Tesla’s robotics team landed in Ningbo, China, on September 16, 2026, and opened a new round of production audits for Optimus the following day.
According to supply-chain sources cited by Chinese finance outlets and confirmed in a Bloomberg report published September 18, the audit is meant to verify that Optimus components can be built with the exclusivity and consistency mass manufacturing requires, and it doubles as the mechanism for migrating parts of Optimus’s manufacturing know-how from Tesla’s U.S. factories to its Chinese suppliers.
The company’s stated target is approximately 50,000 Optimus units built in 2026 for deployment across its gigafactories worldwide. But the orders suppliers currently hold in hand are far smaller — in the thousands, not tens of thousands, as Tesla’s first production order in September showed.
That gap between a stated goal and a confirmed order book is precisely what an audit like this one exists to close.
News of the audit rippled through China’s A-share market on September 17, according to a BigGo Finance report: Ningbo Joyson Electronic hit its daily up-limit, Tuopu Group rose 4.36%, and Sanhua Intelligent Controls gained 2.70%. The Solactive China Humanoid Robotics Index climbed as much as 2.4% on the report.
Tesla Optimus audit, what the Ningbo visit actually covers
Three things happened at once when the audit team touched down in Ningbo. First, engineers began verifying that Optimus components — precision parts like planetary roller screws, dexterous-hand sensors, and torque sensors — can be produced with automotive-grade consistency across multiple supplier lines, not just as one-off prototypes.
Second, the team began helping supplier facilities debug new production equipment on-site. Third, and most consequentially, the audit is the vehicle for shifting manufacturing capability that used to live inside Tesla’s U.S. plants over to its Chinese supply base.
The order book behind this audit has grown in steps. According to a Jiemian News report, Tesla placed a batch order for roughly 5,000 sets of Optimus components with core suppliers in early September — the first time an order reached four-digit volume, after earlier trial runs of only a few hundred units.
A separate LatePost report says Tesla simultaneously issued capacity guidance asking suppliers to reach production of 1,000 units a week by the end of September, rising to 2,000–2,500 units a week by year-end.
At that year-end weekly rate, suppliers would have the capacity to support roughly 100,000 units annually — well above the 50,000-unit figure Tesla has stated as its 2026 target.
That mismatch — a supplier capacity build-out sized for 100,000 units a year, a public target of 50,000, and confirmed orders sitting at 5,000 to 15,000 — is the numeric core of this story. It reflects a company building manufacturing headroom ahead of firm demand, a pattern Tesla has used before in vehicle production.
The companies drawn into this round include familiar names from Tesla’s automotive supply chain: Ningbo Tuopu Group, Sanhua Intelligent Controls, Zhejiang Rongtai, Xinquan Auto Parts, and Joyson Electronics.
Audit standards reportedly borrow directly from automotive practice — fast cycle times, high yield rates, and per-unit cost breakdowns aimed at converging toward a roughly $20,000 target selling price, alongside requirements that suppliers show they can manufacture overseas and manage their own second-tier vendors.
Tesla Optimus audit, why a second round was needed
Optimus’s road to this point has been anything but linear. Elon Musk first unveiled the “Tesla Bot” concept at AI Day in August 2021. A prototype named Bumblebee followed a year later, able to walk but still needing human help onstage.
Gen 1 arrived via video at the March 2023 Investor Day, shown tightening bolts on a factory floor; Gen 2 followed in December 2023, roughly 10 kilograms lighter, about 30% faster on its feet, and carrying new foot-torque sensors and full-finger tactile sensing.
The “We, Robot” event in October 2024 showed Optimus units bartending and dancing — but by most accounts at the event, those demonstrations were remotely operated rather than autonomous.
Behind the scenes, Tesla was reportedly struggling through late 2024 and early 2025 with an unacceptably high failure rate in Gen 2’s transmission components, a problem serious enough to block commercial-scale production.
The program’s leadership then turned over: Optimus engineering lead Milan Kovac departed in mid-2025, and Optimus AI lead Ashish Kumar left for Meta that September.
