Tesla has officially stated in its latest annual shareholder meeting materials that the third‑generation Optimus humanoid robot is ready for large‑scale mass production. A concrete start date is still missing, but it’s clear Tesla sees Optimus as a core product beyond EVs.
Tesla Optimus humanoid robot on display at an exhibition. — Image: Benjamin Ceci / Wikimedia Commons, Public domain
Three thresholds every humanoid must cross
Despite huge hype, virtually no humanoid robot company is profitable yet. The industry is stuck on three hard problems:
- Dexterous hands
- Supply chain & rare‑earth constraints
- The "embodied brain" – whole‑body intelligence
High‑end robot hands now reach 20+ degrees of freedom, but they’re too expensive and fragile for real‑world deployment. For Optimus Gen3, Tesla designed a tendon‑driven 22‑DOF hand. Still unresolved are the lifetime of finger joints and tactile sensors, and practical, replaceable electronic skin.
Supply chain pressure and memory cuts
Optimus will be built in Fremont, but Tesla knows a US‑only supply chain can’t support volume and cost targets. It has started auditing Chinese‑rooted suppliers like Sanhua Intelligent Controls, Tuopu Group, and Joyson Electronics, which will ship parts via their Mexican factories to ease tariffs and logistics.
China’s new rare‑earth export rules reportedly delayed Optimus by 2–3 months. At the same time, rising memory prices pushed Musk to halve AI5 platform memory to 72 GB and cut AI6 to 144 GB, claiming Optimus performance will remain unaffected.
Software is the real bottleneck
Hardware is no longer the main limiter. Current VLA models can output only around 10 motion control vectors at a given timestamp, so many fine‑grained actions still rely on teleoperation.
Musk argues that pre‑scripted motions are not real capability; robots must handle never‑seen‑before situations. The first batch of Gen3 Optimus units will therefore stay inside Tesla factories, gathering training data via an Optimus Academy before any broader commercial rollout.
Can Tesla repeat the Model 3 moment in robotics?
Chinese players like Unitree, Fourier Intelligence and Ubtech are already shipping thousands of humanoids into warehousing, logistics and retail pilots. Hardware gaps between leading vendors are now relatively small; the real race is in AI, control software and scenario‑specific solutions.
Musk claims that Optimus plus the Cybercab robotaxi will account for the vast majority of Tesla’s future value, and predicts up to billions of humanoid robots in the coming decades. Whether Optimus can become for robotics what Model S/3 were for EVs will depend not only on Tesla’s engineering, but also on how fast it can industrialize the supply chain and deploy robots into real, messy environments.
Source: Gündəm · eu.36kr.com
