Margins Rebound and Cash Turns Positive, but the Payoff From Robotaxis and Robots Slides to Late 2027

Results vs. Consensus

Q3 2026 Scorecard

Year-Over-Year and Quarter-Over-Quarter Comparisons

Quality of Beat

• Revenue: In line and flat. Total revenue of $28.05B was within 0.2% of consensus, 0.2% below Q3 2025 and 0.7% below Q2's record. The mix moved hard: automotive fell 3% (deliveries -2%, credits -73%, lower ex-FX ASP), energy fell 16% (partly from tariff-related customer refunds), and Services & Other grew 33%. FX added about $0.3B. Tesla has now posted two consecutive quarters of trailing-twelve-month revenue above $100B ($103.6B), but there is no top-line growth here; the quarter's improvement is entirely in margin.
• Margins: All three segment margins rose sequentially. Automotive gross margin was 18.5% GAAP (18.1% ex-credits), energy 33.1% and Services & Other a record 15.9%. The automotive gain had no disclosed one-time item, which is the opposite of Q1's 19.2%. Energy's gain was mostly one-time by management's own description. Operating margin of 2.6% remains the binding fact: gross profit grew 8.7% YoY while operating expenses grew 39.2%.
• EPS: The $0.03 non-GAAP beat is operational, driven by the gross profit line. Non-operating items were modest (net interest income of $355M against $117M of other expense, mostly FX), the effective tax rate was a low 15.9%, and diluted shares were flat at 3.54B. The GAAP line tells the investment-cycle story more honestly: GAAP EPS of $0.22 is down 44% YoY, with operating income down 56% and stock-based compensation inside it up 76% to $1.17B pre-tax.
• Cash: Positive free cash flow of $1.53B reversed July's "negative" framing and lifted cash and investments to $45.4B. Year-to-date free cash flow is +$1.88B. The quality caveat in the callout above applies: the quarter was helped by supplier and customer float, and the guided Q4 capex of roughly $7.0B makes Q4 free cash flow likely negative again.

Revenue assessment. A flat top line with deliveries down 2% YoY and the largest order backlog in four years is not a demand problem; management says production, gated by battery availability, is the constraint. The more interesting signal is the composition. Services & Other is now 16% of revenue and growing by a third a year, and it holds supercharging, insurance, service, used vehicles and the nascent Robotaxi service. FSD subscriptions sit inside automotive revenue as ancillary sales, which is part of why automotive revenue per vehicle held up while ex-FX vehicle prices fell. Both are where the future revenue mix is being built, and neither is disclosed on its own.

Margin assessment. We downgraded in July because the clean auto margin base had re-set to ~16.3%. The 18.1% print is the first evidence that the re-set was not a floor going lower. It is a single quarter, and regional mix can reverse, but it was achieved with credits down to $113M and with the subvention and commodity headwinds still present. Energy's 33.1% should be read as a one-time high: the clean level is unknowable from the disclosure, and Tesla's own July guide put long-run energy margins in the "mid- to low 20%" range.

EPS assessment. Year-to-date non-GAAP EPS is $1.22. With operating expenses guided to keep climbing "till we commercialize these new products," we see FY26 non-GAAP EPS of $1.65–1.80, up from our July range of $1.40–1.60. The beat moves the earnings base up, but it does not change the shape: earnings are suppressed by design through 2027, and the multiple rests on businesses that do not yet report revenue.

Segment Performance

Automotive: Supply-Gated Volume, and a Unit-Cost Story Instead of a Price Story

Deliveries of 486,532 beat the 461,974 company-compiled consensus by about 5% and made a second consecutive quarter above 480,000 for the first time. Production of 464,391 trailed deliveries by about 22,000 units, and days of supply fell to 11, the leanest since Q3 2025. Management said the quarter ended with the largest passenger-vehicle backlog in four years, records in Canada, Australia, Japan and Taiwan, record production at Gigafactory Berlin, and the Model Y as Europe's top-selling BEV and South Korea's top-selling passenger car by preliminary data. The stated constraint is battery availability, not demand.

"On automotive margins, excluding credits, increased sequentially from 16.3% to 18.1%. The improvement in margins is driven by a higher mix of deliveries in APAC. While commodity price increases and interest rate changes continue to add to our cost and create further headwinds, we have adjusted the vehicle financing programs to partially offset for such costs." — Vaibhav Taneja, CFO

Our per-unit math on the income statement (automotive sales revenue and cost of automotive sales, divided by deliveries excluding operating leases) shows where the margin came from. Revenue per vehicle was about $41,900, down roughly $440 from Q2. Cost per vehicle was about $34,550, down roughly $1,140. Gross profit per vehicle therefore rose to about $7,350 from $6,650 in Q2 and $6,150 a year ago. A higher share of deliveries in Asia-Pacific and adjusted financing programs explain most of it, by management's account. One cross-current deserves attention: the CFO disclosed that Tesla is "taking extra steps to sell certain cars in Europe with some benefits" ahead of broader FSD approval, which is a price concession in a region where the brand is otherwise supply-constrained.

