As a multinational corporation that represents U.S. investment in frontier industries such as AI (Artificial Intelligence) and robotics on one hand, and relies heavily on China's manufacturing capabilities and supply chain system on the other, Tesla has now sent another team to China to find the answer to the mass production of Optimus.
According to the South China Morning Post, Tesla executives and engineers have recently begun auditing Chinese suppliers, involving Zhejiang-based companies such as Sanhua Intelligent Controls, Joyson Electronics, and Tuopu Group. Information subsequently obtained by Chao News from manufacturing enterprises in Ningbo also confirms that the Tesla team has indeed visited multiple factories recently. Currently, the focus is mainly on auditing small-batch production, and large-scale orders have not yet been determined.
These seemingly scattered pieces of news actually overlap on the same backdrop: on one side, Washington is releasing a rare gesture of cooperation ahead of the presidential visit; on the other side, the U.S. Federal Communications Commission (FCC) just set up a higher threshold for Chinese robots to enter the U.S. market with a new regulation covering foreign-produced robots at the end of July. Politics is telling a story of "easing," while the industry is playing out another story of "indispensability."
This easily brings to mind Tesla seven years ago. At that time, the Model 3 experienced a "production hell," and the Shanghai factory and Chinese supply chain later became crucial fulcrums for Tesla to rapidly expand capacity and reduce manufacturing costs. Now it is Optimus's turn to face mass production issues, and Tesla has once again begun intensive inspections of Chinese factories.
But this time, Tesla's renewed closeness with the Chinese supply chain is taking place in a more polarized policy environment.
On July 28, the FCC added "foreign-produced advanced robotic equipment" to the Covered List. According to the regulations, such equipment cannot obtain the FCC equipment authorization required to enter the U.S. market unless it receives conditional approval from the U.S. "Department of War."
In other words, for a foreign-produced advanced robot to enter the U.S. market, it must first pass the hurdle of the "Department of War."
Reuters directly pointed this policy at Chinese humanoid and quadruped robots in its report, naming Chinese manufacturers such as Unitree as potentially affected. As early as May 2025, all members of the U.S. House Select Committee on the Strategic Competition Between the United States and the Chinese Communist Party jointly wrote to the U.S. Department of Defense, the Department of Commerce, and the FCC, requesting an investigation into Unitree and considering its inclusion in the "Chinese Military Company List," the Commerce Department's Entity List, and the FCC list.
Further back, the U.S. has already set access restrictions on the software and communication hardware of Chinese connected vehicles, and drones subsequently entered a similar security review framework. A clear U.S. policy line is taking shape: Chinese smart hardware equipped with cameras, sensors, and communication modules, capable of moving, collecting data, and even executing actions, is being re-evaluated as "security products" rather than just "industrial products."
By now, the contradiction is very clear: the U.S. wants to keep Chinese robots out of its market, yet U.S. companies still need to go to China to find robot supply chains. Governments can redraw market boundaries, but engineers cannot bypass industrialization.
What Tesla Lacks Now Is Not a Robot Blueprint
Optimus has been showcased for many years, but the real problem that remains unsolved is how to turn dozens or hundreds of robots into tens of thousands or even more products that can be stably delivered.
These two things are not at all the same level of difficulty. Building robots in a laboratory allows for picking the best parts, manual debugging, and solving problems one by one. On a mass production line, motors, reducers, ball screws, sensors, controllers, and housings—every single component must find a new balance among cost, lifespan, consistency, and delivery speed.
A joint weighing tens of grams may require redesigning the motor, structural parts, and wiring; an actuator with insufficient lifespan may force simultaneous modifications to materials, machining precision, and control strategies.
Therefore, what Tesla is auditing in China now is less about a few suppliers and more about whether Chinese manufacturing can transform Optimus from a "robot product" into an "industrial product."
Tuopu Group has already cooperated with Tesla on linear actuators, rotary actuators, and dexterous hand actuators; Sanhua Intelligent Controls is also developing robot joint actuators. These companies did not suddenly switch to the robotics business; their capabilities in motors, motor control, thermal management, precision manufacturing, and quality control, formed from previously serving NEVs (New Energy Vehicles), naturally migrated to robotics.
What Tesla is looking for is not just a few cheaper parts, but utilizing China's manufacturing system that has been running for many years.
Tesla itself does not avoid this point. In April this year, Wang Hao, head of Tesla China, publicly stated that Shanghai's manufacturing capabilities could be the key to solving the mass production challenges of humanoid robots.
Although Tesla is building an Optimus production line in Fremont, U.S., "where to do final assembly" and "who to rely on to make the product" are inherently two different things. After the Shanghai Gigafactory went into production in 2019, the Chinese supply chain helped Tesla solve a large-scale manufacturing problem for cars once. Seven years later, the problem it encounters with robots is, to some extent, the same question: the design can be completed in the U.S., but ultimately, it must find enough people who can quickly turn the design into a product.
Only this time, what Chinese manufacturing can provide is no longer just capacity.
The Chinese Supply Chain Is Beginning to Step into Product Definition
When we talk about Chinese manufacturing, the most frequently used word is cost. Simply put, things made in China are cheap to buy, which is why foreign companies come to use them.
This explanation is obviously insufficient when applied to the Chinese robotics industry.
Humanoid robots are far from forming a stable technological route like automobiles. How to build joints, what structure to use for dexterous hands, where to place sensors, how to allocate computing power, and even what the robot should ultimately look like, are all still in the stage of continuous modification.
At such times, how close the supplier is to the R&D team is much more important than how much cheaper a single screw is.
Every modification to the product means redesigning, prototyping, machining, assembling, and testing. If the motor is in one city, the ball screw is on another continent, and the structural parts need to be transported across borders, one round of modifications could take weeks or even months. However, in industrial clusters like the Yangtze River Delta and the Pearl River Delta, many things can be compressed into a few days.
