Tesla Optimus Gen 3 Finalized: Fremont Factory Retooled for 1,000 Units/Week from September

> A July 9 LatePost report, citing multiple Tesla supply chain sources, says Tesla has issued parts procurement guidelines for Optimus Gen 3: suppliers need to reach 1,000 units/week of capacity by September, rising to 2,000-2,500 units/week by year-end, translating to 100,000 units of annual capacity. Musk approved Optimus Gen 3 at an executive meeting in late June; after more than three years of development, the robot is officially leaving the lab and entering mass production. Musk reportedly declared: if capacity targets are not met by year-end, the entire Optimus procurement team will be fired. The Fremont factory has been converted into an Optimus production line, and the last Model S/X rolled off that line in May this year. > > This is the first concrete timetable for the global humanoid robot industry to move from "demo" to "mass production and delivery" — Musk, in his trademark "miss the target and people get cut" style, has set a deadline-backed commitment for the supply chain, capital markets, and the entire humanoid robot industry.

---

1. Gen 3's Mass Production Timeline and Capacity Ramp

Development milestones:

• Early 2023: Optimus project initiated (alongside Dojo supercomputer and 4680 battery programs)
• 2024: Gen 1/Gen 2 public demos (dancing, folding laundry, carrying boxes)
• March 2026: Musk first describes Optimus 3 as "the most advanced robot in the world by far"
• April 2026, Q1 earnings call: Musk confirms Fremont production start in late July or August
• Late June 2026: Executive review approves Gen 3; Musk sets year-end capacity deadline
• July 9, 2026: LatePost exclusively reports supply chain details

Hard supply chain targets:

• September: 1,000 units/week
• Year-end: 2,000-2,500 units/week
• Annualized capacity: 100,000 units/year
• Specific orders for several hundred units in August have already been placed (placing orders 2 months ahead is standard Tesla practice)

Fremont factory conversion:

• Former Model S / Model X line (last vehicle in May) retooled for Optimus
• ~10,000 unique parts — far more complex than a car
• Low-volume production starts this summer; high-volume production begins in 2027

Musk's statements:

• July 1 post on X: touring the Optimus production line at Fremont
• On speculation that production was ahead of schedule: "No, Optimus production will be extremely slow at first, since everything is brand new. It's not like making cars."
• Late June executive meeting: "Hit capacity targets by year-end, or the entire Optimus procurement team is fired"

---

2. Why Gen 3's Finalization Is an Industry-Level Event

The humanoid robot landscape as of mid-2026:

| Player | Current Stage | Key Metric | Commercialization Path | |------|---------|---------|-----------| | Tesla Optimus | Gen 3 finalized; 1,000 units/week from September | 100k units/year (year-end) | Internal factories + future external sales | | Figure AI | Gen 2 commercial deployment | BMW / Mercedes factory pilots | B2B industrial leasing | | Unitree | H1 / G1 on market | CSRC approved STAR Market IPO registration (July 2) | B2B industrial + research/education | | Zhiyuan Robotics | Expedition A1 / Lingxi X1 | BYD / Geely factory pilots | B2B industrial | | Apptronik | Apollo | Mercedes / Google DeepMind partnership | B2B industrial | | 1X Technologies | NEO Beta | OpenAI-backed / home services | B2C long-term |

Optimus Gen 3's finalization and production timetable is the first commitment in this table with concrete numbers. Other players are still demoing (Apptronik), in IPO processes (Unitree), or in early B2B pilots (Figure). Tesla is directly promising 100,000 units by year-end — a number that, in the context of the global humanoid industry in H2 2026, is aggressively extreme.

But it is not mere bluffing. At the late-June executive meeting, Musk used "firing the entire procurement team" as credibility collateral — his way of signaling to the supply chain, capital markets, and the robotics industry: "I don't believe numbers in OKR spreadsheets; I only believe deadlines." This bet-your-personal-credibility approach has been repeatedly validated in Tesla's history (Model 3 production hell, 4680 battery ramp, Cybertruck delivery).

