Why Tesla's Robot Hands Are So Hard to Build
Tesla made a few hundred Optimus robots per week last month, far below its stated goal of 20,000 per week. Hands and small-part fitting are slowing the ramp. The Verge
Earlier this year Tesla reused its Model S and Model X car lines to make Optimus. That gave it floor space and factory equipment, but the problems since then are specific to humanoid robots, not cars.
The hands are the holdup. Each V3 hand and forearm has more than 100 tiny screws and parts that a person must put together by hand. If car assembly is like building furniture, hand assembly is closer to watch repair. Robot parts are much smaller than car parts and must line up more exactly.
The V3 version is lighter and has more cameras than older versions. It still lacks a sensing glove that is in development. The glove would let Tesla swap out worn touch sensors without replacing the whole hand. Right now a broken hand means replacing the whole hand, which is a bigger repair job.
Tesla only uses Optimus inside its own buildings for simple, limited jobs. The robots are not doing general work and no person is driving them remotely. Tesla plans to lease the robots at first instead of selling them, and use what they learn to make them better. That keeps learning inside Tesla for now.
From small batches to line production
In the second quarter Tesla made only a few dozen robots per week while testing in small batches, according to The Information. Making a few hundred per week now shows work at its Fremont factory has moved beyond test benches, where Tesla says it figures out how to build the lines.
Progress has not been smooth. Tesla stopped buying robot parts while it changed the hardware and software, a pause reported in July 2025. The Information In September 2026, other coverage cited a report that hands break and suppliers cannot keep up. Electrek
Job ads point to the same problem. Tesla is hiring a Manufacturing Engineer, Optimus Hands Launch to lead hand production, and a Robotics Manufacturing Engineer to launch full-robot production including Full Bot Validation Testing.
Milan Kovac, the head of the Optimus program, said in a June 2025 post on X that he would step down. Reuters
Timelines and product definition
Elon Musk has said Optimus will be Tesla's biggest product of all time. Dates for public sale have moved. In April 2024 he said Tesla could start selling Optimus by the end of next year. Reuters In January 2026 he said Tesla will probably sell Optimus to the public by the end of next year. Bloomberg The current plan to lease first would keep early robots under Tesla control while it gathers training data.
Tesla calls Optimus a "general-purpose, bi-pedal, autonomous humanoid robot capable of performing unsafe, repetitive or boring tasks." Tesla Current limited internal use fits that as a future goal, not what the robot can do today. Recent reports describe no general use and no remote control.
The broader context here is a different kind of factory work, not just faster work. Cars allow small differences in big metal parts. Robot hands do not. Tiny errors across many small screws and wires add up fast, and building by hand hides differences that machines will later need to control.
In my view, leasing first makes sense. It keeps breakdowns and touch-sensor wear inside Tesla while the hand still needs hand-building and full replacement. The gap is large. Going from a few hundred to 20,000 per week is almost 100 times more, and hiring more people alone will not get there.
Worth flagging for other robot teams, Tesla now faces a common hardware problem. Simpler design, fewer screws, easy-to-swap sensors and reliable suppliers matter more for schedules than smart software. If Tesla finishes the glove, automates hand building and keeps part quality steady, leased robots doing simple jobs could collect the hands-on lessons needed for wider use. That would make Optimus factory tools first and public products later, which is still how useful robots usually reach people.
