Tesla increased Optimus production at its Fremont factory roughly tenfold, now manufacturing several hundred units per week—a jump from the few dozen units produced during small-batch testing in the second quarter.
The ramp followed a major strategic bet in early May: Tesla ended Model S and Model X production to convert the Fremont line for Optimus, reassigning workers and engineers from those programs and dozens more from Model Y. Management aims for over 1,000 robots per week by year-end, with a long-term target of roughly 20,000 units weekly.
But the math on unit economics is brutal. The robots leaving the line today cannot perform generalized tasks. Most are used internally for testing and data collection. Those deployed in Tesla factories operate only in tightly controlled, supervised areas and are programmed for specific tasks—hardly a proof point for the general-purpose machine Musk has promised.
The core problem is mechanical. The robot's hand and forearm contain over 100 screws and small components still assembled by hand. Assembly line fixtures for hands, joints and electronics testing fail to consistently align parts. These tolerances are far tighter than automotive manufacturing requires, forcing significant rework. Some touch sensors in the hand exhibit reliability issues; Tesla plans a replaceable "sensing glove" next year to avoid swapping entire hands for repairs.
Supply chain constraints compound the issue. Motors and precision gears for the robot's joints come from outside suppliers, many in China, that struggle to maintain quality at higher volumes despite producing good prototypes. This is a classic scaling trap: suppliers who perform at low volumes often cannot hit the consistency required for thousands of units weekly.
The deeper bottleneck is artificial intelligence. Three individuals with knowledge of the system report that Optimus' AI cannot reliably handle a wide range of tasks and behaves unpredictably in situations for which it has not received specific training. Tesla is building a library of basic movements for the robot to combine, but Optimus still requires several days to learn even elementary tasks—a massive gap in autonomous learning.
The V3 robots in current production are not the version intended for commercial sale. Tesla's commercial model still needs to pass stricter durability and reliability thresholds. Until the AI can handle variable real-world conditions without days of retraining per task, the path from test line to paying customer remains theoretical.
