Hardware, data, and foundation models — how China's supply chain density, state-backed industrial policy, and EV-adjacent manufacturing base have made it the world's humanoid production capital.
Read the report → Download PDFIn 2025, Chinese companies shipped roughly 80% of the world's humanoid robots. In 2026, that share is projected to rise as Unitree and AgiBot alone target 75,000+ unit annual capacity — more than the entire Western supply combined.
China is not merely a participant in the global robotics industry. It is the gravitational center. With more than 70% of global industrial robot installations in 2025, an estimated $14.2 billion domestic market (up 47% year-over-year — the fastest growth of any major economy), and the most ambitious national humanoid robotics program on earth, China's influence on the trajectory of global robotics is profound and growing.
What makes China's position remarkable is not just its scale — though the scale is extraordinary — but the depth of the ecosystem that supports it. The Shenzhen hardware supply chain enables robotic arm prototyping in days rather than months. The manufacturing giants that drive demand — BYD, CATL, Foxconn — represent a buyer base that no other country can match. And China's research institutions are producing AI and robotics work that is globally competitive and increasingly collaborative.
Three forces define China's position. First, supply chain integration: the Yangtze River Delta hosts the world's most vertically integrated humanoid supply chain — Unitree manufactures motors, reducers, and sensors in-house; component suppliers (DJI, DeepSeek, EV actuator makers) are within 2-hour logistics radius. Second, national policy coordination: the 2025 Humanoid Robot Action Plan, issued by MIIT and five ministries, targets 100,000 deployed humanoids by 2027. Third, demand-side scale: BYD (world's largest EV maker), CATL (dominant battery manufacturer), and Foxconn alone represent demand for tens of thousands of systems.
A prototype that takes 12 weeks to produce in the US or Germany is turned around in 10–14 days in Shenzhen, at a fraction of the cost. The same EV component supply chain that made China the world's largest car manufacturer now serves humanoids — actuators, gears, precision motors can often be repurposed directly between the two.
SVRC views China not as a competitive threat but as the single most important partnership opportunity in global robotics. This report reflects that view. We believe that robotics companies, researchers, and operators who engage constructively with China's ecosystem will build better products, collect better data, and ultimately deploy more capable robots.
Manufacturing scale, supply-chain vertical integration, and national industrial policy — the gravitational center of global robotics production.
The China robotics market reached $14.2B in 2026, growing +47% year-over-year. This trajectory reflects the confluence of labor-market dynamics, policy incentives, and foundation-model-enabled deployment velocity discussed throughout this report.
Unit shipments tell a more revealing story than market dollars. Shanghai's AgiBot led the world with 5,168 humanoid units in 2025 (39% global share per Omdia), followed by Hangzhou's Unitree at 5,500 (Unitree's own figure disputes Omdia's second-place ranking). Combined with UBTECH (7% share) and Leju (~5%), Chinese firms control nearly 90% of global humanoid installations.
276,000 industrial robots installed in a single year — more than Japan, the US, Germany, and South Korea combined.
The numbers are staggering. China installed approximately 276,000 industrial robots in 2025, according to preliminary IFR data. Japan — the second-largest market by installation volume — installed approximately 46,000 units in the same period. The United States installed approximately 35,000. China installs more industrial robots in a single quarter than most countries install in a year.
Through much of the 2010s, the majority of robots installed in China were manufactured by foreign companies — FANUC, ABB, KUKA (now Chinese-owned), and Yaskawa. By 2025, Chinese-manufactured robots have captured approximately 40% of the domestic market by unit volume, up from 27% in 2020. Companies like Estun, Siasun, Inovance, and STEP Electric have built competitive products for the mid-market, while a new generation of startups targets the emerging sub-$10K segment.
China's robot density — 392 robots per 10,000 manufacturing workers in 2024 — has grown extraordinarily fast. In 2019, this figure was 187. The 2.1x increase in five years is the fastest density growth of any major economy. At current trajectory, China will surpass Germany (397 robots/10k) within 12 months and could approach Japan (399) by 2028. South Korea (1,012) remains the global leader and is unlikely to be overtaken within this decade, but the direction is clear.
China's installation volume has second-order effects that extend far beyond its borders. First, it creates the world's largest training ground for robot integration engineers, maintenance technicians, and application developers. The human capital being built in China's factories will shape the global robotics workforce for decades. Second, the sheer volume of deployed robots generates more operational data than any other country — data that, properly collected and curated, could accelerate AI-driven robotics across the entire industry. Third, the price competition driven by Chinese manufacturers is making robots more accessible globally, lowering barriers for adoption in developing economies across Southeast Asia, South America, and Africa.
Every robotics market has its flagship firms — the companies whose trajectory shapes the country's narrative. China's humanoid ambitions are among the most comprehensive and well-resourced in the world.
