Tesla Q3 was framed as the setup for a much bigger merger-and-autonomy story: FSD, Cybercab, Optimus and energy drove the quarter, while a Tesla/SpaceX deal was treated as urgent because delay could change the exchange math, voting outcome, and tax timing. The deck showed 1.5B of free cash flow, >45B of cash and investments, 33.1% energy gross margin, >75% FSD attach in North America at delivery, and Cybercab scaling from Austin toward Dallas and other cities. The merger debate turned on committee conflicts, a December 3 vote stress test, and how much premium or share ratio is needed to get to 50%+1 with brokers/non-votes against. The later segments pushed the long-run thesis hard: Tesla is building toward robotaxi scale, AI5/Optimus integration, in-house chips, and a research fab, while SpaceX/compute/solar and power constraints shape the capital structure and timeline.
Tesla’s Q3 deck showed 1.5B in free cash flow, >45B in cash and investments, and capex guided to ~20B this year, funded mainly from operations with up to 30B of borrowing capacity left unused in 2026.
Automotive margins excluding credits moved from 16.3% to 18.1%; energy gross margin rose from 20.4% to 33.1%; services and other hit a record 15.9%.
Tesla ended Q3 with its highest backlog in four years, two straight quarters above 480k vehicle deliveries, and inventory at 11 days, described as the biggest car backlog in over four years.
North America FSD attach at delivery was over 75%, with more than 1.7M paid FSD customers globally and over 50% of paid customers on subscriptions for the first time.
Cybercab launched in Austin, cumulative unsupervised miles reached nearly 2M by quarter-end and over 3M later, and expansion to Dallas next month was explicitly laid out.
Tesla said Cybercab rides in Austin expanded service hours from 6:00 a.m.–9:00 p.m. to 10 p.m., 11 p.m., and soon 2 a.m., with a path to 24/7 operation by year-end across more weather, highways, and higher speeds.
Model Y set delivery records in several markets, preliminary data pointed to number one BEV in Europe and number one passenger car in South Korea, and new record markets included Australia, Japan, and Taiwan.
Optimus production began on Fremont high-volume lines, Tesla expects a first test chip from the Austin research fab around May or June next year, and AI5 integration into Optimus is planned for about April next year.
Tesla said the Austin research fab will start commissioning machinery in 1H next year, with useful chip volume targeted around Q3 2028 and real production late 2028.
The Tesla/SpaceX merger was described as needing speed because delays could trigger lawsuits, rerate SpaceX, and shrink the Tesla-side stake from the eventual exchange ratio.
The committee math was framed around conflicted directors and special committees: Elon, Kimble, Ira Aaron Price, and Antonio Garcia were treated as conflicted; Robin Denholm, Kathleen Wilson-Thompson, and Jack Hartung were treated as not conflicted, while Joe Gabia, James Murdoch, and JB Stroble had question marks.
December 3 was framed as a shareholder-vote stress test, with a letter for retail holders to sign in the next 1–3 days, before the larger vote next year.
Voting math was the core hurdle: 3.95B shares outstanding, Elon at 1.1B shares, about 15% of votes historically not cast and ~10% broker non-votes counted as no, leaving a 25% no-vote wall before getting to 50%+1.
Alexandra’s timing case pushed proposal publication into Q2 2027 and closing into Q4 2027 or later, tied to the mid-January 2028 tax bill from the 2018 compensation package, possible SpaceX S&P 500 inclusion, and aligning merger timing with tax and index dates.
The proposed Tesla/SpaceX exchange math ranged from a 4.2-for-1 share ratio to premium floors around 15%–30%, with examples that 30% premium implies ~$493, 520 is about 37%, 550 about 45%, and a 3.5-share SpaceX ratio was described as very lucky.
The call-linked deck said Tesla will get to 400MW of AI training by year-end, with Cortex 1/2, megapods, car-based compute, and 4M AI4 chips from cars framed as about 1GW of compute.
SpaceX was framed as potentially bigger than Azure and AWS, with higher margins, spot pricing, and fewer clients; Tesla’s Cybercab and FSD subscription ARR were contrasted with compute price slumps and client concentration.
The merger and factory narrative was tied to power and materials: Tesla’s 4680 ramp supports Cybercab and Semi, AI5 comes from Samsung and TSMC, and Tesla is building a research fab, an Optimus factory, and more in-house compute and chip capacity.