Tesla has moved its Optimus humanoid robot into production at its Fremont, California, factory, with the company claiming the robot learns from visual observation rather than pre-programmed instructions2.
"Those demonstrations are pre-programmed or remote controlled. There is no humanoid robot that is actually able to do generalized tasks. Optimus will be the first one that is capable of doing that," CEO Elon Musk said on the company's second-quarter earnings call on July 22, 2026.
The underlying AI approach mirrors the data-flywheel strategy Tesla uses for its Full Self-Driving driver-assistance software. Ashok Elluswamy, Tesla's vice president of AI, said on the same call: "Just like Full Self-Driving, we have access to a broad fleet of humans giving us data from all of the workers at our factory. Optimus can learn quite a bit from observing them perform their tasks".
Tesla frames the observation-based learning model as the key to scalability. Programming a robot for every possible task is not feasible, but learning by observation draws on virtually limitless data: factory floor footage runs continuously, and video of humans performing everyday tasks already exists at internet scale. The company says Optimus "practices, fails, learns and improves" in a manner analogous to a new employee.
Manufacturing the robot at scale presents its own challenge. Unlike cars, which draw on decades of existing supplier networks for components such as glass and body panels, Optimus has no established supply chain, and Tesla is building one from scratch.
The Optimus announcement landed alongside a mixed financial picture. Tesla posted revenue growth in Q2 but missed earnings-per-share expectations, with operating margins dropping to 1.4% and free cash flow turning negative1. Costly projects including Optimus, Cybercab, and Megapod lack clear profitability visibility, according to one analysis of the quarter's results.
ANALYSIS The learning-by-observation claim, if validated at scale, would represent a meaningful departure from the scripted or teleoperated approaches common in humanoid robotics demonstrations. However, Tesla's framing — that no other humanoid robot can perform generalized tasks — is a company assertion that has not been independently verified. The financial context adds a layer of tension: Optimus is entering production during a quarter in which Tesla's margins compressed and cash flow turned negative, raising questions about the capital intensity of bringing a novel hardware platform to scale without an existing supplier ecosystem.