Assessment: This is the first quarter since we downgraded in which the automotive margin moved for structural reasons we can observe in the cost line rather than through one-time credits. We do not call the pillar repaired on one print, because Asia-Pacific mix is exactly the kind of driver that reverses, and Europe is getting incentives. But the clean base is now demonstrably above 17% in at least one quarter of the post-credit era, and the FSD-attach dynamic (below) gives a mechanism for keeping it there.

Energy: A One-Time Margin on Shrinking Revenue, With a Sold-Out 2027 Behind It

Energy was the quarter's most confusing segment. Revenue fell 16% YoY to $2.88B, partly from tariff-related refunds passed to customers, even as deployments rose 10% to 13.7 GWh (and missed the ~15.9 GWh consensus). Gross margin jumped to 33.1% from 20.4%, but management attributed the improvement primarily to two one-time items in cost of revenues: tariff refunds recognized in the quarter, and warranty true-downs as Tesla aligned with its cell vendor on reimbursement of the cell-related costs it had charged in Q2. Neither was sized. Gross profit of $954M was still 11% below the year-ago $1.07B.

"Energy gross margins improved from 20.4% to 33.1%. The sequential margin improvement is primarily attributed to tariff refunds recognized in the quarter and warranty true-downs resulting from alignment with our vendor on certain cell-related costs to be reimbursed by them." — Vaibhav Taneja, CFO

The forward picture changed more than the quarter did. Megafactory Texas began production with 50 GWh of planned capacity, Tesla shipped its first Megapacks built entirely with its own LFP cells from Nevada, and the order rate and backlog were the highest ever. The CEO said Tesla is "almost sold out of all Megapacks for 2027" and expects gross margins "to increase meaningfully," and the CFO tied that to ASP improvement from data-center demand. In July the guide was normalization to the "mid- to low 20%" range.

Assessment: The direction of energy guidance has reversed, from margin compression to margin expansion, on data-center demand that lets Tesla price up. That is a genuine positive for the bear point we carried, but this quarter's 33.1% cannot be used as evidence for it, because both cited drivers are one-time. Year-to-date energy gross margin is about 30%, flattered in Q1 and Q3 and depressed in Q2. Deployment volumes were flat sequentially despite record demand, so the binding constraint is again production, which Megafactory Texas and in-house LFP should address in 2027.

Services & Other: The Line That Will Carry Autonomy Revenue

Services & Other revenue grew 32.5% YoY to $4.61B and gross profit doubled YoY to a record $734M, a 15.9% margin. Management credited strength across several businesses, naming supercharging and insurance. The Robotaxi service, now including Cybercab rides in Austin, is reported here and is too small to see.

Assessment: Services margins have risen for three straight quarters (8.8% in Q4 2025, then 9.2%, 14.1% and now 15.9%), and gross profit has gone from about a tenth of automotive's a year ago to about a fifth. This is the line where Robotaxi revenue would first appear, so the fact that management will not break it out is the main disclosure gap in the print.

Key KPIs

Key Topics & Management Commentary

Overall Management Tone: Management was more specific about dates than at any call this year, and for the first time in three quarters the financial results supported the confident tone instead of contradicting it. The cost of that specificity is that most of the dates landed in late 2027, and management delivered them as matter-of-fact engineering timelines rather than as setbacks. Compared with July, when management leaned into the investment cycle to defend a 1.4% operating margin, this call was less defensive about the present and more measured about the future, and it was disciplined about refusing every SpaceX question.

1. The Clean Auto Margin Recovers to 18.1%

The single most thesis-relevant number in the print is 18.1%. In July, the CFO's bridge showed that the clean automotive margin had sat near 16.3% for two quarters once Q1's $230M of warranty and tariff benefits were removed. This quarter the bridge contained no such items: the CFO named Asia-Pacific mix as the driver and described financing-program adjustments as a partial offset to rising commodity and rate costs. The deck's year-over-year profit bridge lists "improved vehicle region and product mix, partially offset by lower vehicle average selling price" alongside higher FSD-driven ancillary revenue.

"Demand is growing rapidly for our vehicles. It's exceeding our production by a significant margin. And obviously, we're going to be ramping and we are ramping production meaningfully. The biggest driver for demand is full self-driving." — Elon Musk, CEO

That demand framing matters for the margin because a supply-constrained manufacturer can choose its mix. Tesla sent more cars to higher-margin Asia-Pacific markets, and it reduced financing support where it could. The countervailing evidence is the Europe incentive program and the deck's admission that ASPs excluding FX fell again.

Assessment: One of the two clean quarters above 17% that we required in July is now in hand. We move the margin pillar from broken to at risk rather than on track, because mix-driven gains are the least durable kind and the next print will tell us whether 18% is a level or a spike.

2. Capex Cut to About $20B, Funded From Operations, Debt Undrawn

The July guide was "more than $25 billion" for FY26, with capex growing for two to three years and up to $30B of debt capacity arranged. Year-to-date capex is $13.0B, and the new guide of about $20B implies a Q4 of roughly $7.0B and H2 spending near $11.7B, against the more than $16.7B of H2 spending the July guide implied. The CFO said FY26 spending will be funded "primarily from our cash flows from operations" and that no draws are planned under the new facilities this year.