The Tiangong humanoid, developed by the Beijing Humanoid Robot Innovation Center, represents China's flagship government-backed humanoid effort. The robot features 42 degrees of freedom, has demonstrated walking, stair climbing, and basic manipulation, and is designed as a platform for both research and eventual commercial deployment. The Tiangong program is notable for its consortium structure — bringing together multiple universities, research institutes, and companies under a coordinated national framework.
Galaxy Robotics, one of the newest entrants, has raised over $50 million and represents the next wave of Chinese humanoid startups building on the infrastructure and talent pool created by earlier companies. With 330+ distinct Chinese humanoid models unveiled in 2025 per MIIT, the ecosystem is generating extraordinary variety and competitive pressure.
| Company | Robot | Funding | Distinctive Approach |
|---|---|---|---|
| Tiangong (BHRIC) | Tiangong | Government-backed | National consortium; 42 DoF platform |
| AgiBot | AgiBot series | $100M+ | Manufacturing-first deployment |
| Unitree | H1, G1, R1 | $80M+ ($3B val.) | Aggressive pricing, global distribution |
| UBTECH | Walker S | $820M cumulative | Longest iteration history; consumer DNA |
| Fourier Intelligence | GR series | $120M+ | Rehabilitation crossover; biomechanics |
| Galaxy Robotics | Galaxy series | $50M+ | Next-gen; built on ecosystem talent |
| XPENG Robotics | IRON | EV-backed | Automotive AI and manufacturing scale |
| Leju Robotics | Kuavo | Tencent-backed | Education and research focus |
If you want to understand why Chinese robotics companies can iterate faster and produce hardware more cheaply than their competitors anywhere else in the world, the answer begins and ends in Shenzhen.
Within a 100-kilometer radius of central Shenzhen, a robotics company can source brushless DC motors, harmonic drives, strain wave gears, CNC-machined aluminum frames, injection-molded housings, custom PCBs, depth cameras, IMUs, force-torque sensors, and power electronics — often with first samples delivered within a week. This density of suppliers creates a compound speed advantage: each iteration cycle is shorter, each prototype is cheaper, and the ability to experiment with multiple design approaches in parallel is dramatically greater than in any other geography.
The numbers tell the story. A robotic arm prototype that costs $15,000–$25,000 and takes 10–14 weeks to produce in the United States can be produced in Shenzhen for $4,000–$8,000 in 10–14 days. This is not a marginal difference. It is a structural advantage that shapes the entire competitive landscape for sub-$10K robotic hardware.
Of the fourteen manufacturers globally producing robotic arms priced under $10,000, eight are Chinese companies. Companies like Unitree, AgileX, Elephant Robotics, LEBAI, Flexiv, and others have leveraged local manufacturing capabilities to produce arms that are competitive on both price and technical specification. Several of these companies have also invested in software — ROS 2 compatibility, URDF models, and integration with standard robot learning frameworks — making their hardware accessible to the global research and development community.
The Shenzhen advantage extends beyond the robots themselves to the data collection hardware that trains them. Chinese manufacturers are producing leader-follower teleoperation systems for under $2,000 — systems that are functionally equivalent to the $5,000–$8,000 devices available from US and European suppliers. This cost reduction has direct implications for the economics of robot learning: it makes large-scale teleoperation campaigns economically viable for a wider range of organizations and tasks.
The Shenzhen hardware advantage is not merely a Chinese advantage. When Chinese manufacturers produce affordable robotic arms and teleoperation hardware, they reduce costs for robotics researchers and companies everywhere. A research lab in Brazil, Nigeria, or India that could not afford a $15,000 arm can now purchase a $4,000 arm with comparable capabilities. This democratization of access is one of the most important developments in global robotics.
Three companies — BYD, CATL, and Foxconn — together employ millions of workers and are actively investing in robotic automation at a scale that makes them, collectively, the largest robot buyers on earth.
Where robots are actually working in China today — and where growth is accelerating fastest. SVRC's estimates reflect operational stock, not cumulative installations.
| Vertical | Deployed Units (2025E) | YoY Growth | Leading Form Factor |
|---|---|---|---|
| EV / Automotive | 28,000 | +42% | Precision 6-DoF + humanoid |
| Electronics (Foxconn etc.) | 19,500 | +35% | Bimanual arm system |
| Battery (CATL, BYD) | 12,500 | +51% | Precision arm + inspection |
| Logistics / JD, Cainiao | 8,200 | +39% | Mobile manipulator |
| General Manufacturing | 5,800 | +28% | Fixed industrial arm |
| Consumer / Retail demo | 3,400 | +124% | Humanoid (demo/PR) |
BYD has become the world's largest electric vehicle manufacturer, with production capacity spanning passenger cars, buses, trucks, and battery systems. The company employs over 700,000 workers. BYD's automation journey is driven by growth velocity (scaling production faster than human hiring can keep pace) and quality requirements (EV manufacturing demands precision that exceeds what manual assembly can reliably deliver). A single BYD factory operating 50 robotic arms for 12 months generates more task-specific demonstration data than most academic datasets contain in total.
