Note: I am currently travelling. Hence the more limited postings. Actually, Suzanne, David and myself are in China for a 10-day China tech tour.
December 12, 2025
Shanghai is China’s robotics hub.
Like Shenzhen, Shanghai is very electric, an eerily silent city when no ICE trucks around.
No BYD taxis here like in Shenzhen (see Shh…). This is SAIC Motor’s (state owned Shanghai Automotive Industry Corporation) turf. Many Teslas though, thanks to its gigafactory here.
When Suzanne and I were here in 2004, young Chinese often came to us to chat in English, seeking to learn what they perceived was needed for their future.
One generation later, they still get English classes at school but very few speak much English. They don’t seem to think they need it now. Mandarin and tech are their future. We sense them solid, determined and assertive, and not at all cocky.
And very welcoming.
On the battlefield for technological supremacy, robotics is now on the front line and China’s efforts are centered in Shanghai.
So far, robotics was essentially supplying fixed industrial robots. China already totally dominates this field. At the end of 2024, China had over 2 million installed industrial robots. The US had only 542,000 and showing no momentum since 2018.
The US accounted for 6% of world robot installations in 2024, in third place behind Japan (8%). China: 54%
The battle is now getting more serious: this is the “bots on the ground” era.
The Shanghai Humanoid Robot Kylin Training Ground, covering approximately 5,000 square meters (54,000 sq ft) in the Zhangjiang area of Shanghai, is China’s first national-level, public facility designed to mass-train humanoid robots from different companies in a shared environment.
Think of it as a massive “gym” or “driving school” for robots, where instead of building muscles, they build AI intelligence through endless repetition of tasks.
China clearly displayed its robotics objectives by inserting the word “kylin” above. “Kylin” is a standard brand name for homegrown Chinese “core technologies” intended to replace foreign dependencies.
The facility’s core purpose is to accelerate the entire Chinese robotics industry by centralizing data collection and standardizing how robots learn complex tasks.
It is a “heterogeneous” facility, meaning it trains robots from competing manufacturers (like AgiBot, Fourier, and Unitree) side-by-side rather than serving just one company.
It aims to collect 10 million high-quality physical data entries (robot movements, interactions) to build a shared “brain” for the industry.
The facility is divided into specialized zones that simulate real-world environments. Robots run 24/7 to practice specific skills until they achieve high reliability.
Robots practice more than 45 basic manipulation skills, such as grasping irregular objects, using tools, and navigating cluttered spaces.
The facility also acts as a standard-setter. Just as cars need to pass crash tests, robots here must pass proficiency exams in these standardized environments to prove they are ready for factory work.
This project is a state-backed initiative led by the National and Local Co-Built Humanoid Robotics Innovation Center. It represents a “public utility” model for AI development:
By providing the physical space and testing infrastructure, it saves individual startups from building their own expensive and time-consuming labs.
It creates a unified standard for what a “work-ready” humanoid robot looks like, preventing fragmentation in the fast-growing Chinese market.
I could not find a direct US equivalent to China’s approach to such crucial strategic technology.
The US National Institute of Standards and Technology in Maryland focuses on metrology, the science of measurement, creating the “ruler” by which robot capabilities are judged. It does not operate as a mass-training facility for commercial AI models.
-
NIST creates standards and exams (certifying that a robot can do X), whereas Shanghai Kylin provides the practice gym (helping the robot learn to do X through repetitions and interactions).
-
NIST facilities are primarily for research and standards development, though they collaborate with industry. They do not typically house hundreds of commercial robots simultaneously for mass training.
-
The US model (NIST) emphasizes “measurement science” to ensure safety and reliability in a competitive market. The Chinese model (Kylin) emphasizes “infrastructure sharing” to accelerate the entire industry’s development speed.
Tesla’s Optimus Lab in Texas is strictly private and proprietary.
MassRobotics (Boston) offers shared workspace and equipment for startups but operates as an innovation hub/incubator rather than a large-scale data generation factory.
The “US model” relies on private enterprise, venture capital, and a university system that fosters fundamental research and innovation.
The “China Model” offers several strategic advantages over the decentralized US approach.
Its primary superiority lies in treating AI training data as a public utility rather than a private asset, which theoretically allows the entire industry to advance faster than any single company could on its own.
In the US, valuable training data is trapped in “walled gardens.” Tesla does not share its Optimus data with Boston Dynamics, and Figure AI does not share with Agility Robotics. This means every US company must solve the same basic problems (e.g., how to open a door, how to walk on gravel) from scratch.
The Kylin center aggregates data from 100+ (1000+ in 2027) different robots into a single, massive dataset targeting 10 million entries. This creates a shared “industry brain.”