"We're revising our CapEx guidance to be approximately $20 billion for the current year as a result of managing our capital deployment efficiently and plan to fund it primarily from our cash flows from operations. Additionally, during the quarter, we secured capacity to borrow up to $30 billion to support the expected growth in CapEx." — Vaibhav Taneja, CFO

The CEO's framing in Q&A was that capex is "ending up being a little lower than we anticipated" and will become "more meaningful next year and then noticeable like in the tens of billions in '28." The CFO also said capex "for the next few years would continue to expand" across the Robotaxi fleet, Optimus capacity, solar manufacturing, SI compute and the semiconductor fab. Tariff refunds are being reinvested in US production capacity.

Assessment: The FY26 cash risk that drove our July downgrade is materially smaller: we now see FY26 free cash flow between about ($1.1B) and breakeven instead of ($7B) to ($5B). But management did not say which projects slowed, and every forward statement points to higher spending in 2027 and 2028. We read the cut as a timing shift with genuine efficiency elements, not as a smaller program, and we hold FY27 capex above FY26.

3. Energy: Margin Guidance Flips Upward on Data-Center Demand

The energy disclosure changed character. In July the message was normalization to the "mid- to low 20%" range after a warranty charge. In October the message was extreme data-center demand, the highest-ever order backlog, a nearly sold-out 2027, and margins rising with ASPs. Megafactory Texas started production and the first Megapacks with Tesla-made LFP cells shipped from Nevada.

"We're actually almost sold out of all Megapacks for 2027, and we're ramping production as quickly as possible to meet demand. We also expect gross margins on that to increase meaningfully." — Elon Musk, CEO

"As Elon said, the business is witnessing extreme demand from data centers, and we are doing our best to increase production. And accordingly, we'll see improvements in ASP and further improvements in margins as we move forward." — Vaibhav Taneja, CFO

The tension is in the numbers: segment revenue fell 16% YoY and deployments missed estimates by about 14%. Management linked part of the revenue decline to refunds of tariffs that had been passed to customers, which means some of Q3's margin uplift and some of its revenue decline are two sides of the same tariff unwinding.

Assessment: We move the energy-margin bear point back from materializing to emerging. Pricing power from data-center demand is a better setup than we modeled in July, and in-house LFP cells reduce the vendor exposure that caused Q2's charge. But a sold-out 2027 also means 2027 volume is capped by Megafactory ramps, so the upside runs through price and margin, not volume.

4. FSD Is Now the Sales Pitch, and Its Revenue Is Smaller Than the Customer Count Implies

FSD paid customers reached 1.74M, up 260,000 in the quarter and 67% YoY. In North America, over 75% of deliveries included a paid FSD subscription sign-up at delivery, up from more than 55% in Q2, even though every new car includes a free month. For the first time, half of paid customers are on subscriptions rather than upfront purchases. Approvals in Croatia, Slovenia, the Czech Republic and Slovakia brought FSD (Supervised) to nine EU countries and over 100,000 European users.

"In Q3, over 75% of deliveries in North America included a sign-up for paid FSD subscription at the time of delivery, which is approximately 20% higher than Q2. Overall, the FSD attach rate continued to improve, reaching over 1.7 million paid customers globally, of which, for the first time, 50% is based on subscriptions." — Vaibhav Taneja, CFO

In Q&A, the CFO confirmed an analyst's estimate that FSD recurring revenue is approaching $1 billion a year. That is a useful anchor, and it exposes an error in our own July framework, which carried an FY26 exit FSD ARR of $2.3–2.7B and in effect priced every paid customer as a monthly subscriber. Half of them paid upfront, so the recurring stream is roughly $1B today.

Assessment: FSD has become Tesla's primary demand driver in the US, and the attach curve is steeper than we expected. But the recurring-revenue base is about half the size we had modeled, so the pillar stays on track on adoption while our FSD revenue line is cut. The more important consequence is indirect: FSD pull is what lets Tesla choose its mix and trim subvention, which is the mechanism behind the margin rebound.

5. Robotaxi and Cybercab: Purpose-Built Vehicles in Service, Eightfold Miles, Six Cities

Tesla started providing Robotaxi rides in Cybercabs in Austin on September 3, and the CEO said the Austin Cybercab fleet has grown tenfold since launch. Cumulative unsupervised miles reached nearly 2 million by quarter-end, about eight times the start-of-quarter level, with double-digit weekly growth throughout. Austin service hours now run 6am to midnight, with 2am next. The Robotaxi service operates in six cities, and Dallas gets Cybercab next month. Management targets 24/7 operation and higher-speed roads by year-end, Cybercab sales to customers by the end of 2026, and the first customer-owned vehicles joining the Robotaxi fleet around Q2 2027.

"The Cybercab fleet in Austin, for example, has grown by a factor of 10 since launch about a month ago." — Elon Musk, CEO

"We suspect somewhere around Q2 next year is when we can turn on some of the first customer vehicles to be part of the Robotaxi fleet." — Ashok Elluswamy, VP of AI

Two things were absent. First, no base for the tenfold fleet claim and no revenue or cost figures. Second, the July call's Phoenix and Las Vegas preparation did not come up, and the city count is unchanged at six. Separately, NHTSA opened an audit in early September into how Tesla certified the Cybercab, which has no steering wheel or pedals, against federal motor vehicle safety standards. A retail question about that review was answered only on the technology side.