This is also why the Chinese robotics supply chain is increasingly difficult to be simply understood as a "contract manufacturer." When engineers need to work with suppliers to determine actuator dimensions, machining methods, materials, tolerances, and even assembly processes, manufacturing has already begun to enter R&D.
Whoever is responsible for production will participate in telling the OEMs (Original Equipment Manufacturers) "what can be built," "what is too costly," and "what design is suitable for mass manufacturing." The supply chain thus begins to influence the product in return.
What Tesla really needs to confirm in this round of factory audits is not only whether the companies have enough machines, but whether they have the ability to keep iterating along with Optimus.
If robots remain in a period of rapid iteration in the coming years, this capability is even more important than a fixed order.
What Is More Interesting Is That Tesla Is Not the Only One Coming to China
If it were only Tesla, this could still be explained as a continuation of the automotive supply chain. But over the past year, chip companies at the other end of the global robotics industry have also begun to lean towards China.
On September 15, the Dutch company NXP established a China Robotics Innovation Center; three weeks prior, the Japanese company Renesas launched a Physical AI and Robotics Laboratory in Beijing. The German company Infineon is cooperating with a Shenzhen company to develop a flexible tactile dexterous hand, and the U.S. company AMD has also built an embodied robot R&D platform with a Shanghai company.
What they are doing is exactly the opposite of Tesla. Tesla is searching from downstream to upstream, looking for companies that can build robots; chip manufacturers are moving from upstream to downstream, hoping their processors, control chips, and software platforms can enter the OEM design earlier. But in the end, they all arrive at the same place: Chinese robotics companies and supply chains.
The reason is not complicated. Physical AI is not like mobile phone chips; many capabilities cannot be verified independently in an office. A failed grasp could be a problem with the model, or it could be an issue with sensors, finger structures, joint control, or communication latency. Algorithms, chips, and mechanical structures must be put together and continuously tested. This means that whoever owns more robot bodies, more suppliers, and more real-world application scenarios can provide more opportunities for engineering iteration.
Data from the International Federation of Robotics shows that in 2024, China added approximately 295,000 industrial robots, accounting for 54% of the global total; the number of industrial robots in service domestically has exceeded 2 million, and the market share of local manufacturers in China has also surpassed that of foreign companies for the first time.
Centered around motion control, servos, motors, reducers, sensors, machine vision, precision machining, and system integration, China has formed a sufficiently large engineering network. Now, humanoid robots are directly inheriting this network.
The Real Contradiction Is Not "Whether Decoupling Has Succeeded"
From the government's perspective, setting restrictions on Chinese robots has its own security logic. Robots are different from ordinary mechanical equipment in the past; they come with cameras, microphones, communication systems, and increasingly strong autonomous decision-making capabilities. U.S. regulators and members of Congress are concerned about data collection, remote access, and dual-use risks, and these concerns have been written into policy documents.
But companies face a different set of problems. Are there enough reliable actuators today? Can costs continue to drop? How soon can suppliers re-prototype after a design change? If tens of thousands of parts are needed next year, who can deliver them stably?
National security policies cannot solve these engineering problems. This is also a very interesting split in the current robotics industry: market access is being politicized, while the supply chain continues to cluster according to efficiency. The former attempts to reduce dependence on Chinese technological products, while the latter continues to concentrate in China due to cost, manufacturing capabilities, and development efficiency.
In the future, a situation could completely emerge where fewer humanoid robots of Chinese brands are seen in the U.S. market, but within U.S. robotics companies, a large number of Chinese-made actuators, structural parts, and transmission components are still used, and they even rely on Chinese suppliers to complete product iterations.
In other words, "de-Chinese robots" does not automatically equal "de-Chinese robot supply chain."
Behind the Chinese Supply Chain Are Chinese OEMs
Seven years ago, the Chinese supply chain was mainly helping a U.S. NEV (New Energy Vehicle) company increase production and reduce costs. Now, this supply chain serves not only overseas customers. This is also the biggest difference between today and when Tesla entered Shanghai back then. When reporting on Tesla's inspection, Chao News mentioned that domestic robotics companies are also looking for suppliers in Ningbo.
Unitree, Agibot, Fourier Intelligence, and an increasing number of Chinese robotics companies are themselves continuously putting forward new requirements for this supply chain. Unitree has already sold humanoid robots to universities, research institutions, and developers; in NVIDIA's GR00T N1 research, China's Fourier GR-1 was directly used for model deployment and dual-arm operation verification.
Thus, a cycle begins to emerge: OEMs demand lighter, cheaper, and more durable joints, and suppliers follow to modify them; suppliers produce the products, allowing the next robotics company to adopt them directly; new robot bodies and scenarios increase, which in turn attracts chip, algorithm, and component companies to enter.
This is completely different from simply relying on a single large customer. Tesla is, of course, still important; it can bring huge orders and provide suppliers with global endorsement. But what it faces is no longer a group of factories waiting for overseas orders, but an industrial system that is being fed and grown simultaneously by China's local robotics companies.
This is the hardest part of the Chinese robotics industry to replicate. Parts can be bought, factories can be built, and subsidies can be given. But putting together hundreds of companies, multiple technological routes, a large number of engineers, and continuously emerging new projects, and having them repeatedly test, modify, re-quote, and re-produce every day takes time.
The competition faced by Chinese robots is shifting from "whether products can enter overseas markets" to another more complex contest: even if Chinese brands are kept out, how deep can China's manufacturing capabilities, engineering experience, and supply chain leave their mark in the global robotics industry? The answer is, at the very least, that it is not so easily cut off.