Tesla’s third quarter was flat on the top line and brutal on the bottom line. Revenue reached USD 28.046 billion, down 0.2 per cent year on year. Net profit attributable to ordinary shareholders fell 43 per cent to USD 787 million, or about RMB 5.25 billion.
The detail underneath explains the direction of travel. Automotive revenue slipped 3 per cent to USD 20.56 billion on 486,500 deliveries, down 2 per cent. Energy generation and storage revenue fell 16 per cent to USD 2.88 billion. Services and other revenue, the smallest line a year ago, grew 33 per cent to USD 4.61 billion.
Margins up, profit down, capex through the roof
Gross profit rose 9 per cent to USD 5.494 billion and the GAAP gross margin climbed to 19.6 per cent from 18 per cent a year earlier. Operating cash flow held at USD 6.263 billion.
Then the investment line arrived. Capital expenditure more than doubled to USD 4.732 billion, up 110 per cent, which pulled free cash flow down 62 per cent to USD 1.531 billion. Cash, equivalents and short-term investments stood at USD 45.447 billion at the end of September, up 9 per cent.
Production reached 464,400 vehicles, up 4 per cent, with Model 3 and Model Y accounting for 457,400 of them. Other models, meaning Model S, Model X and Cybertruck, collapsed: 7,004 built, down 40 per cent, and 8,295 delivered, down 48 per cent.
There were two genuine upside markers. The passenger vehicle order backlog hit a four-year high at quarter end, and storage orders and backlog both set records. Active FSD subscriptions rose 67 per cent to 1.74 million. FSD is now live in nine EU markets with more than 100,000 users, who have covered over 160 million kilometres since first approval. Storage deployments reached 13.7 GWh, up 10 per cent, and Tesla added more than 2,100 superchargers worldwide.
Seven answers, two deadlines
The call was a scheduling exercise. On China, executives said Tesla is working hard to win FSD approval and that some vehicles already run it, while arguing that solving the long tail of autonomy to fleet-level safety is the part rivals will not match for years.
Unsupervised FSD was guided to some customers around the middle of next year, with early features reaching first customers within roughly twelve months. Cybercab sales to outside buyers could begin by the end of this year, though early units stay in Tesla’s own fleet. Robotaxi gross margin is improving but below target, with the service running 9am to 9pm in Austin and expanding to more cities.
Musk put a unit economics number on autonomy: USD 0.37 per mile at scale. He added that if unsupervised miles keep compounding at roughly 15 per cent a week, Robotaxi becomes a material part of revenue and profit in 2027, and more important around the third quarter of next year.
The humanoid line carried the sharpest guidance. Optimus is in early production at Fremont, and Musk described the ramp as slow and bumpy because the product is new. He conceded that the AI, the electromechanical design, reliability and service life still need work, that a robot cannot break after a month, and that roughly another year is needed. External deliveries should start around the end of 2027, reaching 20,000 units a week by the end of that year. A video of non-preprogrammed, AI-driven behaviour is expected within about a month, and the AI5 chip should land in Optimus around April.
TeraFab, the Austin fab, starts tool installation in the first half of next year with first chip testing around May or June. Musk wants something useful by 2028 and real production around May 2028, with capital spending of at least tens of billions of dollars by early 2028 and more than USD 100 billion over the long run.
What the capex line is actually buying
Tesla is funding three parallel futures out of one cash pile: a robotaxi network, a humanoid robot, and its own silicon. Any one of them can justify the multiple. Together they raise the cost of being late, because the spending is fixed while the revenue is not.
The battery constraint is the boring part that still decides the quarter. Tesla said cell supply remains the main limit on vehicle ramp, and it is expanding pack output in Berlin, cathode and lithium refining in Texas, and LFP cell capacity in Nevada. The first Megapacks built with Tesla-made LFP cells have shipped.
Gross margin recovery plus USD 45 billion of liquidity is what lets Musk keep answering questions with dates instead of numbers. 2027 is now the year every one of those dates matures at once.
Editor’s note: This is an adapted translation of the original Chedongxi report. It has been trimmed and restructured for readability for an international business audience. The full original (in Chinese) is at https://chedongxi.com/p/378738.html.