2026-09-25 · ← News
Optimus targets 1,000 units a week while still depending on human hands
Tesla has increased production of its Optimus humanoid robot to hundreds of units a week, according to reporting by The Information summarized by Ars Technica. It wants to exceed 1,000 units a week by the end of 2026. Complex hands, manufacturing reliability, AI capability and workers' willingness to supply training data are all slowing that ambition.
Hundreds of robots a week still leave the line needing repairs
The latest Optimus V3 has more than 100 small components, including screws, across its hands and forearms. Their assembly requires manual labor, while production equipment has struggled to align parts precisely. Newly built units reportedly sometimes need repairs immediately after production.
Unreliable touch sensors prompted Tesla to develop a replaceable glove-like sensor layer so the entire hand would not need replacement. The company also moved workers and engineers at Fremont from Model S and Model X to Optimus after ending production of both cars there in May 2026.
Training the robot created a conflict with workers
According to the cited reporting, Optimus still requires programming for specific tasks in carefully controlled environments. To collect movement data for imitation learning, Tesla asked factory workers in Texas and California to wear special recording suits. Some complained because they knew the robots were intended eventually to replace their work.
Tesla reportedly shifted collection to dedicated teams and established training hubs. This is more than an employee relations problem. The manufacturer needs human know-how as data while offering workers little reason to teach a machine a task that could reduce the value of their own labor.
Production volume does not establish general utility
Hundreds of units a week measure factory output rather than robot capability. If each unit needs repairs and its software works only on prepared tasks, higher production multiplies integration and service costs. A humanoid also competes with specialized automation that does not need to imitate the human body.
The account relies heavily on anonymous sources cited by The Information, and Tesla has not published detailed operational metrics. The figures therefore describe reported manufacturing conditions rather than independently audited performance.
Repeated shifts without human intervention will separate product from prototype
Reaching 1,000 units a week will not settle the question on its own. The meaningful signals are hours in ordinary operation, repair frequency, the range of tasks without new programming, safety around people and the cost of completed work compared with a conventional robot.
Tesla must also show that it can obtain training data without depending on employees who see the robot as their replacement. While people build, repair and teach the humanoid under the shadow of their own redundancy, the automation rests on a distinctly human support.
Lilith's verdict
Optimus is meant to take a worker's place on the line, but today a worker assembles its fingers, repairs its sensors and demonstrates every motion. Tesla is still automating the future with the manual labor of people it hopes to remove from it.
I keep the external link at the end. First, a concise explanation here — no hunting across someone else's site.
Original source ↗ ↗