A CITIC Securities research note from December 2025 marks that June 2025 leadership change as the pivot point toward a genuine mass-production plan, with Tesla announcing that September it would expand robot production and launch a redesigned Gen 3 before year-end.
Supply-chain sources say the first intensive round of China-based audits began that same September and ran in waves through roughly February 2026 — a round focused on moving suppliers from prototype-stage validation to manufacturing-readiness, covering joint-module integration, motor-reducer-controller-encoder assembly, and repeatable production of the same precision parts now under a second round of scrutiny.
This built on an earlier first production order Tesla placed with suppliers.
This is, in other words, Tesla applying the discipline it already uses for cars — audited supplier qualification, cost target convergence, tier-2 governance — to a fundamentally different kind of product.
No other automaker is running a comparable audit process for a humanoid robot supply chain at this scale, and the fact that Tesla can lean on suppliers it has already qualified for vehicle production (Tuopu and Sanhua both supply Tesla’s EV lines) gives it a head start that a robotics-only startup would not have.
Tesla is also visibly hedging the geographic concentration this process creates.
Filings with Thailand’s Board of Investment show five of the Chinese suppliers involved in or expected to join this audit round — Xinjian Transmission, Beite Technology, Sanhua Intelligent Controls, Tuopu Group, and Xusheng Group — have been approved to build factories in Thailand to produce robot skeletons, joints, and finger/arm control systems.
Relocating a slice of this capacity to a third country is a direct response to tariff and geopolitical risk, not an afterthought.
What would have to go right, and what still hasn’t
The dependency this audit is trying to manage is real and, according to Morgan Stanley’s estimate, expensive to unwind: if Optimus Gen 2 excluded Chinese components entirely, the bank calculates material costs would rise from roughly $46,000 to roughly $131,000 per unit.
That estimate helps explain why Tesla is deepening its China supplier relationships even as it opens a Thailand hedge rather than choosing one path over the other.
Raw materials add another layer of exposure.
After China imposed export controls on seven rare-earth elements and related magnetic materials in April 2025, Musk acknowledged on that quarter’s earnings call that the controls had affected Optimus production, saying Chinese authorities wanted assurance the magnetic materials “won’t be used for military purposes” — assurance he noted was easy to give, since “they’re just for humanoid robots.”
Tesla’s own production guidance for Optimus has moved several times, and it is worth laying out plainly: at the June 2024 shareholder meeting, Musk said limited 2025 production of “over 1,000 units, possibly several thousand.” By the January 2025 earnings call, that had been revised up to roughly 10,000 units produced internally in 2025.
By March 2025, it was revised down again — 5,000 units in 2025, 50,000 in 2026. By the October 2025 earnings call, Musk acknowledged the 5,000-unit 2025 target had been shelved entirely, pivoting instead to a Q1 2026 Gen 3 launch and a stated ambition of a million-unit-per-year line by the end of 2026.
September 2026’s capacity guidance to suppliers — 1,000 units a week rising to 2,000–2,500 — is the latest data point in that sequence, and it describes supplier capability, not confirmed shipped units.
Two more limits are worth stating clearly. Ordinary consumers still have no way to buy an Optimus unit; the units coming off this line go first to Tesla’s internal “Optimus Academy” training program and its own factories, with public sale still targeted for the end of 2027.
And Tesla has not officially confirmed the 50,000-unit figure itself — it comes from supply-chain sources, not a company statement, which is a distinction worth keeping in mind given Tesla’s history of guidance revisions.
None of this erases what actually changed this month: a real audit team, on the ground in China, checking whether parts can be built the same way twice in a row — which is the unglamorous, necessary step between a company saying it will build tens of thousands of robots and a supply chain that can actually do it.
References
BigGo Finance, “Tesla’s Optimus mass-production audit lands in Ningbo”, Sept. 17, 2026
Bloomberg, “Chinese Robotics Suppliers Rise on Tesla Optimus Audit Report”, Sept. 18, 2026
NextBigFuture (Brian Wang), “OPTIMUS CONFIRMED: 15,000 Bots This Year,” Sept. 7, 2026
CITIC Securities research note, Dec. 2025 (as cited by BigGo Finance)
