The task was unglamorous: carry a part, seat a screw, route a wiring harness. The setting was not. For the first time, Tesla has put its Optimus humanoid robots to work on core automotive assembly steps inside its Fremont, California factory, according to people familiar with the deployment. Dozens of units are now on the line, and the company’s chief executive, Elon Musk, told an internal meeting that the robots should spread to the Texas gigafactory and to Tesla’s Shanghai plant by the end of the year.
The deployment moves humanoid robots from laboratory demonstrations into live production. Musk has long argued that Optimus could eventually be worth more than Tesla’s entire automobile business, and the Fremont line is the first concrete test of that claim. The factory has been preparing for it: this spring Tesla wound down production of the Model S and Model X — the last of the sedans rolled off the line in early May — and began converting that assembly space into a dedicated Optimus production line. According to the company, the converted line is designed to reach capacity of one million robots a year, though Musk has cautioned that output will ramp slowly at first.
From training floor to production line
The distinction matters. Tesla had already run hundreds of Optimus units inside Fremont for data collection and training, with robots moving battery modules and loading EV packs under supervision. The new step is different: the robots are now assigned to assembly tasks that go into the vehicles themselves — parts handling, screw tightening, wire-harness organization — alongside human workers on the line.
Each Gen 3 Optimus contains roughly 10,000 unique parts, and Tesla has built new production processes for actuators, hands and structural components that have no equivalent in its automotive operations. The company describes the ramp as deliberately slow, with the robots learning by observation and by demonstration before taking on more complex steps. The internal bet is that the same factory that builds the cars can build the robots that build the cars, and then export the model elsewhere.
The expansion plan
Musk’s internal comments, reported by attendees, outline the next two sites. Giga Texas, already the company’s largest factory, is the likely second location for Optimus deployment, with Shanghai after it. The sequence matters for Tesla’s narrative: a robot that works in California, Texas and China is a product with global demand, not a lab exhibit. The company has also said a second, larger Optimus factory is under construction at Giga Texas, with volume production targeted for next year.
The commercial stakes are large. Humanoid robots have been a fascination of the robotics industry for decades and a business failure for most of it, because general-purpose manipulation in unstructured environments is exceptionally hard. Tesla’s approach is to sidestep the hardest part of the market at first: the early customers are Tesla’s own factories, where the environment is controlled and the use cases are repetitive. From there, the company argues, the same robots can move into logistics, warehousing and eventually other manufacturers’ plants — a market that industry analysts put in the hundreds of billions of dollars.
The risk
The risks are equally visible. Humanoid robots remain unproven at industrial scale; even Tesla’s own suppliers have noted that the initial ramp will be limited. The robots also carry a heavy cost structure — each unit involves dozens of actuators, advanced sensors and a battery — and Tesla has not said what a production Optimus will cost or when the program becomes self-funding. Skeptics note that Musk has predicted near-term humanoid deadlines before, and the program has slipped.
The Fremont line’s history is part of the story. The factory, acquired from a joint venture between Toyota and General Motors in 2010, built Tesla’s first Model S in 2012 and has been the company’s U.S. manufacturing anchor since. When Musk announced on the fourth-quarter earnings call that the S and X would be phased out, the decision to convert the space into a robot line rather than retool it for another vehicle was a clear signal of where the company’s priorities sit. Tesla’s factory page now lists Optimus alongside the Model 3 and Model Y as core products.
Tesla’s own robots were part of the preparation. More than a thousand Gen 3 units were already operating inside Fremont in early 2026, according to the company, handling battery-module assembly, EV pack loading, cable routing and connector seating — work that generated the training data for the newer, more sensitive tasks now being introduced. That closed loop is central to the strategy: the robots being manufactured at Fremont will, in part, help build the next generation of themselves, generating operational data no competitor can easily replicate.
Analysts who follow the robotics sector are watching the ramp for what it reveals about cost. Humanoid robots are only commercially viable if they can match a human worker’s output at a comparable or better price, and Tesla has not disclosed production cost per unit. Musk has said the company’s manufacturing know-how — the same supply chains, quality systems and automation expertise used for cars — gives it an advantage in driving cost down. The Fremont line is where that claim will be tested in volume.
The Fremont deployment is a small number of robots doing work a human could do. But it is the first time a humanoid robot has been assigned to core assembly tasks in a real factory, and Tesla has put its own production floor behind the experiment. If the ramp works, Tesla will have done what the robotics industry has failed to do for decades; if it stalls, the company will have spent billions finding out. Either way, the assembly line at Fremont is now the proof point for the entire humanoid-robot sector.
