When Chunguang Gu tells the origin story of his robot company, Galaxis Technology, he sketches a “smile curve” in the air with his pointer finger: Nobel laureates on one side, scrappy builders on the other. “Those were the choices I faced,” Gu jokes. “Become an academic heavyweight or try to create tangible value for society.”

He chose the scrappy end. After completing his mechanical engineering PhD at MIT, he returned home to a China in 1999 that was full of industrial promise but short on infrastructure and technology. His story isn’t just about a man building robots; it foreshadows an industry tailor-made for Chinese companies to flourish.

Convinced China would close its industrialization gap, Chunguang joined McKinsey in Shanghai to study business transformation and later moved to drug distributor Jiuzhoutong Pharmaceutical. There he encountered logistics chaos firsthand: costly manual labor, imported automation ill-suited to “China speed,” systems designed for another era.

He built a five-hundred-person engineering team to design something better. “I was right at the intersection of the problem and the solution,” he tells us. By 2009 he’d launched an internal R&D effort—backed by Tsinghua University alumni and corporate partners—that became the seed for Galaxis Technology. “There wasn’t a single robot product that could manage tens of thousands, hundreds of thousands of boxes,” he recalls.

By 2014, an exploding e-commerce sector made that need impossible to ignore. Logistics no longer moved in pallets but in boxes, and millions of them. German and Japanese mini-load systems couldn’t keep pace with about 360 million Chinese consumers demanding delivery now . At JD.com, monthly deliveries jumped from 55 million in 2014 to 133 million two years later and eventually to almost 270 million in their last reported year in 2019. Yes, per month .

Local suppliers gave Chunguang fast access to motors and sensors, and rapid iteration with warehouse partners produced his breakthrough: a multi-directional shuttle robot that could move forward, sideways, and up and down racking systems. Part motorized dolly, part minimalist go-kart, it could move goods at two thousand units per hour—four times legacy systems—and it expanded warehouse capacity by 40 percent because it reached above human height. Another crucial feature was the technology’s adaptability to changing demand: light or irregular workloads might require only a single robot, while higher or surging volumes could be handled by scaling up to a coordinated fleet of robots. Meanwhile, traditional German and Japanese systems struggled outside the “steady middle” of, say, five hundred boxes per hour.

When Galaxis robots launched in 2015, the marketplace was already fiercely competitive. Many start-ups in these strategic areas enjoyed speedy local government approvals and could launch with 1 to 2 million RMB (about $143,000 to $286,000). Competition forced rapid innovation. “Hard for us, but good for capability,” Chunguang says. By the time Galaxis began selling overseas, its systems had been battled-tested in China’s demanding marketplace.

Industrial robots aren’t new—they’ve been used since the 1950s, and about 3.5 million units operate globally with 550,000 more deployed annually—but the emerging sector is finally poised for explosive deployment. Demand for automation is rising, robot hardware keeps getting cheaper, labor shortages are increasing, and new AI capabilities allow robots to learn tasks and adapt to environments that weren’t possible before. China’s landscape is crowded with firms racing to seize the opportunity.

Robotics sit squarely in China’s sweet spot of strengths—and we mean the term “robot” broadly: not just industrial arms, shuttles, and warehouse automation, but the growing class of machines that sense, decide, and act in the physical world, from unmanned drones and vehicles to service robots in restaurants, hospitals, and homes. These are the physical embodiments of artificial intelligence. If top US companies lead in the foundational AI frontier, in the “embodied AI” world cup—where intelligence lives inside machines—Chinese companies have the edge in cost, scale, hardware supply chains, and deployment speed. And most of the factories that might deploy their products are right outside their doors.

The country has more than 450,000 “smart robots” enterprises—triple the number just four years ago— and accounts for roughly three-quarters of global production. The biggest industrial robot makers remain the global incumbents—ABB Robotics of Switzerland/Sweden and Japan’s Epson and FANUC. But the center of gravity for the next generation of robots is shifting to China.

Most initiatives start with a use case in a sector the developers know well. From there, the broader ecosystem lifts them: booming demand, real-world testing grounds, supportive local governments, abundant engineers, fast prototyping, cheap AI, and foreign systems that remain world-class but costly and rigid. The result is a swarm of founders attacking every niche—heavy loads, tight spaces, high-speed throughput, food handling, pharmaceutical testing, e-commerce fulfillment, and more. “Of course, there’s involution in robotics,” says Chunguang, laughing. “Everyone’s sprinting.”