Assessment: Operationally, this is the quarter Robotaxi stopped being a pilot: purpose-built vehicles are carrying riders in a commercial service and the miles curve is steep. Financially, nothing changed, and the federal certification audit is a new, concrete regulatory risk to the vehicle that the economics depend on. The pillar stays on track; the disclosure gap stays open.

6. Robotaxi Economics: Short of the $0.30 Goal Today, Revenue in 2027

The unit-economics question came from retail shareholders and from analysts, and the answer was the most specific yet. The CFO said Tesla is not yet at its $0.30-per-mile cost goal but sees a path through utilization, pickup and drop-off, and cleaning. The CEO dated the revenue: not this quarter, noticeable in Q2 2027, a big deal by Q3 or Q4 2027.

"So are we at our goal of $0.30 today? No. But do we see a path to get to that? Very -- yes, we do. And it's all about scaling operations." — Vaibhav Taneja, CFO

The CEO also restated the scope of ambition when asked about pricing below rideshare: absent government protectionism, he expects Tesla to carry "a majority of all passenger miles on Earth in any country where it is allowed to operate freely."

Assessment: This is the first time management has put a quarter on Robotaxi revenue materiality, and it is three to five quarters away. That is useful, because it gives us a dated test, but it also means our July condition for an upgrade (a first economics disclosure showing a credible path to contribution) will not be met in the next two prints. We cut our FY26 Robotaxi revenue estimate and move the first meaningful year to 2027.

7. Optimus: Production Started, Customers at the End of 2027

The July commitment was Optimus production "later this year." Production began on the Fremont high-volume line in Q3, and the deck calls these the first production-quality bots Tesla has built. The CEO used most of his opening remarks to reset the slope: the vast majority of a 1-million-unit-per-year line is installed, but reaching that annualized rate will take "at least a year, maybe a little longer," the hand design is still changing, the AI5 chip goes into Optimus around April 2027, and deliveries to companies outside Tesla are expected around the end of 2027. First uses will be industrial, round-the-clock, dangerous or repetitive tasks. A non-preprogrammed demo video could follow in "maybe a month or 2," and an Austin Optimus factory is under construction.

"But we do have the vast majority of a 1 million unit per line production line landed in Fremont right now. But I would expect that it takes us probably at least a year, maybe a little longer to get to that annualized rate to get to 20,000 units a week." — Elon Musk, CEO

"I do expect these issues to be solved around the end of next year. So that's the time to expect when we're delivering Optimus to companies outside of Tesla." — Elon Musk, CEO

The design-maturity detail came from an Optimus engineering executive in Q&A: the electromechanical system is "relatively high confidence," while the hand and some bespoke gears are still in development. The CFO said cost per unit is trending toward where Cybercab cost is expected to land, without giving a number.

Assessment: Tesla met its production-start commitment for the first time this year, after two slips. But a line producing robots whose hands are still being redesigned is a pre-commercial line, and the internal plan to build 10,000 to 20,000 bots for self-play training means early output is a cost, not revenue. We keep the pillar at risk, with outside revenue now a late-2027 event at the earliest.

8. Terafab: First Dates and a Price Tag, With the SpaceX Split Undecided

July deferred Terafab detail to a standalone event. This call supplied a schedule: equipment commissioning at the Austin research fab in the first half of 2027, a first test chip around May or June 2027, and "some kind of useful chip at volume around Q3 of 2028." Most equipment orders for the research fab are placed. In Q&A the CEO put the capital at "tens of billions" of equipment through early 2028 and said it will "probably be north of $100 billion long term," with the split between SpaceX and Tesla still to be decided.

"We'll begin commissioning machinery in the first half of next year. And hopefully, if things go well, we'll make our first test chip just to see if we can make any chip at all around May or June." — Elon Musk, CEO

Assessment: The disclosure improves our ability to underwrite Terafab's timing, and the "see if we can make any chip at all" phrasing is a candid acknowledgment of execution risk. The unresolved cost split is the problem: the largest capital program in Tesla's plan is being negotiated between two companies with the same controlling shareholder at the same time their boards are evaluating a merger. We keep the AI-silicon pillar at risk.

9. Operating Expenses: Growing "Till We Commercialize"

Operating expenses rose to $4.78B, up 39% YoY and 10% sequentially. R&D drove the increase (+69% YoY to $2.76B): SI development, depreciation on added compute, chip development and pre-production costs for Optimus, Megapack 3 and Semi. Stock-based compensation of $1.17B was up 76% YoY, reflecting the 2025 CEO Performance Award. Depreciation, amortization and impairment rose 14% sequentially to $1.85B.

"This trend of growing OpEx would continue till we commercialize these new products." — Vaibhav Taneja, CFO

Assessment: With Robotaxi revenue a 2027 event and Optimus customer revenue late 2027, "till we commercialize" means operating expenses outgrow gross profit for at least four more quarters. That caps operating margin in the low single digits through mid-2027 unless automotive margins keep rising, and it is why our FY27 EPS framework of $1.95–2.25 still leaves the stock near 180x forward earnings.

10. Capacity: Battery-Gated Today, 5 Million a Year by 2030

The deck says battery availability remained the limiting factor on vehicle volumes. Management pointed to Berlin pack capacity, Texas cathode and lithium refining, and Nevada LFP cells as the supply fixes, and the vehicle engineering lead said factories ran at record rates as last year's debottlenecking investments came through. The CEO set a new long-range goal.

