Tesla Optimus production has started. On the same earnings call, Musk said getting to 20,000 robots a week takes “at least a year, maybe a little longer,” and that robots reach companies outside Tesla around the end of next year. If you own the stock, or run a warehouse that might one day buy a humanoid, that second sentence is the news.
The quarter itself was ordinary. Revenue of $28.046 billion fell under the $28.34 billion analysts expected while non-GAAP EPS of $0.48 beat the $0.45 estimate, per Teslarati. Against a year earlier, Tesla’s own table shows revenue flat, $28.046 billion against $28.095 billion.
So the robots and the robotaxis carry the story, and I wanted to see what the story actually says.
Tesla Optimus Production Has Started. The Hands Are Not Finished.
Tesla’s wording, from its Q3 update:
“Optimus production began in Fremont on our high-volume lines. The bots coming off this line are the first production-quality bots that we have manufactured. While we will continue to iterate on the hardware and manufacturing process, we have proven that the current manufacturing process for Optimus is suitable for a commercial product.”
Same day, same company, different mood. On the call, Musk called it “early production” and described a growth curve with a shallow slope that is “somewhat bumpy.” His reason: “There’s like 10,000 things that have to go right for production of the robot,” and the ramp moves at the pace of the slowest one.
Then, one answer stopped me. An analyst asked about ramping before the design is final, and Musk said Tesla is “building robots that do not have perfected hands.” Konstantinos Laskaris, who took the hardware question for Tesla, called the hand one of the most challenging hardware areas and said some bespoke gears that carry a lot of load for reliability are still in development.
So one day gives us the first production-quality bots and robots without perfected hands. Both can be true. I don’t regard it as contradictory.
The 20,000-a-Week Target Is a 2027 Story.
Twenty thousand a week works out to about a million robots a year, and a million a year is the line Musk says is mostly installed already: “the vast majority of a 1 million unit per line production line landed in Fremont right now.” By his own account, floor space and machines are not what hold the ramp back. He gave the reason in one sentence:
“There are still parts of the design that are changing.”
I searched the full reference for the two numbers a buyer or an investor would want: the current weekly output and the price. The call has neither. Tesla’s finance chief, Vaibhav Taneja, said that Optimus cost is tracking well, but that it is “a little too early to say that we have got there” because the design is still iterating. On October 9, I argued in my article “Tesla Scrapped Two Car Models for Optimus. The $20,000 price Won’t hold.” A first-year price near $80,000, about €71,000, looks more realistic than $20,000.
Nothing said on this call moves that either way.
What the call gives is a sequence. Musk wants the AI5 chip inside Optimus around April, “if things go according to plan,” which is the chip I covered in “Tesla Doesn’t Need AI5 in Its Cars. It Needs It in Its Robots.”
He expects the AI, the electromechanical design, and the robot’s usable life to be solved around the end of next year, and that is the point he gave for delivering Optimus to companies outside Tesla. The first customers are industrial, doing dangerous, repetitive work around the clock. Homes come later, “as we scale production.”
He also floated an Optimus video “in a month, or so” that shows the robot doing things that are not pre-programmed.
Tesla Robotaxi: Six Cities, Two-Thirds of the Miles in One
The second half of the call was louder, so it gets less benefit of the doubt from me. Musk said the Cybercab fleet in Austin has grown by a factor of 10 since launch about a month ago. A multiple tells you direction, not size. The one count Tesla has given came on October 10, when its Robotaxi account said it had 300+ Cybercabs serving Austin, an 8x increase in 30 days.
Teslarati reported the same week that Texas registrations had gone from 169 to 319, and that a community tracker counting sightings and state records listed 105 Cybercabs in unsupervised service that Saturday, adding that tracker counts can lag.
Elluswamy said Tesla now does six cities, and Tesla’s robotaxi page lists Austin, Dallas, Houston, Miami, Orlando and Tampa. He framed it as breadth across cities and depth in Austin. The miles show how lopsided the depth is. Through October 18, the fleet had driven 3.1 million autonomous miles, and the per-city counts from Tesla’s new safety page put 2.12 million of them in Austin. That is 68%. Dallas already shows 608,130 miles, so what Elluswamy promised for next month is Cybercab rides there, not robotaxis.
The growth is real. Tesla’s update counted nearly two million cumulative unsupervised miles at the end of September, and the safety page counts 3.1 million autonomous miles by October 18. Different labels, same company, and a steep line either way.
The Cybercab Promise Has a Federal Deadline
Asked when customers can buy a Cybercab, Lars Moravy said Tesla is “trying to get Cybercabs to customers by the end of this year,” while everyone built so far goes to Tesla’s own fleet. A few minutes later, Musk said, “The vast majority of the Cybercab stuff will be essentially Tesla-owned and transport as a service.” In a separate answer about unsupervised FSD in customer cars, Elluswamy said, “Somewhere around Q2 next year,” Tesla can turn on some of the first customer vehicles as part of the Robotaxi fleet. Three answers, three original pictures of the customer.
The car also has a regulator attached. NHTSA’s special order has 21 requests asking Tesla to list every federal safety standard it certified the Cybercab to and explain how it determined compliance with each, including the one that says service brakes must be activated by a foot control. The Cybercab has no pedals. Tesla was given until November 20 to answer. NHTSA gave it until October 30.
I don’t know how that review ends, and the call didn’t say. The call named the dependency. Taneja said the key thing is “all the regulatory approvals where we can start deploying the Cybercabs as fast as we can produce them.” Moravy added that the line can run faster than its listed rate and that Tesla is building cars only as fast as the service needs them. So the factory is not the limit on the Cybercab either.
Every Date on the Call Came With a Hedge
I went back through the call and marked every date. Musk said, “I would expect” and “probably” about the Optimus ramp. The AI5 chip comes “if things go according to plan.” Elluswamy said “we suspect” about customer vehicles joining the fleet and “hopefully” about 24/7 service and highways by year-end. Moravy said, “We’re trying.”
Lay them on one line, and the year that matters is 2027. For investors, the arithmetic is plain. Tesla guides to about $20 billion of capex this year, has lined up capacity to borrow up to $30 billion, and plans no draws this calendar year. Revenue is flat. And Taneja said that operating expense keeps growing “till we commercialize these new products,” a group that includes Optimus. Its outside-delivery date is the end of 2027.
For a buyer, there is nothing to buy yet. No price, no stated current output, a first-customer list that starts with industrial work and reaches homes at a date nobody gave. Musk also repeated that Optimus will be the biggest product on Earth, in the same answer that told investors to wait about a year.
My opinion: production started, and delivery is the thing to watch for timing. I would treat “around the end of next year” as the earliest date on the board, not the middle of the range, because every date I found came with a hedge in front of it. Tesla gave no count of Optimus robots built in the call. Until it does, “production” is a pronoun, not a number.
Is Tesla’s Flying Car the Last Great Argument for Owning One?
Tesla revealed the second-generation Roadster on October 15, 2026, on the penultimate day of the Mondial de l'Auto in Paris. I’ve been reporting on this company for a while, and while I’m not a Tesla devotee, I have to admit that what they revealed is genuinely bizarre and worth examining.
If you read earnings calls for the hedges too, follow Nov Tech and subscribe because I’ll keep counting them as the NHTSA answer, Dallas, and the next Optimus video arrive.