Humanoid robots field on the World Robotics Convention in Beijing. Credit score: Georg Stieler
The newest knowledge from the Worldwide Federation of Robotics signifies that, out of roughly 520,000 industrial robots put in worldwide in 2024, 54% went to China, 17% to the European Union, 8% to Japan, and seven% to the U.S. Whereas China continued to develop from an already excessive base, shipments declined within the different three areas.
Robotic density, a key indicator of a nation’s automation stage, tells an analogous story. Twelve years in the past, the U.S. charge was virtually 5 instances China’s. Since then, the nation has overtaken the U.S. and is now virtually than twice its stage.
Motivated partially by the prospect that good, linked manufacturing might set off reshoring to early‑industrialized economies, no nation has automated its factories as decisively as China.
China scales automation trade
Backed by state coverage and a willingness to just accept quick‑time period ache to achieve technological management, China has constructed a extremely aggressive industrial ecosystem. Home distributors’ share of the industrial robotic market has risen from below 30% a decade in the past to greater than 50% at the moment. In collaborative and cell robots, native suppliers maintain about 90%.
The nation now units the tempo in electrical automobiles (EVs), batteries, photovoltaics, and drones; software know‑how is more and more flowing from China to Europe. In autonomous driving, China is working neck and neck with the U.S. Greater than half of the publicly listed firms within the humanoid robotic provide chain are Chinese language, estimates J.P. Morgan.
In mechatronics, nobody brings new merchandise to market sooner, at top quality and aggressive costs. Chinese language industrial robots are roughly one‑third cheaper than these from Europe or Japan. Robotic exports have grown about 65% yearly since 2022.
The features attain past native manufacturers: a considerable portion of the {hardware} for Tesla’s Optimus humanoid is reportedly sourced from China. To remain aggressive, German automation corporations are increasing analysis and growth in China. One latest instance is STIHL’s determination to maneuver growth of its robotic lawnmowers to China.
China has invested in STEM training. Credit score: Georg Stieler
Expertise pool and AI present tailwinds
China’s emphasis on science, expertise, engineering, and arithmetic (STEM) training is paying off. The nation graduates about 3.5 instances as many science majors because the EU and roughly 4.5 instances as many because the U.S.
At neural data processing methods convention NeurIPS final 12 months, U.S. establishments nonetheless produced probably the most papers, however students of Chinese language origin fashioned the one largest group, underscoring the depth of the expertise pool.
Distributors akin to DeepSeek are setting new markers with succesful, useful resource‑environment friendly open‑supply giant language fashions (LLMs), enabling purposes that beforehand lacked a enterprise case.
As The Economist reported final month, 80% of startups pitching at Andreessen Horowitz have been utilizing a Chinese language open mannequin as an alternative of a U.S. one. In robotics, entry to low‑value {hardware} encourages experimentation and opens new use instances.
Semiconductors and state assist
A key weak point stays the scarcity of excessive‑efficiency home chips, although progress towards self‑reliance is arriving sooner than anticipated. U.S. sanctions have created new alternatives alongside provide chains.
On the similar time, demand for compute is surging. Fashions that seize the bodily world are much more advanced than LLMs or simple imaginative and prescient‑language fashions (VLMs). Huawei is working to place itself as a nexus on this stack.
China applies EV playbook to bodily AI
After a number of financially lean years for the Chinese language tech sector — with a drastic drop in enterprise capital and a document low in IPOs — Beijing is making an attempt to reinvigorate the sphere. In February, Xi Jinping, president of the Folks’s Republic of China, met with main tech entrepreneurs. Alibaba’s Jack Ma was publicly rehabilitated. Corporations akin to DeepSeek and Unitree stand for a brand new dynamism.
Humanoid robots and “bodily AI” are the most well liked sector. By the tip of Might 2025, funding in robotics and embodied intelligence exceeded the total‑12 months 2024 whole.
That tempo accelerated in Q2 2025. By July, we calculate that China invested $3.4 billion in new robotics ventures — 42% greater than the U.S. and 5 instances that of Europe. Enterprise funds are again, giant tech firms akin to Meituan and Alibaba are lively once more, and state‑backed funds are co‑investing alongside them.