"We're, I think, aiming for 5 million vehicles a year by approximately 2030. And that's -- I think we've got a good shot of achieving that goal." — Elon Musk, CEO

Assessment: Five million units a year is roughly 2.8x our FY26 estimate of about 1.81 million and implies a volume CAGR near 29% from 2026 to 2030. The 2026 evidence (flat deliveries, a four-year-high backlog, battery-gated production) supports demand but says nothing about Tesla's ability to triple output, so we treat 2030 volume as an aspiration, not a model input.

11. SpaceX: Special Committees, and Silence on the Call

The merger question was settled before the call started. On SpaceX's own call a day earlier, the CEO said both boards had formed special committees of independent directors at shareholders' request, that he is excluded as a conflicted party, that he thinks a combination would be a good idea but is not a certainty, that Tesla would need a vote of a majority of all outstanding shares, and that the committees have not agreed on terms. Tesla's IR opened analyst Q&A by ruling out merger questions.

"With regard to a potential merger with SpaceX, I think I said everything that should be said yesterday, tap-dancing through a minefield there." — Elon Musk, CEO

The financial entanglement continued in the quarter: Tesla booked a $142M pre-tax markdown on its SpaceX stake, now marked to a public share price, and the CEO credited "the Tesla and SpaceX teams" for the Roadster demo.

Assessment: The governance bear point has moved from a deflected question to a formal process, which is better procedure and higher stakes. A combination's terms (the exchange ratio, any premium, and how Terafab costs are allocated) will matter more to a Tesla shareholder's 12-month return than anything in this print. We stay at materializing and treat an announced deal as an immediate rating review.

Guidance & Outlook

July Commitments Graded

New Dated Commitments From This Call

The outlook section of the deck remains qualitative: no volume, margin or EPS guidance, a commitment to "maximum capacity utilization," and an expectation that hardware profits will be "accompanied by an acceleration of SI, software and fleet-based profits." The only hard number on the call was the capex revision. The CEO framed the company as leaving a flat period: "after many years of sort of grind and things being fairly flat, I think we're now headed for a lot of fun and significant growth."

Implied Q4: Year-to-date deliveries are 1.32M. Production is battery-gated near 465,000 a quarter, and 11 days of supply leaves only about 70,000 units of inventory to draw. If production edges up to 470,000–490,000 and inventory falls modestly, Q4 deliveries land at 470,000–505,000 and FY26 at 1.79–1.83M, down from our July range of 1.90–1.98M. The ~$20B capex guide implies Q4 capex near $7.0B. If operating cash flow returns to a $4–5B quarterly pace without the Q3 working-capital assist, Q4 free cash flow is roughly ($3B) to ($2B), and FY26 lands between about ($1.1B) and breakeven.

Street at: FY26 revenue consensus of roughly $105–107B needs about $26.6–28.3B in Q4, at most a repeat of Q3. We see that as achievable on vehicles; energy is the swing line given the Q3 deployment miss and Megafactory Texas ramp timing.

Guidance style: Unchanged in form (no numeric operating guidance), changed in content. Management for the first time put quarters on Robotaxi revenue materiality and outside Optimus deliveries, and both sit three to five quarters out. The capex revision is the first time this year that a Tesla financial guide moved in the shareholder's favor.

Analyst Q&A Highlights

Whether Rival Driver-Assistance Systems Could Deflate FSD Pricing

The opening analyst question asked whether FSD's success could invite a wave of competing advanced driver-assistance products built on merchant chip platforms, and whether that would pressure FSD pricing in the US and in China, where autonomy is already table stakes. Management answered with the difficulty of the long tail rather than with pricing data, and on China pointed to FSD operating on "certain cars" and to the release of Tesla's safety reports to regulators.

Q: "I guess in a weird way, could you guys be a victim of your own success in seeing other carmakers noting the success and trying to aggressively come up with similar L2+ type products, maybe partnering with NVIDIA or Qualcomm? And if that happens, could that be deflationary for FSD pricing?"
— Tom Narayan, RBC Capital Markets

A: "Yes, it's easy to get to the first 90% and make like a cool-looking demo. But when you get in the long tail of this problem, it's so long that people don't realize anyone with their sane mind should not attempt to solve this problem."
— Ashok Elluswamy, VP of AI

Assessment: The technical argument is credible for unsupervised autonomy, but the question was about supervised products and price, and management did not engage on price. FSD's US attach rate shows no competitive pressure yet. China is where the risk is live, and the answer there was a regulatory hope, not a timeline.

When Robotaxi Miles Turn Into Revenue

A question translated the double-digit weekly mileage growth into a year-end mileage figure and asked how many miles the service needs to generate gross profit. Management did not give a breakeven, but it gave the first dated revenue timeline for the business.

Q: "If you sustain, let's say, 15% week-over-week growth in these miles, you'll probably cross 10 million miles by the end of the year. How many of those -- how many miles would you need to generate gross profit?"
— Emmanuel Rosner, Wolfe Research

A: "Yes. I wouldn't say it's not like going to move the needle this quarter. But if you say like maybe -- I think it starts to become noticeable probably in Q2 and then probably a big deal by Q3 or Q4 of next year."
— Elon Musk, CEO

Assessment: The breakeven question went unanswered, and the revenue timeline is wrapped in "probably" twice. It is still the most concrete monetization statement Tesla has made on Robotaxi, and it sets a falsifiable test for the Q2 2027 print. It also confirms the business contributes nothing measurable to FY26.

