TL;DR
Tesla is building hundreds of Optimus robots per week but cannot yet make them useful, reliable, or ready to sell, while Chinese rival XPeng has already opened an automated humanoid production line targeting commercial sales.
What happened
- Production jumped roughly 10x: Tesla went from a few dozen Optimus units per week in Q2 2026 to several hundred per week by August, per The Information.
- Fremont retooled: Tesla ended Model S and Model X production in early May and converted that line to Optimus, pulling engineers from S/X and Model Y.
- Targets are aggressive: managers aim for 1,000-plus robots per week by year-end and eventually 20,000 per week.
- Almost none are deployed externally: most units go to internal testing; factory-floor robots are confined to supervised, task-specific areas, not general use.
- The V3 being built now is not the commercial version: a stricter durability threshold must be cleared before any customer product ships, and the promised V3 reveal (due mid-2026) still has not happened as of late September.
Why it matters
- Generalization is unsolved: Optimus AI cannot reliably handle a wide range of tasks, behavior is unpredictable outside trained scenarios, and learning a single basic task still takes several days.
- Hardware reliability is a crisis: hands contain over 100 hand-assembled screws and small components, touch sensors have had failures, and Tesla's fix (a replaceable sensing glove) is not arriving until next year.
- Supply chain is a bottleneck: outside suppliers, many in China, can produce precision motors and gears in prototype quantities but struggle to hold quality at volume.
- The business model mirrors FSD: Tesla plans to lease (not sell) Optimus to a short list of companies with factory layouts similar to its own, collect deployment data, and iterate, repeating the same promise-now, deliver-later pattern that drew criticism in autonomous driving.
- China is not waiting: XPeng launched an automated IRON humanoid production line in Guangzhou this month, targeting commercial sales in 2027, and Chinese firms broadly benefit from EV supply-chain crossovers, state funding, and lower unit costs.
What to watch next
- The V3 reveal: a public demo was promised for mid-2026 and is now overdue. A credible showing with durability data would signal the commercial timeline is real.
- First external lease deployments: which companies receive Optimus units, under what task constraints, and whether Tesla discloses any reliability or productivity metrics.
- Chinese cost curves: MERICS notes Chinese humanoids need to cut costs by at least half for commercial viability. Watch for price announcements from XPeng IRON and peers as volume scales in 2027.
Originally published on Present of AI, a daily source-linked AI news timeline. Read the full timeline or browse the open dataset.