A startup’s robot can theoretically learn from the mistakes made by a competitor’s robot yesterday, accelerating the baseline intelligence for everyone.
The US market typically waits for a clear winner to emerge before setting standards (e.g., waiting for Tesla’s NACS charging port to become the standard). This can lead to years of fragmentation where parts and software aren’t compatible.
The Kylin center acts as a central certification authority from day one. By forcing all robots to pass the same “exams” in the same physical facility, it enforces early standardization. This ensures that supply chains (batteries, joints, sensors) can be standardized faster, lowering component costs for the whole industry.
The Kylin center is “heterogeneous,” meaning it trains robots with different body shapes and mechanical structures side-by-side. This diversity is rare. Most US labs only contain one type of robot. Training AI on diverse hardware makes the algorithms more robust.
The China model (centralized infrastructure) and the US model (market competition) generate completely different economic benefits. While the US model maximizes corporate value and radical innovation, the Chinese model is designed to maximize manufacturing scale and supply chain dominance.
The Shanghai Kylin model offers many benefits to the Chinese economy compared to the US approach:
1. The “EV Playbook”: Commoditization vs. Value Capture
China is applying the same strategy to robots that it used to dominate the Electric Vehicle (EV) and Solar industries: commoditize the hardware to own the global supply chain. By standardizing data and components through a shared facility, China drives down the unit cost of robots rapidly and significantly.
China controls a vast and integrated hardware supply chain for robotic components (motors, actuators, sensors), allowing for rapid prototyping and mass production at much lower costs. Chinese-made humanoid robots can cost a fraction of their US equivalents.
The strongest evidence of the “China Price” is the Unitree G1, released in 2024/2025 at ~$16,000 USD.
There is currently no US-made humanoid robot you can buy for this price.
-
Agility Robotics Digit: Costs approximately $250,000 (or is leased for ~$30/hour).
-
Boston Dynamics Spot: Even their dog robot (not humanoid) costs ~$74,500 (base), nearly 5x the price of the Chinese humanoid G1.
This positions China to eventually flood the global market with “good enough,” affordable robots, undercutting Western competitors on price. The economic win is export volume and manufacturing jobs.
US companies (like Tesla or Boston Dynamics) focus on proprietary technology to build a “moat.” They aim to sell premium products with high profit margins or recurring software subscriptions (“Robot-as-a-Service”). The economic win is high corporate valuation and wealth creation for shareholders.
2. Solving the “Demographic Crisis” vs. Labor Augmentation
China faces a shrinking workforce and rising wages. The Kylin model acts as a subsidy to accelerate automation adoption for everyone, not just rich companies. By lowering the R&D barrier, small and medium-sized factories can access cheap robots sooner, keeping China’s manufacturing sector competitive despite demographic decline.
The US labor shortage is acute but less existential. US robotics is driven by returns on investment, deploying expensive robots only where they replace high-cost labor (e.g., warehousing, logistics). The economic benefit is increased productivity per worker in high-value sectors.
In China, robots are seen as essential for national development, with less public concern about job displacement than in the US.
The Chinese public is generally more trusting of AI, and broad access to large datasets, facilitated by government policies, accelerates the training of AI models.
3. Data as Public Infrastructure vs. Private Asset
Treating data as a public utility changes who captures the value. The “shared brain” approach prevents wasted effort. This creates a resilient, diverse ecosystem where the industry is the winner, not just one company.
Intense competition forces US companies to push the absolute limits of AI to survive. While inefficient (everyone reinvents the wheel), it often leads to higher peaks of innovation. The winner (e.g., a potential “Google of Robotics”) could capture immense value, but the losers could go bankrupt, leaving the ecosystem more concentrated.
The above analysis is partly based on my queries to Perplexity.ai and Gemini 3.0 and my extensive discussions with David. Here’s how Google’s Gemini LLM concludes our “conversation”:
Neither model is strictly “superior” overall; rather, they excel at different stages of the development cycle. The US is currently winning the “0 to 1” race (innovation and invention), while China is winning the “1 to 100” race (mass production, deployment, and commercialization at scale).
The US winning the “0 to 1” race is very debatable (where exactly?) and, in any case, likely only temporary.
China is winning the important “0 to 1” race in Embodied AI (robots that can think and act). Unitree’s G1 humanoid robot entered mass production in late 2025 with a price tag of ~$16,000. This is the “iPhone moment” for robotics.
These robots are not just demos; they are being deployed in factories (like UBTech’s Walker S2) to generate proprietary physical-world data.
US labs have text data; Chinese labs have physical data. This will create a new “data wall” that the US cannot easily climb because it lacks the manufacturing environments to deploy these robots at scale.
“The 0 to 1 race” is also mainly academic compared with the Chinese approach towards practicability. The “1 to 100” race is much more rewarding for the winner.
