EN / 中文

2026 World Robot Conference · Industry Leaders Discuss Market Landscape: From "Diverse Proliferation" to "Tiered Differentiation"

by zhongguodianzibao·August 26, 2026

Editor's Note: The embodied intelligent robot industry has reached a critical intersection of technological iteration and large-scale deployment, accelerating its integration into thousands of industries. The entire sector is swiftly transitioning from the "technology validation phase" to the "commercial popularization phase," with the outline of a new round of growth becoming increasingly clear. Meanwhile, more practical propositions are placed before every industry practitioner: where should technological breakthroughs be made, where should the market be deeply cultivated, and how should the ecosystem be co-built... Based on this, on the occasion of the 2026 World Robot Conference, China Electronics News invited representatives from multiple robotics companies to discuss core topics such as technological evolution, market deployment, ecosystem synergy, and future trends. By collecting voices from the front lines of the industry, we aim to leave a vivid footnote for the next chapter of the robotics industry.

What changes are occurring in the current industry competition landscape? What do you consider to be the core competitive advantages and potential shortcomings of Chinese enterprises in the global embodied intelligence track?

The industrial landscape is shifting from "diverse proliferation" to "tiered competition and cooperation." The robotics industry will not see an oligopoly eating up the entire market like the semiconductor sector: only about a quarter of the market—general-purpose humanoid hardware and general-purpose brain foundations—might be monopolized by large corporations; the remaining three-quarters, including humanoid-specific software and non-humanoid hardware and software, cannot be monopolized, allowing for tiered complementarity among large corporations, tech giants, and startups. Chinese enterprises have three advantages: the world's most complete hardware supply chain (taking Hunan as an example, the local area can provide 80% supply chain support); a testing ground and data mine composed of massive application scenarios; and an implementation path featuring modular reconfiguration and localized operation and maintenance. Shortcomings must also be faced squarely: the underlying "brain" is about 1.5 to 3 years behind overseas leaders, and the hardest gap to close is the real-machine data closed loop, though language understanding and scenario decision-making have basically caught up; high-end chips and precision reducers still face the risk of supply chain "choke points"; and a data sharing mechanism is lacking. Chaonengyuan is willing to work with partners, driven by scenarios and co-building the ecosystem, to bring robots into thousands of households.

First, China's hardware supply chain has achieved a substantive breakthrough. China possesses the world's most complete robotics hardware supply chain, covering the entire industry chain including motors, reducers, lead screws, sensors, and controllers, with costs only 40% to 60% of those in Europe and the United States. In 2025, the top six global shipments of humanoid robots were all from Chinese enterprises. These six manufacturers collectively accounted for 74.1% of the global shipment share, and China's total humanoid robot shipments accounted for 84.7% of the global total. Industry institutions predict that in 2026, the total annual shipments of domestic humanoid and quadruped robots are expected to exceed 100,000 units, marking the official entry of the industry into the year of mass production.

Taking Xinghaitu as an example, from motor reducers to complete machine delivery, Xinghaitu is rooted in China's complete hardware industry chain, achieving a hardware breakthrough model of "made locally, used globally." With thousands of orders and a global developer ecosystem of over a hundred partners, Xinghaitu has proven the irreplaceable competitiveness of China's robotics hardware supply chain with its actual performance of ranking first globally in the market share of wheeled dual-arm robots. In less than three years since its establishment in September 2023, it has ranked first in the global market share in the field of wheeled dual-arm robots, accumulating orders in the scale of thousands. Product shipments are mainly to the developer community, with over 150 developer customers, covering three profiles: university and research institute laboratories, basic research departments of large tech companies, and leading embodied startups. Its R1 series products have covered over 90% of the world's top developers, with clients including Fei-Fei Li's team at Stanford, Physical Intelligence, Huawei, BYD, and other well-known institutions and companies, becoming a typical sample of China's embodied intelligence hardware breakthrough.