China is now, by far, the world’s largest end market for robots, installing more than half of all industrial robots globally with demand continuing to accelerate at home and abroad. The global market, valued at roughly $50 billion in 2024, is projected to double by the end of the decade.

Increasingly, Chinese firms are designing and deploying robots not only for domestic use but for export markets—creating an ecosystem that both produces and absorbs automation at a large scale. As automation moves beyond cutting-edge factories and into ordinary industries, robots stop being productivity tools and start replacing labor outright, driven by rising costs and a shrinking workforce. In that world, automation isn’t just a nice-to-have; it becomes core infrastructure. This scale may be both China’s competitive advantage as well as a prerequisite for future growth.

Nick caught an early hint of China’s future robotics trajectory a decade ago in a Beijing electronics shop. On a shelf sat one of Xiaomi’s first robot vacuums—a smooth white puck with an orange-rimmed laser sensor. At the time, iRobot dominated Western living rooms, but its Roombas were pricey, battery-hungry, and notorious for getting stuck under furniture. Xiaomi’s version offered longer battery life, smarter navigation, and a friendly price tag—an early sign of how Chinese companies would compete: rapid cycles of improvement paired with cost discipline.

Nick bought a half dozen Mi Robots and carried them abroad. One landed in Switzerland, where his sister-in-law unboxed it. The device whirred to life, introduced itself in Mandarin, and began vacuuming the living room. Unfortunately, Nick’s sister-in-law is fluent in Swiss German, High German, French, Italian, and English, while the Mi Robot could only make pronouncements in Mandarin. “She couldn’t actually understand what it was saying,” Nick recalls, “but she was determined to make it part of the family.”

Her 2016 Mi wasn’t perfect: once it got caught outside in the rain, but it dried itself out and was diligently working again the next day. Even then, while the US robot vacuum remained a luxury curiosity, the Chinese counterparts were smaller, fully functional, ubiquitous, and offered a price point most people could afford.

This story foreshadows what would come: Chinese companies using engineering efficiency to leapfrog incumbents before the rest of the world noticed. Xiaomi would soon apply its electronics expertise to phones, appliances, and eventually vehicles—essentially a four-wheeled robot. Today, Chinese home robots don’t just clean; they pick up socks, climb stairs, and pluck tissues out of corners. What began as a glimpse of promise has grown into a vibrant, hyper-competitive robotics ecosystem with use cases across consumer, industrial, logistics, and healthcare.

Ten years ago, China’s automation systems were built almost entirely on foreign components—ABB and FANUC robots powered by German controllers and Japanese cameras, recalls Cedric Leleu, a McKinsey partner who runs the digital manufacturing transformation practice. Today the landscape has flipped, he says, with Chinese suppliers now accounting for half of the world’s top twenty automation-component makers, with three ranking in the top five. “These local players now offer fully integrated systems—robots, controls, and vision—that are reliable, high-performance, and dramatically cheaper.”

China’s robotics edge isn’t just one advantage but the interplay of many.

To understand China’s robotics advantage, you might start with Huaqiangbei Road in the southern city of Shenzhen. Often called “China’s electronics street” or “the Silicon Valley of hardware,” it’s less a Sand Hill Road than a multi-block labyrinth of towers and alleys housing the world’s largest electronics market. Thousands of tiny stalls sell everything a robot maker might need: actuators, torque sensors, reducers, control boards, casings, circuit boards, computer numerical control (CNC) machined parts.

Huaqiangbei is the kind of place where an engineer can walk in with an idea and walk out with the components—and even a consultant’s expertise—to turn it into a working machine. Founders describe riding escalators through seven-story buildings—sometimes with a couple of engineering classmates—moving among more than three thousand vendors, negotiating part by part. The talks might soon shift to WeChat; regardless, custom pieces can be fabricated in hours, not weeks, at a fraction of Western prices.

One shop might harvest chips from discarded phones and laptops and sell them by the gram. Another offers server hardware at half of branded server prices and can assemble a team to build a robot prototype in under two weeks. Yet another can customize humanoids, teaching them to tap-dance or speak German. With a few hundred dollars, vendors can assemble a kid-sized robot prototype in a matter of days.