Chinese language humanoids transfer to mass manufacturing
Chinese language corporations are surging forward towards scale. The query for all humanoid firms is whether or not they can ship robots at quantity and combine them usefully into actual situations.
Based on a report earlier this 12 months, six of the nation’s 11 main humanoid‑robotic producers, together with Unitree, AgiBot, Galbot, Engine AI, and Leju Robotics, launched mass‑manufacturing initiatives in 2024, aiming to supply over 1,000 items per 12 months by the tip of 2025.
Collectively, Chinese language corporations are projected to fabricate greater than 10,000 humanoid robots in 2025, accounting for over half of worldwide output. UBTECH, AgiBot, and Unitree captured roughly 60% of the home market within the first half of this 12 months. Sarcastically, there may be nonetheless substantial handbook labor in key parts for small‑batch humanoids.
Can humanoids combine into actual, helpful situations?
Regardless of the optimism and capital inflows, many consultants warning that humanoid robotics are usually not but mature. Vital technical and industrial hurdles stay within the close to time period.
Current procurement offers are notable. In June 2025, a China Cell-affiliated platform awarded a CN¥124 million ($17.4 million U.S.) procurement cut up between AgiBot and Unitree. In July, UBTECH gained a ¥90.5 million ($12.7 million) order from a Shanghai‑based mostly EV startup. It was the most important identified single‑firm humanoid robotic order worldwide at the moment.
Noetix disclosed 2,000+ unit orders in H1 2025 (price over ¥100 million or $14 million), primarily from Okay‑12 faculties, universities, and vocational institutes.
As of mid‑2025, the overwhelming majority of humanoids in China are prototypes or pilot items, not mature mass‑produced merchandise. A big portion of deployments are demonstrations, performances, or knowledge assortment quite than mission‑important duties, with deliveries scheduled over the following two to a few years.
Commercialization timelines (trade steerage)
Unitree stated it expects seen features in humanoid intelligence inside the subsequent one to 2 years. It stated actual commercialization will start inside three to 5 years, probably first in public service or choose manufacturing duties.
Inovance, the most important industrial automation firm in China, is getting ready for a viable market by 2030.
Investments shift towards VLA fashions
Attendance was robust at China’s World Robotics Convention. Credit score: Georg Stieler
The essential query is about software program. When will versatile, semi‑autonomous robots turn into actuality? The present development targets bodily AI plus movement management “moats” and favors higher‑physique humanoids on wheeled bases – two arms, not essentially legs.
Capital is gravitating to the “brains.” Almost each July deal earmarked greater than 40% for knowledge‑engine technology and enormous embodied‑mannequin coaching.
Imaginative and prescient-language-action (VLA) fashions from Spirit AI, XSquare, and Galaxea demonstrated job suites akin to fabric folding, drivetrain meeting, mattress‑making, and different dealing with duties. Whereas these methods should still path Western leaders like Google Gemini Robotics or Bodily Intelligence in breadth of generalization, knowledge variety, and benchmarked reasoning depth, the Chinese language corporations are transferring sooner in deployment and job variety inside their chosen niches.
Many Chinese language firms demonstrated their bodily AI on the World Robotics Convention in Beijing final month. XSquare founder Wang Qian argued that progress is outpacing expectations.
Fashions with GPT‑3‑like functionality for the bodily world might emerge inside a 12 months, with industrial use one to 2 years later and family adoption in three to 5 years.
Corporations in China assault knowledge bottleneck
China is attacking the embodied AI knowledge bottleneck on a number of fronts:
- Giant actual‑world knowledge factories, akin to AgiBot’s and new hubs in Shanghai, Beijing, Shenzhen/Tianjin
- Human demonstration and simulation pipelines like these of Parsini and academia
- Regional consortia in Hubei and Guangzhou
- Open datasets together with AgiBot World and RoboMIND
Collectively, they type a nationwide knowledge engine for multi‑talent, multimodal robotic studying. As facilities interconnect through knowledge exchanges, throughput is rising: AgiBot’s ~50k/day seize is now not distinctive; friends goal related output, and Parsini goals an order of magnitude greater.
