Tesla has begun installing its first dedicated Optimus humanoid robot production line at its historic Fremont factory — but the company has quietly softened its own timeline, replacing an earlier "late July or August" start with the vaguer promise of production "later this year." The shift, buried in Tesla's second-quarter shareholder update, captures the gap between the industry's soaring humanoid ambitions and the hard reality of standing up an entirely new kind of factory.
From Model S to Machines
The Fremont line exists because Tesla cleared space for it. On its fourth-quarter 2025 earnings call in late January, CEO Elon Musk announced the wind-down of the Model S and Model X, framing the decision as an "honorable discharge" for the aging programs and a pivot toward what he called an autonomous future. The last of those cars rolled off the line in early May, and the freed floor space is being converted into Tesla's first humanoid production line.
On the Q1 2026 call in April, Musk said Optimus production would begin at Fremont in late July or August — roughly four months after the sedans departed. He was blunt about the risk, warning that initial output would be "quite slow" and calling the year's production rate "literally impossible to predict" given that each robot carries some 10,000 unique parts across a brand-new line.
The Quiet Slip
By the July 22 second-quarter update, that specific window had all but vanished. Tesla confirmed the Model S and X lines were decommissioned and that first-generation Optimus lines were being installed, but production was now "anticipated later this year." As of late July, the company had not disclosed a single verified Optimus production unit — no start date, no unit count and no price. Reports noted that initial builds are earmarked for Tesla's internal Optimus Academy rather than external customers.
In other words, Tesla walked back the August target before the month even arrived. Given the program's history of moving reveal dates and ramp goals, analysts have suggested treating the direction as real and the date as soft.
Ambitions That Dwarf the Car Business
What keeps Optimus in the headlines is the sheer scale of the goal. Musk has said he wants Fremont to reach a run rate of one million units per year by late 2026, with Gigafactory Texas adding humanoid lines targeting 10 million units per year by 2027. For perspective, a factory producing 10 million humanoids annually would be larger than any car plant ever built — Tesla makes roughly 1.8 million vehicles a year worldwide.
Those numbers are aspirational, and the ramp is genuinely difficult. Building a humanoid at automotive volume means solving new problems in actuation, dexterous hands and onboard perception that have no direct precedent on a car line. A vehicle is a rigid product assembled from mature supply chains; a general-purpose robot must sense an unstructured world and manipulate it reliably, and the components that make that possible are neither as standardized nor as proven as automotive parts.
Why It Matters
Tesla is far from alone, and the contrast with rivals is instructive. Across the industry, humanoid deployments are quietly accumulating: Figure recently passed 1,000 units of its latest robot and retired an earlier model after nearly a year of work at a BMW plant, where the machines helped build tens of thousands of vehicles. Chinese players are moving fast too — BYD just unveiled its first humanoid, and Xpeng is pushing a next-generation robot toward mass production. Automotive suppliers, sitting on deep expertise in perception, computing and actuation, are pivoting that know-how into embodied intelligence.
- The convergence is real. Carmakers and robot makers increasingly draw on the same core technology stack.
- Execution, not vision, is the bottleneck. Almost everyone can demo a humanoid; almost no one can yet build them at scale.
- Watch disclosed units, not slogans. Verified production counts are the honest measure of progress, and right now Tesla's stands at zero.
The Fremont story is a useful reality check on a field prone to hype. The line is being built, the intent is clear and the long-term prize — general-purpose robots doing physical work — is enormous. But the retreat from a firm August date to "later this year" is a reminder that in robotics, the distance between a cleared factory floor and a shipping product is measured in engineering problems, not press releases.