This ecosystem has launched dozens of hardware champions. RoboSense, headquartered nearby, became the world’s largest lidar maker in under a decade. Founder Chunxin Qiu started with 800,000 RMB (about $114,000), an amount that in Silicon Valley might cover an office’s annual coffee budget, he jokes. “Here a few hundred RMB gets me a hand-polished prototype in a day,” he says.

In Huaqiangbei, Chunxin learned by doing. “I’d explain the torque requirements, and the seller would explain the parameters. I learned while sourcing from all the vendors—it’s very convenient.” For computer numerical control (CNC) parts, he’d revise drawings with a craftsman for “a few dozen RMB” ($5 to $6), and “wait downstairs” for the finished pieces. If they were complicated? He’d need to come back after lunch. “In this way, we could iterate through several generations in two to three months,” Chunxin says.

This ecosystem allows founders to stay lean, as suppliers become extensions of the team. “The chain is so complete that two or three people can start a business,” Chunxin says. “You don’t need to master everything—you can outsource most tasks.” Small factories can handle machining, printed circuit boards, and even final assembly, and a founder can order runs of ten or a hundred units—ideal for rapid demo and proof-of-concept validation.

When RoboSense scaled up, this supply chain expanded with it, partnering with companies that can mobilize any fraction of their technicians for cyclical production.

The same conditions that make this ecosystem so accessible have also pushed the system toward involution. Walk into any Chinese tech expo, and something is always whirring, blinking, staggering, or singing. Companies are building machines to teach children, deliver food, play music, patrol warehouses, clean kitchens, fold laundry; if humans do it, someone in China’s robotics ecosystem is trying to automate it.

This isn’t just start-ups; also crowding in are the mobility divisions of internet giants, logistics companies, and even EV makers. “The future world of mobility won’t just be vehicles; it will be multiple mobility solutions,” says Brian Gu, vice chairman and president of EV maker XPENG. “We’re excited about the robotics space—especially humanoid robotics—where applications could be five times bigger than the automotive space.” XPENG’s leaders describe autonomous driving as “robotics on wheels”—perception, motion, decision—and recently unveiled a humanoid robot that can strut a fashion catwalk.

This sense of piling in is everywhere. “Is there excess competition? Maybe,” one founder says, shrugging. “If you look at the start-ups and the robotics Olympics—there’s certainly a lot of immature companies—and it’s hard to see them all succeeding.”

Capital isn’t slowing down, either. Venture funding for general-purpose robotics globally has surged, tracking the bet that as AI improves, so will the robots built around the models. China has layered on top a $138 billion national innovation fund over the next twenty years for robotics and embodied AI. Private giants—from SF Express to BYD—are investing aggressively in their own robotics teams.

When Joe spoke at a Shanghai robotics and GenAI conference in winter 2025, the real action wasn’t in the lecture halls—it was in the robotics exhibition pavilion. Bright, airy, and smelling faintly of warm plastic, the hall was packed with robots in various states of movement: some moving, some frozen. Start-ups squeezed in next to veterans like Unitree Robotics, whose robot dogs and humanoids are already in global university labs.

The scenes bordered on comical. A “dancing cowboy” wearing a red bandanna and blue vest stumbled through a hip-hop routine, then walked itself back to its recharging rack. A handler kept an orange rope looped around “John Wayne” the entire time, not trusting him to stay upright. Nearby, a gray robot head with lifelike eyes sat mute; “Sing,” then “Sing NOW” produced nothing. A tall white humanoid raised its arms in surrender, again and again, but missed every third command. A motorcycle-sized velociraptor lurched forward, speakers along its spine booming synthetic roars—impressive if slightly terrifying. What’s its use case? No one knew, though a sales rep gamely suggested “tourism… maybe movie sets.” Teams studied competitors’ ideas and coaxed passersby into offering feedback. To Western or Japanese eyes, it looks chaotic. To Chinese founders, this is how you learn: ship half-working robots, collect data, benchmark rivals, and iterate.

Talent keeps the flywheel spinning. China’s robot builders range from scrappy hardware tinkerers to elite software engineers—many grew up fixing scooters or building drones in college and now prototype robots in forty-eight-hour sprints. As we’ve mentioned before, China’s millions of STEM graduates create an engineering base larger than that of any other country, alongside the electricians, robot technicians, and senior experts who design core systems. “Layer on top the energy and adaptability of young engineers—willing to test, fail, and test again,” says Chunguang of Galaxis Technology, “and you get an ecosystem built for massive acceleration.”

