The Backflip Was Never the Point
Somewhere in a stadium in China this past weekend, a robot with a humanlike torso threw a punch, sprinted a lap and marched in formation at the second edition of the World Humanoid Games. The footage travelled fast, as footage of machines doing human things always does. But not far away, in an unremarkable industrial park in Chengdu, a different kind of robot was doing something far less photogenic: scanning a QR code, over and over, while a young engineer nudged it across the floor and hoped it might one day learn to do more.
That contrast between the spectacle built for cameras and the plodding, unglamorous work happening in factories and vocational classrooms is the real story of where China’s automation drive is headed. According to the BBC’s on-the-ground reporting, the robots quietly reshaping Chinese industry do not dance or leap on stage. They weld, paint, lift and assemble around the clock, and some are even built to construct other robots. This is the part of the robot economy that will not trend on social media, and it is also the part that will most directly reshape jobs, factories and supply chains over the next decade.
What Counts as a Robot Revolution, and What Doesn’t
It helps to separate two very different categories that get blurred together in headlines. Humanoid robots are general purpose machines built to move through spaces designed for people, walking on two legs, climbing stairs, using human tools. They are still mostly experimental, expensive and limited to narrow demonstration tasks. Industrial robots are the opposite: fixed or wheeled machines, often just an arm or a cart, purpose built for one repeatable job on a factory floor. They are unglamorous, but they are already doing the work at industrial scale.
The numbers make the gap obvious. The International Federation of Robotics’ World Robotics 2025 report found that China installed 295,000 industrial robots in 2024, the highest annual total any single country has ever recorded, representing 54 percent of everything deployed worldwide that year. The IFR’s global press release notes that, for the first time, Chinese manufacturers sold more industrial robots domestically than foreign suppliers did, a milestone marking the point where China stopped merely buying automation and started building most of its own.
The IFR’s China-focused briefing puts the country’s total operational stock at around two million industrial robots, roughly four and a half times more than Japan, which sits in second place. That stock did not accumulate through one dramatic announcement. It built up arm by arm, welding station by welding station, across nine years of steady deployment by companies most people outside the industry have never heard of.
Inside the Factories Where the Real Transition Is Happening
One of those companies is CRP Technology in Chengdu, which has spent nine years building robotic arms for factories that make cars, electronics and wind turbines. Per the BBC’s factory floor reporting, the firm says a single arm can be programmed to perform over 90 percent of a given repetitive task, and it employs around 300 people building and refining that hardware. On the assembly line, roughly 30 middle-aged workers screw, drill and polish the company’s robotic arms while, in the same room, younger engineers work on a prototype robot intended to help build other robots.
That detail, robots built to build robots, is worth pausing on. It signals a shift from automation as a tool that replaces individual tasks to automation as a self-reinforcing production system, where the machines making cars are themselves increasingly made by machines. One engineer on the research team put the ambition plainly: the goal is a robot that can think for itself, because in roughly a decade China may simply not have enough workers to fill its factories.
At Leapmotor, an electric vehicle maker whose budget-friendly cars have become increasingly popular in Europe and the UK, the picture is further along. A company vice-president told the BBC that stamping, welding and painting have essentially reached full automation, and that a future without human workers on those lines is entirely possible and could arrive relatively quickly. Even so, Leapmotor still employs more than 25,000 people, many of whom perform final quality checks as vehicles roll off the line, a reminder that automation tends to move task by task rather than eliminating entire workforces overnight.
Why Beijing Is in a Hurry: The Demographic Clock
China’s push into automation is not simply an industrial ambition. It is increasingly a demographic necessity. Research from the Rhodium Group estimates that China will lose close to 60 million people between 2026 and 2035 alone, roughly the population of France, with the annual rate of decline widening as the decade goes on.
The workforce implications are stark. A Pew Research Center analysis shows China is approaching what state media call a “moderately aging” society, with 20 percent of the population already 60 or older, a share expected to climb to 30 percent, or more than 400 million people, by 2035. Coverage from ABC News Australia notes that 400 million figure is roughly equal to the combined populations of the United States and Italy, meaning hundreds of millions of people will be leaving the workforce just as pension systems come under increasing strain.
Against that backdrop, automation stops looking like a choice made purely for efficiency and starts looking like an attempt to keep the factory floor staffed at all in the coming decades. Around 120 million people still work in China’s manufacturing sector today, and the central risk officials face is timing: move too slowly and the labour shortfall becomes an economic drag; move too quickly and job losses hit millions of workers who currently depend on factory wages.