CATL (Contemporary Amperex Technology), the world's largest battery manufacturer by market share, faces automation imperatives driven by the extreme precision required in lithium-ion cell production. Cell assembly involves handling thin electrode materials, maintaining clean-room conditions, and executing thousands of identical operations per shift with near-zero defect tolerance. CATL has begun piloting AI-driven manipulation systems that can adapt to the subtle material variations inherent in battery production.
Foxconn employs over one million workers across its global operations, with the majority based in mainland China. The company's "Lights Out Factory" vision has been a stated goal for over a decade. Foxconn's investment in flexible, AI-driven manipulation systems reflects the reality that consumer electronics production requires frequent line changeovers that traditional fixed automation cannot accommodate efficiently.
China's largest manufacturers are not merely buying robots — they are generating the deployment data that will train the next generation of robotic systems. This creates a virtuous cycle: more deployment generates more data, which trains better models, which enables more capable deployment. Understanding this cycle is essential for anyone seeking to understand where global robotics is heading.
China's AI research output has grown remarkably over the past decade, and the quality of that output has risen in parallel with its volume.
Chinese universities and research institutions are now publishing work in robotics, computer vision, natural language processing, and reinforcement learning that is competitive with the best work produced anywhere in the world. This is supported by citation metrics, acceptance rates at top conferences, and the practical impact of Chinese-origin research on tools and methods used globally.
Tsinghua University (THU), Peking University (PKU), Shanghai Jiao Tong University (SJTU), and Zhejiang University are among the leading institutions. Tsinghua's Cross-Disciplinary Institute for Robotics has become a significant center for humanoid robotics research, while SJTU's robotics group has contributed influential work on manipulation planning and dexterous hand control. These institutions are not working in isolation — collaborative projects with MIT, Stanford, CMU, ETH Zurich, and the University of Tokyo are common and productive.
Chinese researchers and companies are increasingly active contributors to open-source robotics and AI projects. Contributions to LeRobot, open-source URDF model libraries, ROS 2 packages, and robot learning frameworks have expanded the global toolkit. Unitree's open-source releases have been widely adopted by the international research community. This pattern reflects a pragmatic understanding that shared infrastructure benefits everyone — including the contributing company.
The bridge between academic research and industrial application in China is shorter than in many countries, partly because of the scale of industrial demand. The cultural expectation that academic work should have practical applications creates a fast feedback loop — ideas that work in the lab are tested in factories months later, and the results inform the next round of research.
China's government support for robotics is comprehensive, coordinated across multiple levels of government, and explicitly linked to national economic strategy.
The Made in China 2025 strategic plan, announced in 2015, identified robotics as one of ten priority industries. The plan set targets for increasing domestic market share, improving robot density, and developing core technologies including precision reducers, servo motors, and controllers that had previously been imported from Japan and Germany. A decade later, many of these targets have been met or exceeded.
The 14th Five-Year Plan included specific provisions for robotics development, emphasizing intelligent manufacturing, service robots, and AI integration. The plan directed provincial and municipal governments to develop local robotics industrial parks, resulting in dedicated robotics zones in Beijing, Shanghai, Shenzhen, Guangzhou, Hangzhou, and Chengdu. These parks offer subsidized facilities, streamlined permitting, and local government investment.
The 2025 Humanoid Robot Action Plan, issued jointly by MIIT and five other ministries, is perhaps the most ambitious government humanoid robotics initiative anywhere in the world. The plan sets a national target of 100,000 humanoid robots deployed by 2027 and calls for developing a complete domestic supply chain for humanoid components including high-performance actuators, dexterous hands, and perception systems. The plan also establishes national standards for humanoid robot safety, interoperability, and performance testing.
The scale of Chinese government support for robotics is significant, but what may be more important is the coordination. National, provincial, and municipal programs are designed to complement rather than duplicate each other, creating a layered support structure that addresses everything from fundamental research to factory-floor deployment. International partners benefit from understanding this structure.
Despite the geopolitical complexities of the current moment, China's robotics industry is deeply and productively engaged with international partners — and this engagement is growing, not shrinking.
Chinese robotic arm exports have grown exponentially, with an indexed volume increase of 4.45x from 2020 to 2025. These exports serve customers across Southeast Asia, South America, Europe, the Middle East, and Africa — markets where affordable robotic hardware is unlocking automation for the first time. Chinese manufacturers have invested in international distribution networks, multilingual support, and compliance with international safety standards (CE, UL).