Second, the gap between China and overseas in the "brain" level is rapidly narrowing. In core algorithm fields such as VLA (Vision-Language-Action) foundation models and world models, overseas indeed has a first-mover advantage, but the catching-up speed of China's leading enterprises exceeds expectations. Taking Xinghaitu as an example, as mentioned above, the new generation VLA foundation model G0.5 we recently released and open-sourced directly benchmarks against Physical Intelligence's π0.6, ranking first in China and top three in the world. In the direction of world models, the Fast-WAM released by Xinghaitu in March 2026 has also attracted attention from the academic community. The related paper, after being made public on arXiv, sparked widespread discussion. Xie Saining, a well-known scholar in the AI field, specifically reposted and commented, suggesting to "best read it together with Yann LeCun's new work"—discussing Xinghaitu's work alongside the latest achievements of a Turing Award winner is an extremely rare recognition in the academic circle and the most intuitive international endorsement of Xinghaitu's algorithmic strength.

The industrial shortboard is the severe scarcity of high-quality physical world data, which has become the biggest bottleneck in the industry. Experts from the Development Research Center of the State Council pointed out that due to the highly unstructured nature of real scenarios and the inconsistent standards of different hardware embodiments, data collection costs are high and difficult to reuse, and the scale of valid data available for training is far from keeping up with the needs of model iteration. Industry calculations show that training an embodied model with general generalization capabilities requires tens of millions or even hundreds of millions of hours of real operation data. Data is becoming the biggest "invisible bottleneck" for embodied intelligence to move from the laboratory to the real world.

Combining Xinghaitu's data company "Yishu Intelligence" and the "1 Million Hours of Ultra-High-Quality Real Data Plan," we believe we can explore a government-enterprise collaborative data infrastructure model. Moreover, we believe that embodied intelligence is an "infinite game"—there is no endgame, only continuous iteration. As long as this system remains open and sensitive to real needs, shortcomings can be turned into springboards, and bottlenecks can be turned into breakthroughs.

Industry competition is shifting from single-point technology contests to a contest of full-stack capabilities and ecosystem synergy. In the past, making a dexterous hand or an algorithm could attract attention; now, the industry requires end-to-end delivery capabilities of "hardware-perception-data-model-scenario."

The core competitive advantages of Chinese enterprises lie in the industry chain and scenarios. We have the world's most complete robotics component industry cluster, with globally leading new product iteration speeds; at the same time, the rich domestic industrial, power, and logistics scenarios provide massive real data for technology deployment. The potential shortboard is underlying original theories and high-end precision components.

Of course, in areas such as underlying original theories, general algorithm model frameworks, and high-end sensors, Chinese enterprises still need to continue catching up.

The competition landscape is shifting from "diverse proliferation" to "tiered differentiation." Relying on mass production capabilities, scenario accumulation, and ecosystem layout, leading enterprises are widening the gap with followers. Meanwhile, cross-boundary entrants—including tech giants, automakers, and home appliance enterprises—are accelerating their penetration, making industry boundaries increasingly blurred.

The core competitive advantages of Chinese enterprises lie in three points: first, the world's most complete manufacturing system and supply chain advantages, providing a solid foundation for the scaled production and cost control of humanoid robots; second, rich application scenarios and massive real data, which are the "nutrients" for training embodied intelligence models; third, strategic attention from the policy end, providing stable expectations for industrial development.

In terms of potential shortcomings, first, there is still a gap with leading countries such as the United States in basic theoretical research and original innovation; second, in underlying links such as high-end chips and operating systems, dependence has not been completely eliminated; third, brand internationalization and global market expansion capabilities need to be improved.

Leju's strategy is to "leverage strengths and offset weaknesses": fully utilizing the advantages of scenario deployment and full-stack independent R&D, while gathering the wisdom of global developers through the OpenLET open-source ecosystem to promote innovation through openness. Third-party data shows that the company's humanoid robot shipments have entered the top ranks globally, confirming the first-mover advantage of Chinese enterprises in industrialization deployment.