---

3. Why 100,000 Units/Year Matters

Pushing 100k units/year up the supply chain:

Reducers / harmonic joints:

• Each Optimus needs roughly 14-28 harmonic reducers (shoulders, elbows, wrists, hips, knees, ankles, waist)
• 100,000 units/year = 1.4-2.8 million harmonic reducers
• Harmonic Drive Systems (Japan) currently has global capacity of ~300,000 units/year; Leaderdrive (China) ~500,000 in 2024
• Tesla's late-2026 demand equals 2-5x existing global capacity — a hard constraint the supply chain must solve

Frameless torque motors:

• ~28-40 per unit (1-2 per joint)
• 100,000 units = 2.8-4 million motors
• Kollmorgen, Maxon, and domestic players (Kinco, Leili) all need to expand capacity

Six-axis force/torque sensors / electronic skin:

• Dozens of sensors per unit
• Only a handful of companies worldwide can stably supply automotive-grade six-axis force/torque sensors

AI chips:

• Optimus's "brain" is expected to be Tesla's in-house Dojo / AI inference chip (possibly Samsung 4nm or a custom Tesla NPU)
• ~50-100 TOPS per unit
• 100,000 units/year = 5-10 EFLOPS of total deployed compute

Every supplier on this chain faces the prospect of being locked into Tesla orders starting late 2026 — the same script as Panasonic's battery lock-in during the 2018 Model 3 production hell.

---

4. Impact on Chinese Humanoid Robot Companies

Unitree received STAR Market IPO registration on July 2; Zhiyuan Robotics closed a Series B in June (valuation over 10 billion RMB); UBTech, Fourier, Galbot and others are all racing for the window. Implications of Optimus Gen 3's timeline:

Short term (2026 H2):

• Chinese humanoid companies accelerate fundraising — once Optimus mass-produces, the question "does the humanoid track really have commercial prospects?" is answered, lifting valuations across the board
• Optimus's supply chain spillover may reach China — reducer, motor, and sensor suppliers could receive orders from both Tesla and domestic customers

Mid term (2027-2028):

• If Optimus truly hits 100k units/year, unit cost could drop to $20,000-30,000 — half to one-third of current industry forecasts
• Chinese B2B industrial leasing gets price-pressured — customers will ask "why can Optimus do it for $20k when yours costs $80k?"
• Differentiation shifts from hardware specs to scenario fit and software ecosystems (the direction Figure AI and Unitree G1 are already betting on)

Long term (2029+):

• Once Optimus enters external sales, it collides head-on with Chinese price competition
• Chinese players may be forced out of B2B industrial markets, pivoting fully to B2C home services, education, healthcare, and companionship — scenarios Tesla doesn't yet cover
• But hardware requirements (safety redundancy, long-tail interaction) and business models (subscription vs. one-time sale) for those scenarios remain immature

---

5. Risks and Watch Points

Credibility of Musk's deadlines:

• Model 3 production hell (2018) — eventually delivered despite the chaos
• Cybertruck (2024) — timeline slipped repeatedly
• Robotaxi (2024) — delayed multiple times; Cybercab only began production July 1
• Optimus Gen 3's year-end 100k target will likely slip 3-6 months — but won't collapse. Tesla's execution culture ensures the commitment is fulfilled in some form

Supply chain risks:

• Global harmonic reducer capacity can't keep up — Harmonic Drive needs 18-24 months to expand, potentially forcing Tesla to accelerate Chinese suppliers like Leaderdrive into the chain
• Yield on 10,000 unique parts — if any critical component's yield drops to 70%, the whole line stalls
• Automotive-grade reliability of six-axis force sensors — an industry-wide bottleneck no single company can solve

Commercialization uncertainty:

• Optimus's "internal use" value (transport, assembly in Tesla factories) is validated — but only for simple repetitive tasks
• The "general-purpose home service" scenario for external sales is far from mature — hardware cost, software capability, and safety standards are all lacking
• The investor question: post-ramp, what's the unit cost? The price? Where are the orders? Musk has given none of these three numbers

> Source links: > - IT Home (LatePost repost): https://www.ithome.com/0/974/782.htm > - Musk's March "most advanced robot in the world" remark: https://www.ithome.com/0/967/375.htm > - April Q1 earnings call: https://www.ithome.com/0/942/371.htm > - Last Model S/X off the line in May: https://www.ithome.com/0/948/399.htm