Coupled with cloud and chip partnerships, this surge marks an inflection towards broad actual‑world deployment – probably at a scale that outpaces different nations.
Beijing offers coverage assist for bodily AI
Beijing is throwing its weight behind “new productive forces” in bodily AI. “Embodied intelligence” was talked about for the primary time within the premier’s annual work report in March.
Additionally that month, throughout the Two Periods, the Nationwide Growth and Reform Fee (NDRC) introduced a ¥1 trillion ($138 billion) fund for “exhausting tech.” This contains robotics and AI, chips and superior tools, quantum computing, clear power, and different rising applied sciences and can use market‑based mostly fairness with lifecycles as much as 20 years.
On the second Embodied AI Convention hosted by the Chinese language Academy of Sciences along with main universities in March, delegates adopted a white paper with a 3‑stage roadmap:
- 2025 to 2027, basis: Shared datasets and open middleware
- 2028 to 2030, scaling: Deployment in factories, logistics, elder‑care pilots
- Publish‑2030, generalization: Mass‑market humanoids and composite robots
Regulatory scaffolding is rising. The Ministry of Business and Data Know-how’s (MIIT) Clever Knowledge Assortment Normal 1.0 (11/2024) proposes a unified, enforceable framework for synchronizing, formatting, labeling, and high quality‑grading multimodal robotic coaching knowledge. It’s aimed toward interoperability and privateness compliance.
On the municipal stage, Shanghai has issued China’s first plan for embodied intelligence, pairing broad R&D assist with shared infrastructure – compute, testing, pilot manufacturing, and financing. Town additionally added demand‑pull instruments akin to vouchers, demonstration tasks, and modest gross sales incentives to maneuver bodily AI from lab prototypes to scaled manufacturing.
Nationally, the“AI+” plan, introduced final month, seeks to mainstream clever methods throughout China’s economic system. It has set adoption targets above 70% by 2027 and 90% by 2030 on the best way to an “clever economic system” by 2035. This creates steady demand and infrastructure assist for AI in robotics, good automobiles, healthcare, and upgraded infrastructure.
Is Chinese language AI funding a bubble?
Vital observers have requested what number of industries China can maintain with heavy state assist, and for the way lengthy.
When requested why funding in humanoids is booming, one insider at a number one producer put it bluntly: Capital is rotating into this sector as a result of a number of of the beforehand “scorching” performs didn’t earn money.
Constraints are actual: Native governments that underwrite a lot of the assist carry heavy debt burdens. Subsidy-driven dominance is upsetting commerce pushback overseas, and a few goal sectors may be structurally more durable to catch up in.
On the similar time, Beijing views bodily AI capabilities as important to future productiveness, demographic headwinds, and nationwide safety. Previous episodes present China can soak up vital inefficiencies and nonetheless obtain strategic outcomes, albeit at excessive value. Pockets of overbuild and consolidation are probably, however it might be a mistake to underestimate China’s resolve.
Though the U.S. invented the economic robotic, it now not has a scaled, homegrown industrial robotic OEM. Overseas distributors are solely now including meeting capability. There’s a generational hole in manufacturing know-how. Policymakers wish to reshore manufacturing, however expertise shortages, excessive prices, a thinned-out provider base, and a shortfall in core parts are actual hurdles.
The U.S. nonetheless leads in software program and core AI, and its universities appeal to prime expertise. For Chinese language humanoid makers, NVIDIA stays the central compute provider for now. Its toolchains and coaching GPUs are the default worldwide.
Bodily Intelligence and Google’s Gemini Robotics tasks are considered as benchmarks for end-to-end VLA fashions. U.S. humanoid efforts from Determine AI, Agility Robotics, and Apptronik look extra technologically bold than most Chinese language counterparts.
In concept, the U.S. allocates capital extra effectively. The worldwide division of labor has raised dwelling requirements, but it surely additionally hollowed out home manufacturing.
