Tesla reported on October 21, 2026, that Optimus production had begun on a high-volume line at its Fremont, California, factory. The program is at an early ramp: Tesla called the first robots production-quality, while Elon Musk described 20,000 robots per week as a future production rate, not the current stage.
What Tesla has started at Fremont
Tesla’s Q3 2026 update classified its California Optimus program as “Early Ramp” and its Texas site as “Construction.” The different stages are clear in the company’s site update:
| Site | Optimus program status |
| Fremont, California | Early Ramp |
| Texas | Construction |
Tesla said the first robots coming off the Fremont line were production-quality and that the manufacturing process was suitable for a commercial product. The company also said it would keep refining the hardware and manufacturing process. The production start is a manufacturing milestone, not a measure of completed commercial deliveries.
Tesla reported Q3 2026 GAAP net income attributable to common stockholders of $787 million, compared with $1.373 billion in Q3 2025. Revenue was $28.046 billion, versus $28.095 billion a year earlier.
The production targets are still future milestones
Musk said Tesla would need at least a year, and possibly longer, to reach an annualized rate of 20,000 Optimus robots per week. He also described the Fremont line as designed for about one million units a year and said most of its equipment had arrived. Those are descriptions of intended line capacity and a future rate, not reported output.
Musk forecast deliveries to outside companies around the end of 2027, conditional on further work on design, AI, reliability and durability. He said early customers would likely be industrial users, particularly for dangerous or repetitive tasks.
Building robots does not settle how well they can work
Tesla’s stated production readiness sits alongside development work on the robot itself. Musk said AI for open-ended tasks still needed work; Tesla executive Konstantinos Laskaris identified the hand as a challenging hardware area. NeoTeo previously covered reported hand and supplier snags affecting Optimus.
Ken Goldberg, a robotics professor at the University of California, Berkeley, offered a broader field-level caution: manufacturing robot hardware does not by itself solve the software challenges involved in useful performance. Goldberg was discussing robotics generally, not reporting a test of Optimus. His assessment points to a practical distinction in Tesla’s announcement: starting production describes progress on manufacturing, while dependable performance on varied tasks remains part of the development work.
