Q2 2025 estimate was immediately caveated by Ashok as "unvarnished internal estimate"; the HW3 admission is candid but creates a retrofit liability of unknown size; the 10x safety figure is Autopilot-based, not FSD intervention data — the core evidentiary gap persists.
What changed versus the prior quarter- A volume guide exists again. Q2 pointedly did not restate "higher sales than 2023"; Q3 gives 20–30% growth for 2025 — the most concrete volume commitment in over a year.
- The $25K car is officially dead, replaced by Cybercab-as-affordable-car. This resolves the derivative-vs-new-platform ambiguity by eliminating the conventional option entirely.
- FSD revenue quantified for the first time at scale ($326M) and a safety statistic published (10x vs. US average) — partial answers to the disclosure drought, though take rates and intervention data remain withheld.
- HW3 capability doubt disclosed — a new risk and contingent cost that did not exist in prior quarters' framing.
- CapEx raised (>$10B → >$11B) even as management claims to be "judicious" on AI spend and not compute-constrained.
- Semi reappeared with a schedule (pilot H2 2025, production H1 2026) after going unmentioned in Q2 — but external volume is later than Q1's framing implied.
- xAI narrative flipped from defense to synergy ("helpful quite a few times"); the shareholder-vote investment idea went unmentioned.
- Regulatory posture shifted from "morally obligated" rhetoric to process: state-by-state pathway detail, FMVSS compliance emphasis, and a federal-AV-approval push explicitly linked to Musk's government-efficiency role — a new political dimension.
- Margin trajectory improved but with a warning: Q2 was "flat sequentially with record credits"; Q3 improved QoQ with $326M FSD revenue and >$2B YTD credits — and management itself flagged Q4 sustainability as "challenging."
Bull case- Record quarter across the P&L: record deliveries, record $6.3B OCF, lowest cost per vehicle ever, auto margins up QoQ, record >30% energy margins — while the rest of the EV industry is, per Musk, unprofitable and shrinking.
- A dated, funded autonomy path: v13 imminent with claimed 5–6x improvement, safer-than-human estimated Q2 2025, paid driverless rides in TX/CA next year, ride-hail app already built and in employee use, 50K-GPU cluster ahead of schedule, and no training-compute constraint.
- 2025 growth is now guided (20–30%) with identifiable drivers: affordable model H1, financing tailwinds from falling rates, FSD trials driving take rate "substantially" higher post-10/10.
- Energy is compounding: 40 GWh Lathrop run rate + 20 GWh Shanghai from Q1, path to 100 GWh/yr, backlog filling 2025 slots, record margins.
- 4680 nearing cost leadership fully landed in the US — strategically valuable if tariffs raise imported-cell costs.
- Cybercab manufacturing claims (5x factory efficiency, ≥2M units/yr in 2026) would, if real, be a structural cost advantage no competitor is attempting.
Bear case- The 20–30% guide rests on unproven drivers — an undefined affordable model, rate cuts, and autonomy demand — and comes pre-caveated with "force majeure" outs. Musk's volume-prediction record is weak.
- Margin quality still opaque: $326M FSD revenue and >$2B YTD credits flattered Q3; ASPs fell on incentives; management itself says Q4 margins will be "challenging." Ex-credit auto margin remains undisclosed.
- HW3 retrofit liability: if HW3 can't support unsupervised FSD, Tesla owes free computer upgrades to potentially millions of FSD buyers — unquantified cost, unquantified probability ("some chance").
- The $25K car cancellation removes the conventional volume bridge; the growth story now depends on a steering-wheel-less vehicle with no federal approval framework, state-by-state deployment, and a 2026 volume date from a CEO whose autonomy timelines have slipped repeatedly.
- Safety evidence remains mismatched to claims: the 10x figure is Autopilot crash data, not FSD intervention rates; "safer than human by Q2 2025" is an internal estimate with no published methodology.
- Governance/political entanglement is deepening: Musk explicitly plans to use a government role to shape federal AV regulation that directly benefits Tesla — a new regulatory-capture optic on top of the unresolved xAI resource question.
- CapEx up to >$11B with "judicious" spend language suggests the AI buildout's returns are still ahead of the costs.
Next-quarter watchlist- Q4 deliveries and the 2024 exit rate as the base for the 20–30% 2025 guide; any early walk-back of that number.
- v13 rollout: actual vs. claimed 5–6x miles-between-intervention improvement; any published FSD (not Autopilot) safety/intervention data.
- Ride-hailing: Texas/California regulatory filings and approvals; safety-driver removal milestones; fleet size, pricing, and economics of the employee pilot.
- HW3: any quantification of the installed FSD-on-HW3 base and the retrofit cost scenario; whether v13 features ship on HW3 or slip.
- Affordable model: H1 2025 timing language, factory/line evidence, pricing detail vs. the "sub-$30K with incentives" framing.
- Margins: Q4 auto gross margin ex-credits against management's own "challenging" warning; credit revenue trajectory; financing-incentive drag.
- 4680: did dry-cathode launch in Q4 as previously guided? Year-end cost-parity claim status; "most competitive cell in North America" evidence.
- Energy: Q4 deployment rebound to secure the >2x full-year guide; Shanghai Megafactory Q1 start; margin durability after the record 30%+.
- Semi: Reno construction progress toward H2 2025 pilot builds; customer commitments beyond Pepsi.
- Governance: any Tesla–xAI investment proposal or shareholder vote; Musk's government role and its interaction with federal AV rulemaking; Optimus V1 production timing (early 2025 per Q2) and factory unit count.
|
Corbanu