Whether the Cybercab Line Can Keep Pace With Mileage Growth

The follow-up asked how the Cybercab line ramps from its listed capacity of more than 125,000 units a year. Management described the vehicle as one of the easiest products to scale in manufacturing and said the line is deliberately running below its capability, building only what the Robotaxi service can deploy.

Q: "And then if I think about it in terms of fleet size, so I think your Cybercab line capacity is listed at more than 125,000 units a year. Production just recently started. How do you expect the ramp-up curve to be on the Cybercab line?"
— Emmanuel Rosner, Wolfe Research

A: "I think the listed capacity is more of a rate at which we're capable of running today, but the actual capacity of that vehicle line is significantly higher. Right now, we're building at the rate that the Robotaxi service needs the vehicles."
— Lars Moravy, VP of Vehicle Engineering

Assessment: Manufacturing is not the bottleneck; deployment is, through software capability and regulatory approval. That reframes the Cybercab capacity number as irrelevant to 2027 revenue, and makes the NHTSA certification audit and state approvals the variables to watch.

Software Versus Regulators as the Robotaxi Constraint

An analyst asked whether, after the Austin inflection, growth is now limited mainly by regulators. The AI lead said the limiting factor is still the software, citing the staged extension of service hours and a zero-incident operating principle, and said regulation may become the constraint in the first half of 2027.

Q: "So would you generally say that the bottlenecks at this point to growth are more regulatory driven as you think about expanding into new geographies and that you've kind of gotten through most of the hurdles on the software side."
— Andrew Percoco, Morgan Stanley

A: "So we are slowly expanding the availability of Robotaxi, and that is driven by the capability of the AI software. ... So it's not purely regulatory or something like that. It is technology driven, but we do see that pretty quickly become regulatory perhaps in Q1, Q2 time frames."
— Ashok Elluswamy, VP of AI

Assessment: This is a more cautious answer than the headline mileage chart suggests: the system still lacks night, all-hours and highway capability in commercial service. It also lines up the risks in sequence: software through year-end, then possibly regulators in 1H27, which is when the Robotaxi revenue ramp is supposed to begin.

How Much Terafab Will Cost, and Who Pays

An analyst asked for the capital required to reach Terafab's volume target and the milestones along the way. The CEO gave the first order-of-magnitude figures and acknowledged the cost split with SpaceX is unresolved. The CFO added that most equipment orders for the research fab are placed.

Q: "Do you have a sense on how much capital needs to be invested to get there? And also what kind of milestones we should be on the lookout for on the way there?"
— Edison Yu, Deutsche Bank

A: "It's certainly in the tens of billions for Terafab. Although we've got to figure out like what the sort of relative split is between SpaceX and Tesla in terms of who pays for what at Terafab. But it's not going to be less than tens of billions in capital equipment up through early 2028. And then it will be -- it will probably be north of $100 billion long term."
— Elon Musk, CEO

Assessment: A probable long-term cost above $100 billion is the largest capital figure management has attached to any program, and the allocation between two companies with a common controlling shareholder is undecided. This answer, not the FY26 capex cut, is the real capex signal for 2027 and 2028.

The Size of FSD's Recurring Revenue

An analyst laid out a lifetime-value argument (roughly $7,000 of gross profit per car at sale, plus over $1,000 a year of FSD) and asked whether FSD ARR is approaching $1B. The CFO confirmed the order of magnitude and promised broader FSD metrics, including churn, as the business scales.

Q: "And are you at some point going to disclose for us like the annual recurring revenues that FSD represents? And I did my own math, and I'm thinking it's probably now approaching $1 billion of ARR. Is that the right order of magnitude?"
— Pierre Ferragu, New Street Research

A: "Yes. I think the numbers which you shared are correct. And as it scales further, we will start sharing more metrics around FSD, not just the revenue, but also other key metrics around it."
— Vaibhav Taneja, CFO

Assessment: The confirmation anchors FSD subscription revenue near $1B a year, well below what a 1.74M customer headline suggests to a casual reader, and below our own July estimate. Retention was the follow-up, and it was deferred, so the quality of the subscription base is still unknown at the moment attach rates are peaking.

Where Capex Goes After the FY26 Cut

The last analyst question asked how Tesla is securing semiconductor equipment capacity ahead of a larger ramp and how that flows through capex. The CEO's answer is the clearest statement that the FY26 reduction is not a smaller program.

Q: "But are there some more aggressive things that you can do to start getting ready for the larger ramp now? And how should we think about that flowing through the CapEx budget?"
— Colin Rusch, Oppenheimer

A: "It's not a massive impact right now. I think CapEx is actually -- it's actually ending up being a little lower than we anticipated. But it will start to become more meaningful next year and then noticeable like in the tens of billions in '28."
— Elon Musk, CEO

Assessment: The FY26 number came down partly because semiconductor spending has not started in earnest. When it does, the CEO expects it to add tens of billions in 2028 on its own. Investors who read the ~$20B guide as a new run-rate are reading the wrong year.