Galaxis’s early shuttle robots had no competitors, but rivals soon crowded in, forcing Chunguang to “launch faster, refine more, and cut costs.” He likens China’s robotics ecosystem to “rich soil that’s given birth to a pack of wolves—small, fast, and combat-effective, with strong survival abilities.” By contrast, robotics players in the US and Europe are more consolidated, with “a few large elephants.”

The wolf-pack model has limits. Chinese firms file more robotics patents than any other country, but that’s a metric for capacity, not quality. Many fundamental breakthroughs—tactile sensing, dexterous manipulation—still emerge from US, Japanese, or European labs. Of 450,000 registered “smart robot” enterprises, only a few thousand actually manufacture robots.

Moreover, the sheer proliferation of firms makes it difficult for Chinese robotics companies to scale. As Chunguang puts it, “Every robotics firm is chasing a very specific use case. For that context, they specialize and tailor-make the robot that is a perfect fit for purpose. They price aggressively, survive on thin margins, and in doing so make it hard for others to enter. That, in turn, forces competitors to seek out ever more niche applications.” For all the dynamism in the sector, this fragmentation has real limits. Among the top ten robotics companies by revenue, none is Chinese—with the exception of Germany’s Kuka, whose owner is the Chinese multinational Midea Group.

As margins compress and intelligence becomes cheap and interchangeable, “we have the best model—pay us accordingly” becomes harder to defend. Thus, whatever durable payoff exists is more likely to shift into the hardware layer—where AI meets the physical world.

Already most Chinese firms have abandoned model development to focus on applications, echoing the internet browser era, when HTML commoditized and value migrated upward to apps and services.

Similarly, we expect the frontier to move beyond software and into “embodied AI”—robots, drones, agentic devices, smart sensors, and vehicles that use intelligence to act in the physical world.

This is where China’s structural strengths matter most. The country produces roughly three-quarters of the world’s lithium-ion batteries, and it accounts for roughly 90 percent of global rare-earth magnet production—including the neodymium magnets widely used in EV motors—components that underpin embodied AI by powering mobility and enabling precise, high-torque motion. China also maintains scale advantages in power electronics, embedded compute, and industrial automation.

“If intelligence is effectively free,” says tech entrepreneur Jason Chiu, “the real competition shifts toward manufacturing excellence at scale—industrial robots on factory floors, machines that understand homes, drones that navigate dense cities, vehicles that coordinate with infrastructure. In that world, the embodied system becomes the differentiator.” Intelligence alone is inert, say technology analysts such as Packy McCormick; value is created when it’s paired with energy, infrastructure, and the ability to act in the physical world.

Ever-cheaper models are a gift to embodied AI entrepreneurs, who can plug into whichever foundation model best fits their needs. But for model providers, the economics turn harsh: overhead erodes margins faster than small capability gains can restore them.

If China’s robotics ambitions and constraints can be captured in one product, it might be humanoids: glamorous, policy-favored, technically challenging, and financially brutal. Hollywood has long imagined humanoids: C-3PO bumbling in gold, the Terminator stalking in leather, Ex Machina ’s Ava gazing with eerie calm. People have always been obsessed with the human form, which has steered roboticists toward a humanoid look that unfortunately imposes constraints, making it the least effective form of robot. We haven’t met a Chinese robot entrepreneur who hasn’t thought about attempting one.

Humanoids promise versatility; shaped like us, they can, in theory, use our tools, fit through our doors, and operate in our world without costly re-designs of our physical environment. In an aging China facing mounting labor shortages, the appeal is obvious—machines that can lift patients in a nursing home or handle hazardous chemicals without needing lunch, sleep, or a salary. Policymakers tout humanoids as the next “disruptive product” after computers, smartphones, and new-energy vehicles, and investors are piling into the next sector with policy support, and global potential.

This fascination has hardened into an industry, but in 2012, creating one meant inventing the future with primitive technology. UBTech Robotics founder Jian Zhou’s earliest prototypes were palm-sized toys: “We created them at such an early stage that the Wright brothers had better technology when they started on planes. It was like wanting to make a car but starting with a bicycle.” He scavenged parts in Shenzhen’s electronics marketplace and waited for advances in flexible joints, sensors, and control software that would make possible stable movement.