Good robotics corporations arbitrage China’s value and velocity. Broad countervailing duties, whereas typically warranted, would elevate the price of automation and gradual innovation. In the meantime, dependencies have multiplied: rare-earth processing, battery cells for drones, and the focus of superior foundry capability in Taiwan are well-known vulnerabilities.
On this context, the Affiliation for Advancing Automation (A3) has warned that the U.S. dangers shedding its edge in bodily AI. With out management within the bodily manifestation of AI, A3 argues, the U.S. might lose not solely the robotics race however the AI race itself.
What the U.S. ought to do to compete with China
To deal with these issues, A3’s Nationwide Robotics Technique, launched in March, provided suggestions:
- Establishing a central robotics workplace within the federal authorities to coordinate technique
- Providing tax incentives for automation adoption
- Making the authorities itself a lead buyer of robotics
- Increasing STEM workforce coaching
- Boosting R&D funding
- Updating requirements for AI-driven machines
Will that be sufficient to counter China’s momentum? Market economies prize effectivity and near-term returns. As highlighted above, China’s state-led mannequin tolerates extended inefficiency and even years of losses to safe strategic breakthroughs.
What sort of U.S. industrial coverage is sensible on this context? Silicon Valley grew out of protection procurement. The CHIPS Act and Inflation Discount Act confirmed each what is feasible and the place coverage design can enhance. The objective have to be higher system design with market self-discipline, not central planning.
Rebuilding the economic base is the core. Robotics firms want prospects, sparring companions, and supply-chain synergies. An power coverage that delivers plentiful, dependable, and inexpensive energy is a structural benefit — particularly relative to Europe’s greater industrial power prices.
Capital will matter. The U.S. brings many years of enterprise capital expertise, unmatched across the globe. Nonetheless, capital-intensive fields like robotics usually misplaced out to fast-scaling software program performs over the previous 20 years.
The LLM breakthrough and its switch into the bodily world seems to be altering that. Alongside Tesla and Alphabet, Amazon and Meta are constructing AI-enabled robots.
Google co-founder Larry Web page is reportedly exploring AI-reinvented manufacturing at Dynatomics. Sequoia just lately coined the time period “cognitive meeting line.” The relative lack of legacy infrastructure would possibly prove as a bonus throughout the subsequent technological revolution.
The nation that draws the brightest and most bold folks will lead the world. The U.S. ought to subsequently maintain its openness to folks and markets. Particularly, as China eases expert immigration guidelines and promotes its tech stack as an inexpensive different for the World South.
Lastly, rebuild important {hardware} provide chains for actuators, precision reducers, high-efficiency drives, sensors, energy electronics, batteries, and precision gears. With intense competitors and local-content guidelines in China, main European and Japanese automation suppliers are reassessing footprints and are open to North American co-location the place demand and coverage readability exist.
The U.S. ought to courtroom them with fast-track allowing, site-readiness, and mission-driven procurement to localize manufacturing and shut the aptitude hole.
Concerning the creator and RoboBusiness
Georg Stieler advises a number of the world’s largest robotics firms. He spent greater than 10 years dwelling in China and now splits his time between Switzerland and the Folks’s Republic of China. Due to multi‑12 months, shut collaboration with AI startups in Silicon Valley, he’s additionally deeply conversant in its tradition.
At RoboBusiness 2025, Stieler will probably be collaborating in a panel on “Closing the Robotics Hole With China” together with Jeff Burnstein, president of A3, and Eric Truebenbach, managing director at Teradyne Ventures. He can even current in a session on the “World Implications of China’s Robotics Push.”
RoboBusiness, the premiere occasion for builders and suppliers of economic robots, will probably be on Oct. 15 and 16 in Santa Clara, Calif. The occasion will characteristic tracks on enabling applied sciences, humanoids (new), enterprise, design and growth, and discipline robotics, in addition to the Bodily AI Discussion board.
This 12 months’s convention can even embody greater than 60 audio system, a startup workshop, the annual Pitchfire competitors, and quite a few networking alternatives. Over 100 exhibitors on the present ground will showcase their newest enabling applied sciences, merchandise, and providers to assist remedy your robotics growth challenges. Registration is now open.