What They're NOT Saying

1. Robotaxi revenue, fleet size and cost per mile, fourth consecutive quarter: The Cybercab fleet grew tenfold from an undisclosed base, cumulative miles are given only as "nearly two million," and the only economics offered were that the $0.30 cost goal has not been reached. Services & Other is not broken out.
2. The size of the energy one-timers: Tariff refunds and the vendor warranty reimbursement drove most of the 33.1% energy margin, and neither was quantified. Investors cannot compute a clean energy margin for Q3.
3. Whether tariff refunds touched automotive cost: Tesla imports components and received tariff refunds it is reinvesting in US capacity. Management attributed the automotive margin gain to mix alone and did not say whether any refund was recognized in automotive cost of sales.
4. The working-capital lift to cash flow: Payables and accruals rose $4.0B and deferred revenue $1.3B, stretching days payable to 66. Neither was discussed, and positive free cash flow was presented as cash management.
5. Which capex projects slowed: The FY26 guide fell by about $5B "as a result of managing our capital deployment efficiently," with no project detail and no FY27 capex guide, even as management described Terafab adding tens of billions by 2028.
6. The NHTSA Cybercab audit: A shareholder question explicitly cited NHTSA's review of the Cybercab; the answer covered software scaling and Dallas and did not address certification.
7. Phoenix, Las Vegas and the city count: July described both as in preparation. Neither was mentioned, and the AI lead put the current count at six cities.
8. HW3: After featuring on the last two calls, the Hardware 3 upgrade path did not come up at all, despite being a standing shareholder concern and a litigation exposure.
9. Balance-sheet movements: Other non-current assets rose $3.1B in the quarter and current deferred revenue rose $1.2B. Management did not say whether these reflect equipment deposits, customer prepayments for Megapacks, or something else.
10. A China FSD date: April's Q3 2026 target has passed. The answer was that FSD operates "for certain cars" and that regulators should find the safety data hard to refute.

Market Reaction

• Pre-print setup: TSLA closed at $386.47 on October 21, down 14.1% YTD against an S&P 500 up 13.3%, up 3.0% over the trailing 30 days and down 12.7% over the trailing twelve months. The 52-week closing range is $298.32–$489.88. The stock rallied about 9% over the three sessions into the print (from $354.72 on October 16), including a 3.4% gain on October 21, the session after SpaceX's call disclosed the special committees.
• After-hours move: Shares rose about 2% immediately after the 3:00 PM CT release on the EPS beat, the margin rebound and positive free cash flow, then reversed during the call to trade roughly 2–4% lower, in the $370s. The turn coincided with the capex cut to about $20B and the Optimus and Robotaxi revenue timelines.
• At publication (Thursday session, live): TSLA opened at $374.03, 3.2% below the pre-print close, then recovered through the morning to trade around $387, slightly above the pre-print close, on a $373.71–$388.86 intraday range. The session had not closed at publication.

Positioning Ran Ahead of the Print

The stock entered the print with a rally behind it, and its final leg came on the merger-committee disclosure rather than on any Q3 expectation. That left the print with a narrow job: confirm the margin story without disturbing the long-term build-out narrative. It did the first and complicated the second.

A Capex Cut Reads Differently for an AI Stock

For most industrial companies, a $5B capex cut with positive free cash flow would be welcomed. For a stock valued on the future scale of autonomy, robotics and AI compute, the market read it partly as a slower build-out, and the timelines given on the same call reinforced that reading: Optimus outside deliveries at the end of 2027, Robotaxi revenue "a big deal" in late 2027, EU-wide FSD by Q1. The after-hours reversal is consistent with investors reassigning value from 2026–27 to 2028, not with a disappointment in the quarter's numbers, which beat.

The Regular Session Looked Through the Timelines

By mid-morning Thursday the overnight loss was gone. Regular-session buyers appear to have weighted what the quarter fixed over what the call postponed: the problems the stock was penalized for in July (margins, cash burn and debt) all improved, and the dates that unsettled the after-hours market are, for the most part, first-time specifics rather than slips. The round trip from +2% to about -4% and back to flat is the signature of a stock without a consensus anchor. At ~225x trailing earnings, its price rests on 2027–28 businesses that a quarterly print can re-date but cannot yet confirm, and the merger process now adds a second, larger driver that the print does not address.

Street Perspective

Debate: Is the Capex Cut Discipline or Delay?

Bull view: Tesla acts as its own general contractor and found cheaper ways to build; funding FY26 from operations with $30B of undrawn capacity removes the dilution and leverage fears that hit the stock in July. Positive free cash flow during an investment cycle is a sign of a self-funding model.

Bear view: The cut arrived without project detail on a call where management described tens of billions of Terafab spending in 2028 and capex rising for years. Q3 free cash flow leaned on $2.6B of working capital. This is a deferral that flatters 2026 and loads 2027–28.

Our take: Both are partly right, and the bear has the better forward evidence. We credit real efficiency in FY26, but we model FY27 capex above FY26 and treat the multi-year cash question as postponed. The positive is that Tesla can now carry it without drawing debt in 2026, which removes our nearest Underperform trigger.

Debate: Is 18.1% the New Clean Auto Margin or a Mix Spike?