Currently, the competition landscape of the embodied intelligent robot industry has obviously changed. It is no longer about who builds the prototype first to gain an advantage; now, everyone is competing in real operation capabilities, deployment cases, and business models. The track has also differentiated, with multiple routes running in parallel: general-purpose humanoids, industrial-specific embodied robots, and quadruped special robots. Competition has shifted from single-product contests to ecosystem competition. The biggest advantage of Chinese enterprises is the huge manufacturing market, complete supply chain, and strong engineering deployment capabilities; the shortcomings are concentrated in underlying simulation software, insufficient performance of high-end components, lack of data coordination mechanisms, slow construction of safety standards and testing systems, and insufficient investment in basic frontier research.

The most significant change is that the evaluation criteria of capital have shifted from "technical concepts" to "commercial closed loops." Investors no longer just look at how cool the robots' running, jumping, or boxing is, but ask "whether they can create real value." The industry is moving from "who is cooler" to "who is more practical." The core competitive advantage of Chinese enterprises lies in the fact that robot hardware products can quickly form strong competitiveness, while having rich deployment scenarios to quickly verify the commercial closed loop and form scaled replication, thereby further driving the data flywheel and improving the intelligence level of the models.

From the perspective of global industrial competition, the focus of embodied intelligence competition has long shifted from single-point technology contests of laboratory prototypes to a comprehensive contest of industry chain synergy, scenario deployment efficiency, and ecosystem construction. China's industry has risen rapidly in this wave, with both outstanding advantages and areas that need further cultivation.

China has a complete supply chain system for robotics and intelligent manufacturing, with deep linkage among enterprises in the upstream, midstream, and downstream. Upstream and downstream collaborative innovation and mutual empowerment can not only accelerate technological iteration but also amortize the trial-and-error costs of the entire industry. Veichi Electric will continue to play its role, steadily promote the localization of core components, and concentrate on creating innovative products with high performance, high quality, and high reliability. We will continuously refine our technological strength and product competitiveness, constantly strengthen our market response capabilities, promote the deep integration of embodied intelligence technology into various fields of the real economy, drive the value enhancement of the entire industry chain with intelligent equipment upgrades, and activate new driving forces for industrial development with AI innovation.

Currently, four obvious changes are occurring in industry competition: shifting from single-action demonstrations to complete task chain competition; shifting from embodiment parameter competition to system competition of "embodiment, perception, model, data, and scenario"; shifting from prototype releases to mass production delivery and after-sales operation and maintenance competition; shifting from single-point enterprise breakthroughs to industrial ecosystem and development platform competition. In the future, what truly forms a barrier is not necessarily the highest single indicator, but who can first form a positive cycle of "scaled deployment—data feedback—model improvement—cost reduction—further deployment."

The core advantages of Chinese enterprises are mainly reflected in the complete manufacturing system, strong engineering and cost control capabilities, rich industrial application scenarios, and supply chain synergy and product iteration speed. IFR data shows that in 2024, China's industrial robot installations reached 295,000 units, accounting for 54% of global installations, and the share of Chinese manufacturers in the domestic industrial robot market rose to 57%. These data belong to the industrial robot caliber and cannot be equated with humanoid robot sales, but they can illustrate that China has a strong foundation in manufacturing, integration, and application.

Looking ahead to 2030, please make a one-sentence outlook on the ultimate vision of the embodied intelligent robot industry, and put forward some suggestions for industry development.

By 2030, dexterous manipulation will become an important foundational capability in the physical world. Robots do not necessarily have to look like humans, but they must be able to stably complete real tasks like skilled workers.

The industry will gradually abandon short-term concept hype and solidly tackle the core threshold of dexterous manipulation. Taking real scenarios as the examination room, stable work as the standard, and data closed loops as the engine, the government, enterprises, and research institutions will collaboratively promote standard construction and ecosystem sharing, jointly driving China's embodied intelligence industry from scale leadership to technological leadership.

Looking ahead to 2030, we believe that embodied intelligent robots will become the infrastructure of human society just like today's smartphones. At that time, humanoid robots will no longer just be "advanced tools" in factories, but "intelligent partners" deeply integrated into all scenarios such as manufacturing, logistics, services, healthcare, and households.

Leju's vision is to achieve this goal in three stages: the first stage is to establish technological and commercial closed loops in controllable scenarios such as scientific research, education, and commercial services; the second stage is to achieve scaled application and efficiency improvement in core scenarios such as industrial manufacturing; the third stage is to expand to generalized scenarios such as household services, ultimately entering thousands of households.

