Optimus Gen3: production ramp & competition; Model S/X killed; Fremont line pivot; Figure comparison; German subsidies push; LG Energy battery deals; ARK timeline; Unitree IPO; Agility Robotics IPO; UMA launch; Optimus at Diner
Key Questions
What is the production target for Optimus Gen3?
Tesla aims for 1k units per week by September and 2k-2.5k by year-end, with the Gen3 design finalized.
How is Tesla converting Fremont lines for Optimus?
Model S/X manufacturing lines at Fremont are being decommissioned and repurposed for Optimus production as a concrete commitment signal.
Why does Musk call Optimus the hardest product to scale?
Musk cites no existing supply chain, 10k unique parts, and unsolved dexterity issues, with the Q2 deck quietly dropping 'mass production' language.
How is Tesla training Optimus robots?
Optimus learns by watching, using Grünheide employees with backpack cameras to record assembly tasks and generate training data.
Where will the first Optimus units go?
The first units will go to an internal Academy for training rather than factory deployment, with low hundreds built so far.
What competitors are advancing in humanoid robotics?
Unitree is shipping 1k units, Agility is opening an R&D center in Fremont, and XPeng is pursuing open-source VLA approaches.
What investor concerns exist around Optimus revenue?
Ross Gerber questioned the revenue timeline, noting no revenue is coming soon, while a Benzinga poll ranked Optimus as the top growth driver at 32%.
What role will Terafab play in Optimus scaling?
Musk stated Terafab is 'necessary' to scale Optimus, supporting deeper integration with SpaceX and Grok-managed Digital Optimus.
Gen3 design finalized, 1k/week target by Sep, 2k-2.5k by year-end. First units to Academy. Fremont's Model S/X lines converted to Optimus production. Musk admits Optimus is hardest product to scale ever. Line idle, low hundreds built. Optimus learns by watching; Grünheide employees record assembly tasks. Grok managing Digital Optimus, deeper SpaceX integration. Musk says Terafab 'necessary' to scale Optimus. Ross Gerber questioned revenue timeline. Tesla bought a single Virtuix Omni One treadmill for Optimus—exploratory for teleoperation/simulation. Gold Optimus variant spotted handing water bottle—minor design iteration. New detail: Optimus line takt time is 32 seconds per unit, faster than Shanghai Model Y's 40 seconds, but 10,000 new parts and immature supply chain pose massive scaling challenge. Competitors: Unitree shipping 1k units, Agility opening R&D center in Fremont, XPeng chasing with open-source VLA. BYD confirms its first humanoid robot debut in August. US government bans humanoid robots manufactured abroad, potentially advantaging Tesla's domestically-assembled Optimus. China retaliates, threatening rare earth restrictions that could hit Optimus supply chain. Tesla AI boss Ashok Elluswamy corrected Optimus production target to 10 million robots per year at Giga Texas, with a two-phase buildout: 1M at Fremont, 10M at Texas. Cost estimates at that volume remain unclear. A new article (ex-6b6b085f) confirms the Fremont line conversion and provides a realistic timeline, debunking inflated deployment numbers.