Bull view: Unit costs fell about $1,140 per vehicle sequentially with no one-time items, FSD attach above 75% in North America lets Tesla trim subvention, and a four-year-high backlog lets it allocate cars to the best-margin regions. Credits are now too small to matter, so the margin base is fully organic.

Bear view: The driver named was regional mix, which is the least durable source of margin, and Tesla is offering incentives in Europe while ASPs keep falling. Commodity and rate headwinds persist. One quarter does not reverse the evidence of the prior two.

Our take: We side with the bull on direction and the bear on proof. Our July standard was two clean quarters above 17%, and we are holding to it. We model FY26 automotive gross margin ex-credits at 17.4–17.9% and the Q4 print at 17–18%, so a second print at or above 17% is our base case, not a stretch.

Debate: Does Exponential Mileage Justify the Robotaxi Valuation Before Revenue?

Bull view: Eightfold mileage growth in a quarter, paid rides in purpose-built vehicles, a manufacturing line running below capability, and a cost structure no human-driven rideshare can match. Management now dates meaningful revenue to 2027, and the market should look through a few quarters of invisibility.

Bear view: Four quarters into commercial operation there is still no revenue line, no fleet count and no unit economics. The city count has not grown since July, a federal audit is examining how the Cybercab was certified, and management says regulators may become the constraint in early 2027, precisely when revenue is supposed to ramp.

Our take: The operating progress is real and faster than we modeled. But the valuation question is about economics, and the earliest dated test is Q2 2027. Until then the market is paying for a curve without a unit cost, which is why we stay at Hold rather than move up on the operating evidence.

Debate: Does a SpaceX Combination Help or Hurt Tesla Shareholders?

Bull view: A combination would consolidate Terafab, chips, AI software, robotics and connectivity under one balance sheet, remove the related-party friction, and give Tesla holders exposure to a faster-growing space and AI franchise. Special committees and a majority-of-outstanding vote protect minority holders.

Bear view: SpaceX is the larger entity and the likely acquirer in an all-stock deal, so Tesla holders could end up a minority in a company valued on different assumptions, with the controlling shareholder's voting power rising. The Terafab cost split is being decided while merger terms are negotiated, which compounds the conflict.

Our take: The committee structure is the right process, and the high vote threshold is a real protection. The outcome is binary for a 12-month return and cannot be modeled from public information. It is the main reason our fair-value range is wide and our rating is Hold.

Model Update Needed

Valuation impact: We raise our 12-month fair value range to $365–435 from the $350–420 range we set in July. Trailing-four-quarter non-GAAP EPS is $1.72, so the $386.47 pre-print close is ~225x trailing, and our FY27 non-GAAP EPS framework of roughly $1.95–2.25 puts the range at about 175–210x the FY27 midpoint of $2.10. No earnings multiple supports that range on its own; it prices the Robotaxi, FSD, Optimus and energy-for-AI options, and we move it only by what the quarter changed. On the positive side, FY26 non-GAAP EPS rises about 15% at the midpoint, FY26 free cash flow improves by roughly $5–6B, and the debt-funding risk recedes. On the negative side, Robotaxi and Optimus monetization moved one to two quarters to the right, FSD's recurring revenue base is about half what we carried, and Terafab now carries a stated cost of tens of billions with an undecided SpaceX split. Net, the midpoint rises about 4%, to $400.

The midpoint implies about 3.5% upside from the $386.47 pre-print close, and a similar figure from Thursday's mid-morning level near $387, with no dividend, which is a market-like 12-month return and consistent with Hold. Our downside case (merger talks fail or terms disappoint Tesla holders, Robotaxi expansion stalls at regulators in 1H27, auto margin slips back below 17%) points toward the $298.32 low of the 52-week closing range, about 23% below the pre-print close. Our upside case (a second clean margin quarter, a first Robotaxi revenue disclosure, and merger terms that carry a premium for Tesla holders) supports the upper end of the range and above. With both tails live and roughly balanced, the risk/reward does not justify a directional call.

Thesis Scorecard Post-Earnings

Overall: Thesis strengthened on net. The financial half of the story, which broke in July, partly repaired: the clean auto margin rebounded above our 17% threshold, free cash flow turned positive, FY26 capex fell by about $5B and the debt facilities stay undrawn. The operating half kept delivering, with every July product commitment met. What did not improve is the timing of the revenue that justifies the multiple: Robotaxi and outside Optimus revenue are now explicitly 2027 events, the FSD recurring base is smaller than we carried, and a new regulatory risk has appeared around the Cybercab. Overlaying all of it, the SpaceX special committees turn the governance bear point into a live transaction whose terms will dominate the 12-month return.

Action: Maintain Hold, conviction unchanged at 5/10, with fair value raised to $365–435. We would upgrade to Outperform on a second consecutive quarter of automotive gross margin ex-credits above 17% together with a first Robotaxi revenue or unit-cost disclosure, or on a price reset toward the low $300s that pre-pays the 2027 timeline. We would consider Underperform if FY27 capex guidance runs well above our $24–30B range with debt drawn, if the NHTSA audit or state regulators halt Cybercab deployment, or if merger terms subordinate Tesla minority holders. An announced SpaceX transaction triggers an immediate rating review.