The factory floor where Tesla used to build the Model S now builds robots. That’s the headline from Tesla’s third-quarter update, out on October 21. Tesla Optimus production has officially started in Fremont, California. But if you’re hoping to order one for your kitchen, hold that thought.
Here’s what actually happened, what Tesla left out, and why it matters even if you’ll never buy a humanoid robot.
So what actually changed in Fremont?
In its Q3 2026 update, Tesla put it plainly: “Optimus production began in Fremont on our high-volume lines.” It also called these “the first production-quality bots that we have manufactured.” In other words, these aren’t hand-assembled prototypes for a stage demo anymore.
The space itself tells a story. According to TechCrunch, the robots come off a part of the factory that used to make electric vehicles. That lines up with what Tesla said earlier this year. Back in January, Musk said the Model S and Model X would end so the floor space could go to Optimus. Then, in its Q2 update, Tesla said it had shut down those car lines to build its first Optimus production lines.
Tesla also made a bolder claim. “We have proven that the current manufacturing process for Optimus is suitable for a commercial product,” the update says. Still, it adds that the company will keep tweaking both the hardware and the process.
Why Tesla Optimus production is a bigger deal than it sounds
Building one impressive robot is hard. Building thousands that all work the same way is a different problem entirely, and I’d argue it’s the harder one. That’s why a carmaker moving a humanoid onto real production lines gets attention.
The targets are huge, too. On the earnings call, Musk said Tesla aims to make 20,000 Optimus units per week by the end of 2027. Earlier this year, he also set a goal of one million units a year at the California factory.
And Fremont isn’t the only site. Tesla broke ground on a dedicated Optimus factory at its Texas campus in late March, as Tesla North reports. By September, drone footage showed its steel frame nearing completion. For now, Tesla’s own capacity table lists California as “Early Ramp” and Texas as “Construction.”
So where are the first robots going? Tesla said in Q2 that early builds are going toward collecting training data, a program it calls the “Optimus Academy.” Put simply, the robots first need to learn by doing before they can do much of anything useful.
What Tesla didn’t tell us
This is where I’d pump the brakes a little. The update skipped a lot of the numbers you’d want. There’s no production rate and no unit count. There’s also no price, and no date for selling Optimus to outside customers. Tesla didn’t even say which version of the robot it’s building.
Musk himself sounded careful. He called this “early production” and warned the ramp will be “somewhat bumpy because it is such a radically new product.” Back in April, he’d already said the first ramp would be “quite slow.”
He also listed the hard parts that remain. The robot has to be “safe and reliable and has long duration,” he said, so it doesn’t “wear out after a month.” On the AI side, there’s “still quite a lot of work to do” before Optimus is truly useful for open-ended tasks. Honestly, that last bit is the real story for physical AI. Hands, motors and batteries are tough, but teaching a robot to handle a messy, unpredictable room is tougher.
The awkward money part
The robot news landed on the same day as a rough earnings report. Tesla’s net income fell 43% to $787 million, down from $1.37 billion a year earlier. Revenue came in at $28.04 billion, roughly flat against $28.09 billion in the same quarter last year.
Part of the squeeze, TechCrunch notes, came from much higher spending on AI, Optimus and self-driving. So Tesla is paying real money now for a robot business that doesn’t sell anything yet.
It’s also fair to remember Tesla’s track record on timelines. In April, Musk said unsupervised Full Self-Driving would “probably” arrive in the fourth quarter of this year. Now Ashok Elluswamy, Tesla’s VP of AI software, says it’s “somewhere around Q2 next year” before the first customer cars could join a robotaxi fleet. Because of that history, I’d treat 20,000 robots a week by 2027 as a goal, not a forecast.
Okay, but should you care about Tesla Optimus production?
Yes, though maybe not for the reason Tesla wants. Even if the ramp is slow, Tesla Optimus production marks a shift from robot demos to robot factories. Once humanoids get built on real lines, costs, parts and supply chains start to matter as much as flashy videos.
That opens doors well beyond California. The parts that make humanoids move, like the actuators that turn commands into motion, need engineers and suppliers everywhere. Meanwhile, the software that teaches robots to behave is exactly the kind of work small, sharp teams in places like Armenia can compete on.
Tesla also isn’t alone. We’ve looked at how Agility’s Digit 5 is chasing the same prize, and what humanoid robot buyers really pay today. So here’s the question worth watching: when Tesla finally puts a price on Optimus, will it be low enough to change who can afford a robot coworker? That number will tell us more than any launch event.