Joint research programs between Chinese universities and international partners continue to produce significant work. Collaborations with MIT, Stanford, CMU, ETH Zurich, and the University of Tokyo have resulted in co-authored papers at top robotics conferences and shared benchmark datasets. These collaborations focus on specific technical challenges — manipulation planning, sim-to-real transfer, dataset quality standards — where mutual benefit is clear.
Chinese companies and researchers contribute code, models, and datasets to shared repositories that benefit the global community. Contributions from Chinese developers have improved ROS 2 support for Chinese-manufactured hardware, expanded open-source URDF libraries, and contributed training data to shared robot learning datasets. The LeRobot project, in particular, has benefited from active Chinese community participation.
International collaboration in robotics is not a nice-to-have. It is essential. The challenges facing the industry — safety, data quality, affordability, and responsible deployment — require the broadest possible base of contributors. Chinese institutions and companies are among the most capable and productive contributors to solving these shared challenges.
SVRC is committed to building genuine, productive partnerships with Chinese robotics companies, research institutions, and data collection operators.
SVRC maintains direct procurement relationships with Chinese robotic arm and teleoperation hardware manufacturers. We test, benchmark, and certify Chinese-manufactured hardware for our enterprise customers and research partners, providing quality assurance and integration support that helps Chinese manufacturers reach international markets. Our hardware library includes products from Unitree, AgileX, Elephant Robotics, and others.
SVRC is building a data collection operator network that includes Chinese operators. We believe that teleoperation data collected in Chinese manufacturing environments has immense value for training general-purpose robotic policies. Our data partnerships are designed to be fair and transparent: operators retain ownership of their raw data, SVRC provides quality scoring and standardization, and both parties benefit from the resulting curated datasets.
SVRC is actively pursuing research collaborations with Chinese universities and institutes. Our focus areas include VLA model fine-tuning for manufacturing tasks, dataset quality standards, and sim-to-real transfer validation. We contribute our proprietary evaluation infrastructure and standardized benchmarks; our Chinese partners contribute research expertise and access to deployment environments.
Email contact@roboticscenter.ai or visit roboticscenter.ai/contact. We respond to all partnership inquiries within 48 hours. Chinese-language inquiries are welcome — we have team members who read and write Mandarin.
A candid assessment of what China does best in global robotics, where structural vulnerabilities require attention, and the themes that will define the next 18 months.
A successful STAR market listing at $7B valuation would lift the entire Chinese robotics supply chain and likely trigger follow-on IPOs from AgiBot, XPENG Robotics, and Fourier.
The 100,000 humanoid target for 2027 is ambitious, but China has a track record of meeting industrial deployment targets. Whether the full target is met is less important than whether the deployments that do occur generate the operational data needed for the next iteration cycle.
Unitree R1 at $5,900 is the opening salvo. 2027 likely sees sub-$5K consumer-grade humanoids from multiple Chinese firms — a price point that is categorically impossible for Western makers.
The trend from 27% (2020) to approximately 40% (2025) is expected to continue. By end of 2027, Chinese manufacturers are likely to capture 50% or more of their domestic market, driven by continued improvements in reducer technology, controller software, and integration tooling.
Expect new US restrictions on Chinese humanoids in 2026–2027, mirroring semiconductor and EV playbooks. This accelerates rather than blocks domestic demand.
Huawei Ascend and domestic inference accelerators will increasingly replace NVIDIA Jetson in Chinese humanoids. Performance gap is closing; supply is reliable.
As Chinese factories deploy hundreds of thousands of robots, the training data generated by those deployments becomes an increasingly valuable resource. We expect new data partnership models to emerge — structures that allow Chinese manufacturers to share deployment data with international partners in ways that respect both parties' interests.
China's robotics industry is building something remarkable — an ecosystem of scale, speed, and ambition that the global robotics community should engage with, learn from, and partner with. The companies and researchers who do so will build better robots, collect better data, and ultimately contribute more to solving the hard problems that matter. SVRC is committed to being a bridge, and we invite you to cross it with us.
Globally, investors increasingly cite proprietary data collection infrastructure as the primary defensibility argument in robotics. The question for China specifically: do its robotics companies generate deployment-specific data at a rate that compounds faster than foundation model improvements erode it? This is the question that 2026–2027 will answer.
In 2025, Chinese humanoid-related investment reached 39.8B RMB (~$5.5B) across 325 deals — 326% year-over-year growth. Unitree closed Series C at $3B valuation targeting $7B at IPO. Galbot raised over $300M. The first twenty days of January 2026 saw 18 humanoid stocks with net inflows exceeding 100M RMB each. Capital availability is no longer the bottleneck; operational execution is.
Whether you're an enterprise evaluating deployment, a manufacturer considering market entry, or an investor sizing the opportunity — SVRC partners on hardware sourcing, data collection programs, policy navigation, and on-the-ground deployment coordination.