Robot Density and What It Actually Measures
Total robot counts can be misleading for a country as large as China, since a bigger economy simply has more factories to fill with machines. Robot density, the number of robots per 10,000 manufacturing employees, is the metric researchers use to compare automation intensity across countries regardless of size.
By that measure, China’s rise has been unusually fast. Coverage from Business Wire’s report on the 2022 IFR data shows China’s robot density stood at 322 units per 10,000 employees in 2021, putting it fifth globally. By 2023, according to the South China Morning Post, that figure had climbed to 470, pushing China past Germany and Japan into third place worldwide, behind only South Korea and Singapore. The Robot Report’s analysis notes China achieved that density despite still employing around 37 million manufacturing workers, a scale no other country in the top ranks comes close to matching. The most recent 2025 IFR data, summarized in an April 2026 press release, show South Korea still holding the global lead with 1,220 robots per 10,000 employees, a position built on its concentrated electronics and automotive industries.
The comparison to South Korea and Japan is instructive precisely because those countries automated their factories decades ago for almost identical reasons: shrinking, ageing workforces and a need to protect export competitiveness. China is now compressing that same transition into a fraction of the time, using a domestic supply chain that did not exist for Japan or Korea when they went through it.
The Advantage Nobody Can Easily Copy: Vertical Supply Chains
Robot density explains what China has achieved. Supply chain geography explains why it has been able to achieve it so quickly. According to the BBC’s interview with Dr Susanne Bieller, General Secretary of the International Federation of Robotics, sourcing every component needed to build a robot in Shenzhen, from motors to sensors, can take under an hour, while the same process can take up to a week in Europe.
That density of suppliers sitting next to each other is not an accident. It is the byproduct of two decades spent building China into the world’s manufacturing base for consumer electronics, electric vehicles and industrial machinery. The same factories and parts ecosystems that produce smartphones and EV batteries can pivot to produce robot components, because much of the underlying hardware overlaps directly.
Bieller also flagged a genuine gap on the other side of the ledger: China remains ahead on building the physical body of a robot, but the United States still leads on building the “brain,” the artificial intelligence software that lets a robot interpret its environment and decide what to do. That distinction between hardware and software leadership is likely to define the next phase of global competition in robotics more than any stadium demonstration will.
Open Source AI Is Closing the Software Gap Faster Than Expected
The AI side of that gap has been narrowing quickly. Chinese firms including DeepSeek and Moonshot have released AI models their developers say can compete with leading American systems, and much of that progress has come through an open source approach in which start-ups share their underlying code rather than keeping it proprietary. That collaborative model has let Chinese AI development catch up faster than many analysts expected, even as US export controls have restricted access to the most advanced chips.
The strategic implication reaches beyond chatbots and language models. Analysts increasingly believe the next phase of the AI race will not be decided by which company has the single best language model, but by which country can embed that intelligence into millions of physical machines. In other words, the contest is shifting from software running in data centres to software controlling arms, carts and eventually humanoid bodies on factory floors, and China’s hardware manufacturing advantage gives it a natural runway to compete in that race even without an outright lead in raw model capability.
How Vocational Schools Are Building the Workforce That Runs the Robots
Automation does not eliminate the need for workers so much as it changes what kind of workers are needed, and China is trying to engineer that transition directly through its education system. Per the BBC’s visit to the Hangzhou Technician Institute, one of several large, purpose-built vocational schools, students can begin training in robotics from the age of 15, with entire buildings dedicated to subjects such as aeronautics and a restricted floor devoted to processing rare earths, another material central to the standoff with Washington.
A 28-year-old instructor at the institute spends his own spare time visiting factories that are converting to automation, so he knows what to prioritise when teaching students who are learning to programme industrial robots and, in some cases, develop robotic medical instruments. His view of the stakes is blunt: he tells students that anyone not needed by the world of AI and robotics will eventually be eliminated from the workforce.
The institute is partly positioned as an answer to a separate but related problem. With youth unemployment running at roughly one in five young jobseekers struggling to find work, the instructor says the institute’s graduates are already in demand and can be pre-booked by companies before they finish training. Whether that pipeline can scale fast enough to absorb workers displaced from more traditional manufacturing roles remains one of the least certain parts of the whole transition.
What the Global Research Says About Jobs Lost and Jobs Created
China’s experience is not happening in a vacuum, and researchers have been modelling the employment effects of industrial robots worldwide for years. An Oxford Economics study estimated that industrial robots could displace around 20 million manufacturing jobs globally by 2030, equivalent to about 8.5 percent of the world’s manufacturing workforce, with roughly 14 million of those robots expected to be deployed in China alone, according to coverage of the same report.