Three suggestions for industry development: first, maintain strategic focus. Humanoid robots are a long-cycle track, and it is imperative to avoid eager for quick success and instant benefits. Continuous R&D investment and patient market cultivation are needed. Second, adhere to open collaboration. No single enterprise can complete the industrialization process alone; an industrial ecosystem of "competition and cooperation for common prosperity" needs to be built. Third, always take safety and reliability as the bottom line. Before humanoid robots enter human society on a large scale, complete safety standards and ethical norms must be established so that technological progress truly benefits humanity.

Leju will continue to deeply cultivate the embodied intelligence track, based on full-stack independent R&D technology and guided by scenario deployment, joining hands with industry chain partners to jointly promote China's embodied intelligent robot industry to the forefront of the world.

Looking forward from 2030, I expect embodied intelligent robots to be widely used in industrial and special scenarios, becoming collaborative partners of humans, and building an independent, controllable, and sustainable intelligent equipment industry. For industry development, everyone needs to develop rationally, avoid swarming to involution and building general-purpose humanoid prototypes, and prioritize running through the commercial closed loop of industrial-specific solutions. Balancing short-term commercialization and long-term underlying technological research, holding the bottom line of human-machine safety and data security, insisting on real scenarios guiding technological iteration, uniting to improve the domestic industry chain, and participating in global competition.

Deng Xiong, Vice President of Suzhou Veichi Electric Co., Ltd.: Looking ahead to 2030, I hope embodied intelligence will complete the fundamental transformation from a passively executing "tool" to a "collaborator" with complete cognitive capabilities, becoming a wise partner for humans to expand the boundaries of cognition. May the whole industry work together with one heart and one mind, focus on tackling key problems, and join hands to build a future picture of intelligent symbiosis.

To realize this vision, the industry should adhere to long-termism and application guidance: first, strengthen the R&D of original theories, basic algorithms, key materials, and core components to avoid simple copying and parameter stacking; second, evaluate products based on real task success rates, reliability, and unit task costs, reducing demonstration competitions detached from scenarios; third, accelerate the establishment of data, interface, testing, safety, and ethical standards; fourth, support enterprises to cross the early validation stage through patient capital, first-set applications, scenario opening, and government procurement; fifth, adhere to open collaboration and co-build development platforms, data resources, and industrial ecosystems under the premise of ensuring data security and intellectual property rights.

Looking ahead to 2030, we believe that embodied intelligent robots will truly move from industrial scenarios into public life, becoming the new generation of general-purpose intelligent terminals after computers and smartphones, deeply integrating into all aspects of production, services, and life. On the production end, intelligent robots will become the standard in the manufacturing and logistics industries, greatly improving production efficiency and undertaking high-risk and repetitive work; on the life end, service robots will enter more public spaces and household scenarios, becoming intelligent assistants for people's lives. At the industrial level, China will form a complete and independently controllable embodied intelligence industry chain, with local core technologies and products possessing global competitiveness, building an open and win-win industrial ecosystem.

There are three suggestions for industry development:

First, adhere to long-termism and respect the laws of industrial development. Embodied intelligence is a long-cycle track, and short-term quick success cannot be pursued. We need to calm down and polish core technologies and products, avoid blind follow-the-crowd hype, and promote the steady and healthy development of the industry.

Second, focus on tackling key underlying core technologies. Continuously increase R&D investment in key underlying links such as chips, core algorithms, and high-end actuators, consolidate the foundation of industrial independence and controllability, and enhance long-term core competitiveness.

Third, deepen open collaboration and avoid homogeneous involution. Enterprises in all links of the industry chain should give full play to their respective advantages, develop in a differentiated manner, and complement each other synergistically, jointly making the industrial cake bigger rather than falling into homogeneous competition in low-end fields, and joining hands to promote China's embodied intelligence industry to the forefront of the world.

Black Sesame Technologies will also continue to deeply cultivate edge AI computing power technology, polish more competitive embodied intelligent computing power products and solutions, and work with the whole industry to promote the realization of this vision.