At UBTech’s Shenzhen headquarters, videos loop of robots marching up stairs, hiking uneven ground, swapping their own batteries, and working on Geely’s assembly lines. The company now has more than two thousand employees, about half of them engineers, and its Tiangong robot won the world’s first humanoid marathon—of course, this originated in China—its metallic stride going viral.

Costs are falling fast. Industry data show humanoid bill-of-materials costs falling about 20 percent annually between 2022 and 2024. With more than two hundred Chinese companies attempting humanoids, pricing pressure is fierce: a UBTech model once priced at $100,000 will sell for under $50,000; Unitree pushed prices even lower in 2025 with a $6,000 humanoid, down from $16,000 a year earlier, marketing it as a general-purpose platform for developers, researchers, and enthusiasts.

Still, humanoids carry deep structural disadvantages. Lots of motors and sensors make them power-hungry, limiting industrial runtime. How useful are legs if robots must stay plugged in? Wheels are simpler and stabler, and current wheeled versions already have the ability to climb stairs. Meanwhile, humanoid hands still struggle with the finger-bending range of motion and tactile sensing to reliably grip things.

Height adds risk: to see like humans, they must be tall, meaning heavy actuators that raise the center of gravity. In eldercare or childcare, a falling seventy-kilogram (155-pound) machine is no benign helper. Yet designing for safety limits strength. “They end up being able to lift what a twelve-year-old can,” says Ani Kelkar, McKinsey Boston-based partner and advanced industries practice leader. Speed is another constraint: parkour they can do, but when it comes to “doing a useful task like picking up a screw, humanoids remain painfully slow—it’s like watching paint dry,” Ani says.

Economically, the trade-offs are stark: add power, speed, and dexterity, and costs rise. Dial them back, and usefulness suffers: “You might conclude you didn’t need a humanoid—you just needed a dexterous arm on a wheeled platform,” Ani says. It’s not a surprise that humanoid and service robot firms routinely post the sector’s worst losses, but it’s early days, explains Ani. “We hope to see much more technical progress—only then will the hype be realized.”

Humanoids struggle—as do most robots. Many manufacturing sectors, especially consumer electronics, still rely on human labor because product variation is high. Current robots lack the mechanical flexibility and adaptive control to switch from, say, assembling a MacBook Air and then a MacBook Pro without prohibitive changes, says Michael Chang, a McKinsey partner in the AI practice. “We’re getting closer, but it’s still three to four years away, and that timeline keeps shrinking. The hardest part is the mechanics. Software will move faster, because better foundation models make it easier to teach robots new assembly processes.”

For most things, robots aren’t better and cheaper than humans—yet.

China’s robotics advantage is reinforced by a powerful data flywheel, created by pairing hardware with a vast factory base and deep engineering talent. Unlike large language models, robots can’t be trained on text alone. They learn by entering the real world—failing, adjusting, improving, and trying again—reinforced by virtual simulation, human-guided control, and task demonstrations.

Companies that can tap into China’s ecosystems can run this loop faster, cheaper, and at far greater scale than anywhere else. That uptake only creates more data, Ani explains. “The cheaper you are at the beginning, the earlier the foot in the door with deployment, the more data you collect, the better your models become,” he says. “You’re not going to get real-world data if your hardware costs $250,000.”

In China, this cycle is turbocharged by proximity. Factories, logistics, and warehouses sit just outside roboticists’ doors. Deployments remain small-scale, but they’re growing in number, and they force technological leaps, says Shiquan Wang, co-founder and CEO of adaptive industrial robot company Flexiv Robotics, which made an early bet on using “force control”—programming robots to regulate and adapt during interactions with the world around them—over pre-planning rigid trajectories. “When we put our products in the field, we need engineers to use it, and partners to develop our applications—and we get a lot of insights and details,” Shiquan tells us. Flexiv’s office locations are strategic: R&D teams at its US headquarters carry out early-stage research, while engineers in Shanghai, Beijing, and Shenzhen work alongside manufacturing clients, says Shiquan. “Thousands of trivial things will make the product really different.”