The same body of research complicates any simple narrative of robots destroying jobs outright. An American Enterprise Institute analysis of the underlying Oxford Economics data found that, since 2000, industrial robots have displaced about 1.7 million manufacturing workers worldwide, with each robot replacing roughly 1.6 human workers on average. China has absorbed the largest share of that displacement, around 550,000 workers, more than twice the toll seen in the United States. At the same time, the researchers found automation tends to create new jobs at close to the same rate it destroys old ones, through a productivity effect they call a “robotics dividend.”
The catch is distributional rather than aggregate. The same research found job losses fall disproportionately on lower-skilled workers and poorer regions, with workers in lower-income areas losing jobs to automation at roughly twice the rate of workers in higher-income areas, largely because production tasks are easier to automate than the management and oversight tasks concentrated in wealthier regions. For a country attempting to automate 120 million manufacturing jobs while its population ages, that unevenness is not an abstract academic finding. It is the central political risk of the entire strategy.
The Policy Behind the Machines: “New Productive Forces”
None of this automation drive is happening organically. It sits inside a specific and deliberately named government strategy. According to background on the policy’s origins, Xi Jinping first introduced the phrase “new productive forces” during an inspection tour of Heilongjiang province in September 2023, calling for an economic model built on innovation in advanced sectors. Officials describe the concept, as reported by Xinhua’s explainer on the term, as advanced productivity freed from China’s older, more labour and land intensive growth model, built instead on high technology, high efficiency and high quality output.
The phrase draws on a much older idea. As the South China Morning Post notes, “productive forces” is a foundational concept in Marxist economic theory, referring to the combination of human labour with technology and infrastructure. What is new is the specific application to robotics, AI and advanced manufacturing as the vehicles for that idea, formally written into the outline of China’s 15th Five Year Plan, which the IFR’s president has described, per its China strategy briefing, as the primary framework document setting direction for government action on robotics and automation.
The effects of that policy push are already measurable in the supply chain. The same IFR briefing reports that the share of Chinese-made robots among domestic industrial robot installations rose from 30 percent in 2020 to 57 percent in 2024, and in the metal and machinery industry specifically, Chinese suppliers now hold an 85 percent share of their own domestic market. That is the quiet, unglamorous outcome of a state-directed industrial strategy playing out roughly on schedule.
What This Means Beyond China’s Borders
For manufacturers elsewhere, the practical question is not whether China will keep automating, that trajectory looks locked in, but how fast the resulting cost and speed advantages spread into export markets. China already ships industrial robots across the globe, and more than half of the world’s industrial robots are now made there, meaning countries that import Chinese manufacturing equipment are, in effect, importing China’s automation advantage into their own supply chains.
For businesses and workers outside China, the relevant lesson from the IFR data is less about any single robot and more about the compounding effect of vertically integrated supply chains. A factory in Shenzhen sourcing every component within an hour is not a story about robots. It is a story about decades of industrial policy creating a manufacturing ecosystem that is now difficult for any competitor to replicate quickly, regardless of how much capital they commit to catching up.
For workers and students, the more transferable lesson sits in the vocational classroom rather than on the humanoid stadium stage. The skills in demand are not the ability to compete with a robot at doing a repetitive task, but the ability to design, programme, maintain and improve the systems that make repetitive tasks obsolete. That shift in what counts as valuable labour is likely to unfold in every economy that automates, not just China’s, even if the pace and social safety nets available to manage it differ sharply from place to place.
What Remains Genuinely Uncertain
Several important questions in this story do not yet have confirmed answers, and it is worth being explicit about what is established fact versus what is still analysis or forecast. It is confirmed that China leads the world in total industrial robot installations and stock, and that its robot density has risen faster than any other major economy’s in recent years. It is confirmed that its population is ageing and shrinking, and that government policy explicitly links automation to that demographic pressure.
What is not yet confirmed is how China’s information environment will report the social costs of this transition as it accelerates, a concern the BBC’s reporting raised directly. It is also uncertain whether China’s vocational training pipeline can retrain displaced workers fast enough to prevent the kind of regional job losses that Oxford Economics’ modelling suggests are likely, or whether wage pressure in newly automated regions will follow the pattern seen in other countries where automation has proceeded quickly. And whether China’s current lead in humanoid robotics theatre will translate into a genuine lead in general-purpose robot intelligence, closing the “brain” gap that even Chinese industry figures acknowledge still favours the United States, remains an open and consequential question for the rest of the decade.
What is clear is that the backflips and boxing matches were never really the point. The transformation worth watching is the one happening quietly, arm by arm, classroom by classroom, in factories most people will never visit.
