Overview
Per a July 9 LatePost report citing multiple Tesla supply chain sources, Tesla has issued parts procurement guidance for the Optimus Gen 3 humanoid robot:
- September: supplier capacity of 1,000 units/week
- Year-end: 2,000–2,500 units/week, equivalent to ~100,000 units/year
- Design approval: Musk passed Gen 3 at a late-June executive review, moving the project from the lab into mass production after 3+ years of development
- Fremont factory has been converted into an Optimus production line; the final Model S/X rolled off the line in May
- Musk reportedly threatened to fire the entire Optimus procurement team if year-end capacity targets are missed
- Early 2023: Optimus project kicked off (alongside Dojo and 4680 battery programs)
- 2024: Gen 1/Gen 2 public demos (dancing, folding laundry, carrying boxes)
- March 2026: Musk calls Optimus 3 "the most advanced robot in the world"
- April 2026 (Q1 call): production at Fremont confirmed for late July/August
- Late June 2026: Gen 3 approved at executive review; year-end capacity deadline set
- July 9, 2026: LatePost reveals supply chain details
- Harmonic reducers: ~14–28 per unit → 1.4–2.8M units/year needed, vs. global capacity of ~300K (Harmonic Drive Systems) plus ~500K (Leaderdrive, 2024) — a 2–5x shortfall
- Frameless torque motors: ~28–40 per unit → 2.8–4M units/year
- Six-axis force sensors / e-skin: only a handful of automotive-grade suppliers worldwide
- AI chips: Tesla-designed Dojo/AI inference silicon, ~50–100 TOPS per unit → 5–10 EFLOPS total deployed compute
- Short term (2026 H2): faster fundraising; supply chain spillover orders to Chinese reducer/motor/sensor makers
- Mid term (2027–28): if 100K/year is hit, unit cost could fall to $20K–30K (1/2–1/3 of current industry forecasts), pressuring Chinese B2B rental pricing and shifting competition to scenario fit and software ecosystems
- Long term (2029+): Chinese vendors may pivot from B2B industrial to B2C home, education, healthcare and companionship scenarios not yet covered by Tesla
- Musk's deadline credibility is mixed (Model 3 "production hell" ultimately succeeded; Cybertruck and Robotaxi slipped repeatedly) — year-end targets may slip 3–6 months but are unlikely to be abandoned
- Harmonic reducer global capacity cannot keep up (HDS expansion cycle 18–24 months); Tesla may accelerate qualifying Chinese suppliers
- Yield on 10,000 unique parts: any critical part below ~70% yield stalls the line
- Unit cost, selling price, and order pipeline for external sales remain undisclosed
- LatePost via IT Home: https://www.ithome.com/0/974/782.htm
- Musk "Optimus 3 most advanced" remarks (March): https://www.ithome.com/0/967/375.htm
- Q1 2026 earnings call: https://www.ithome.com/0/942/371.htm
- Final Model S/X (May): https://www.ithome.com/0/948/399.htm
This is described as the first concrete timetable for the global humanoid robot industry moving from "demo" to "mass delivery."
Timeline and Key Metrics
Supply chain hard targets: 1,000/week by September, 2,000–2,500/week by year-end (~100,000/year annualized); orders for several hundred units in August already placed. Low-volume production starts in summer, high-volume in 2027. The robot uses ~10,000 unique parts — far more complex than a car.
Musk tempered expectations on X (July 1): "No, Optimus production will be extremely slow at first, as everything is entirely new. It is not like making cars."
Competitive Landscape
| Player | Stage | Commercialization | |---|---|---| | Tesla Optimus | Gen 3 locked; 1,000/week from Sept | Internal factories, future external sales | | Figure AI | Gen 2 deployments | BMW/Mercedes pilots, B2B leasing | | Unitree | H1/G1 on market; STAR board IPO approved 7/2 | B2B industrial + research/education | | Zhiyuan (Agibot) | A1/X1; BYD/Geely pilots | B2B industrial | | Apptronik | Apollo; Mercedes/DeepMind | B2B industrial | | 1X Technologies | NEO Beta; OpenAI-backed | B2C (long-term) |
Tesla's year-end 100,000-unit target is by far the most aggressive commitment in the industry.