There are plenty of these fast-loop iterations available; in short, Chinese manufacturers, squeezed by labor shortages and fierce competition that push them to find cost efficiencies, are gobbling up robots. For twelve consecutive years, the country has been the world’s largest industrial robot market, installing roughly 295,000 units in 2024—and the pace of installation growth is accelerating across China while rates in Europe and the Americas decline. Most new robots roll straight onto Chinese factory floors in the top five hungriest sectors for automation: automotive, electronics, metals and machinery, plastics, and chemicals and food.

More than two million robots already work the lines, and some factories are nearly fully automated, such as those that make power batteries— pushing the country’s robot density past that of Germany, the longtime leader in industrial automation. Large logistics facilities increasingly operate with minimal human presence, and “dark warehouses” are no longer edge cases. If progress continues, the general-purpose robotics market could reach a value of $370 billion by 2040, with China accounting for about half.

Why is this demand snowballing so quickly? Scale. China generates one-third of the world’s manufacturing economic value; more production means more potential places to deploy robots. At the same time, demographic change is turning robotics from novelty to necessity. Policy directives explicitly link automation to eldercare, healthcare, and agricultural modernization, among others, as China’s workforce shrinks and its population ages. More than 310 million Chinese are over age sixty. A decade ago, a 30 percent pay bump might attract workers to hazardous jobs such as industrial glue application involving formaldehyde, says UBTech’s CFO, Ju Zhang. “Now even a 300 percent increase won’t attract workers, and young people might even sue the company. Robots aren’t here to be more efficient than humans—they’re here to fill labor gaps,” says Ju. “Companies can’t find workers.”

Those labor gaps are already reshaping service and mobility sectors as well: coffee chains with no baristas; ride-hailing platforms deploying driverless robo-taxis; residential towers where robots bring packages from lobbies to apartment doors; hospitals where robots handle routine transport and sanitation; ports where cranes load ships with minimal human crews; and public spaces maintained by machines working through the night.

China’s market structure accelerates uptake: compared with counterparts in Germany or the US, Chinese firms don’t have to face unions or regulatory challenges; in fact, the government is actively encouraging the deployment of AI and robots. Chinese leaders are “much more open to innovation and new technology,” says Cedric, who leads McKinsey’s digital transformation services practice. “Sometimes ambition drives them to build fully automated factories, even if it’s not justified financially short term. The vision is to be the most advanced.”

China is poised to produce the next generation of global robotics leaders: its robotics ecosystem sits at the intersection of what Chinese companies do well. Whether and when the market reaches hundreds of billions or a trillion dollars is uncertain, but what’s clear is that Chinese robots are going global.

That doesn’t mean Chinese players will lead in every dimension. Some advanced components—especially sensors and semiconductors—come from German and Japanese champions, and the most advanced humanoid and mobility research remains concentrated in US labs such as Boston Dynamics, Tesla, Agility, and Figure. Further industry consolidation is also required for Chinese players to gain scale and financial muscle. This may take years. Further, Chinese robots might at some point face growing geopolitical and security headwinds—export controls, trust concerns in sensitive sectors, and the emergence of two partially split tech ecosystems built on different fundamentals—that limit how widely they can be adopted globally.

Yet for now, the Chinese industrial environment already contains most of the world’s real end-use cases for robotics: factories, warehouses, logistics hubs, ports, and production lines where machines can be deployed at scale, instrumented continuously, and improved through repetition. These environments generate data, learning cycles, and economic payoff, allowing Chinese robotics firms to lower costs quickly and accelerate deployment. In practice, embedding AI inside structured production lines has proven far easier than applying it to white-collar work, where tasks are diffuse and outcomes harder to measure. In this domain, global companies risk playing catch-up.

For companies trying to understand our robotic future, there’s no substitute for engaging with China’s robotics cluster—in which its companies are increasingly eager to export, partner, and scale globally. In early 2026 in Las Vegas at CES, the world’s largest and most influential technology trade show, about a thousand Chinese robot, drone, and consumer electronics players sought partners, customers, and routes to global scale. In panel discussions, we observed a growing consensus that the trajectory of “physical AI” will hinge less on technical possibility and more on economics. “In a match between the Chinese wolves and Western elephants, what will the future be like?” says Chunguang of Galaxis Technology—which, about a decade after launching its first shuttle robots, listed on the Hong Kong Stock Exchange in March 2026.