After a decade-long journey that saw its valuation surge more than 4,000-fold from a seed round valued at 10 million yuan in 2016, Yushu Technology Co.,Ltd. (688836.SH), dubbed the "first humanoid robot stock on A-shares," is set to list on the STAR Market, marking one of the most closely watched capital market events in August 2026. The company, widely recognized for its multiple appearances on the CCTV Spring Festival Gala, has ignited a frenzy among retail investors, with nearly ten million participants vying for shares in its new stock subscription.

Behind the fanfare, however, lies a significant valuation debate. With a listing market cap of 61 billion yuan and a price-to-earnings ratio of 219 times, the company's offering has split opinion: on one side, the extraordinary multi-thousand-fold valuation growth over the past decade has attracted heavy backing from internet giants, government industry funds, and leading venture capital firms; on the other, there are stark realities, including a heavy revenue reliance on scientific research scenarios, notable gaps in its embodied large model capabilities, and a significant slowdown in earnings growth since 2026.

Yushu's entry into the secondary market is not merely a milestone for a humanoid hardware maker; it has also thrust the entire embodied intelligence sector into a deep-seated struggle over valuation pricing, commercialization pathways, and technology roadmaps.

"Yushu Technology's STAR Market listing signals a profound shift in A-shares' hard-tech landscape, moving from 'digital intelligence' to 'physical intelligence'," said Tian Lihui, a finance professor at Nankai University. He noted that Yushu represents the "embodied" form of AI interacting with the physical world, filling a critical gap in the "physical AI" asset class and completing the humanoid robot piece of the puzzle.

Indeed, Yushu's initial public offering has set a record for the fastest approval on the STAR Market. From the exchange's acceptance of its IPO application on March 20 to the listing committee's approval on June 1, the entire process took just 73 days.

"The STAR Market continues to play a pioneering role in capital market reform," a person close to regulators said. "The robotics industry is a key engine for the next round of industrial transformation and economic growth, as well as a strategic high ground in global frontier technology competition."

"Yushu's A-share listing is a typical signal of domestic industrial policy and capital markets' long-term support for the physical AI track," said Kang Yu, general manager of the capital market department at Shoucheng Holdings. The Beijing Robotics Industry Development Fund, managed by Shoucheng Holdings, is Yushu's largest state-owned single shareholder.

"The listing of domestic embodied intelligence companies is an objective need at this stage of industry development," said Ding Jian, head of the high-end manufacturing research center and chief researcher at Topsperity Securities. "Embodied intelligence is a highly technology-intensive industry. Large model training, real-world data collection, continuous hardware iteration, and competition for top talent all require enormous capital. IPO fundraising is the essential financial support for sustained industry growth."

However, while the embodied intelligence track Yushu represents has vast potential, achieving large-scale commercialization in reality is extremely challenging. As Yushu's listing approaches, discussions and controversies surrounding the company and the sector have not ceased: Can embodied intelligence deliver on investors' high expectations? When will robots that can currently only perform flips achieve large-scale commercial deployment? How should valuation frameworks be established for embodied intelligence companies when valuations outpace earnings? How can the technical bottlenecks of general-purpose embodied intelligence be overcome? What will the long-term industry landscape look like, and how should Chinese robot companies build core competitiveness?

Currently, the industry consensus is that, in the long run, the embodied intelligence track will produce a trillion-yuan-scale leading enterprise. Many companies are now valued at tens of billions of yuan, but if they can grow into industry leaders, their current valuations would not be considered expensive. The core challenge lies in predicting who will ultimately become the leader, as all valuation judgments must be grounded in deep analysis of a company's fundamentals and growth potential.

Behind the scenes, since its founding in 2016, Yushu Technology's valuation in the primary market has grown more than a thousand-fold, consistently exceeding investor expectations.

"We positioned ourselves in the robot hardware track based on our group's strategic planning, invested in Yushu Technology in its Series B round, and subsequently added another round of investment. The core reason was that Yushu's overall development progress and operating results exceeded our initial expectations," Kang Yu told Caijing.

Yushu's offering market cap of approximately 61 billion yuan represents a 4,574-fold increase over its seed round valuation of about 13.33 million yuan at its founding a decade ago. During this period, the company has gathered 46 shareholders, including numerous prominent domestic internet giants, well-known market-oriented VC/PE firms, local state-owned capital, and government guidance funds.

Wang Xingxing, who founded Yushu Technology, serves as the controlling shareholder. He directly holds 86.715 million shares, accounting for 21.44% of the total post-issuance share capital. Through the differentiated voting rights of A and B shares, Wang controls 68.78% of the company's voting rights through direct and indirect holdings, and has consistently served as chairman, general manager, and chief technology officer, shouldering strategic, operational, and technical responsibilities.

The collective presence of internet giants on Yushu's shareholder roster is particularly notable. Major players including Meituan, Xiaomi, Alibaba, Tencent, and ByteDance have all invested. Meituan's affiliates are Yushu's largest external institutional shareholder, holding a combined 35.1236 million shares through Hanhai Information, GalaxyZ, and Chengdu Longzhu, representing approximately 8.68% of post-issuance shares. Based on the 61 billion yuan offering valuation, this stake is worth approximately 5.297 billion yuan.

Meituan founder Wang Xing has previously stated that robotics is one of the company's key vertical investment areas, noting that robots may be the ultimate solution for the last-mile delivery of food delivery services. Additionally, Xiaomi founder Lei Jun's Shunwei Capital fund Astrend IV, Alibaba's Hangzhou Haoyue and Shanghai Yunyang, Tencent's official strategic investment platform Tencent Technology, and ByteDance's Wuxi Jinqiu have all invested in Yushu, with holdings ranging from 2.1792 million to 16.106 million shares.

"Internet giants' cloud services, large models, and digital ecosystems all remain in the virtual world, while embodied intelligence is an important vehicle for AI's physical landing and could potentially be the next-generation universal interaction terminal following smartphones. Therefore, this track is one that major internet industrial capital cannot afford to miss," an industry insider said.

Among market-oriented venture capital institutions, several top-tier VC/PE firms have placed bets on Yushu. Sequoia Capital China, through Ningbo Sequoia and Xiamen Yaheng, holds a combined 25.8995 million shares, accounting for 6.4% of post-issuance shares, making it the largest market-oriented venture capital shareholder. Based on the 61 billion yuan valuation, Sequoia's stake is worth approximately 3.906 billion yuan. Singapore's sovereign investment fund Temasek has also invested through its dollar fund vehicle Vertex Ventures and RMB fund vehicle Xiangfeng Xiamen. Zheng Juncong, founding managing partner of Vertex Ventures, noted that he has invested in Yushu three times, describing Wang Xingxing as a technology-founder he greatly admires. Furthermore, established institutions such as Matrix Partners China, Source Code Capital, Shenzhen Capital Group, Chuxin Capital, and Lightspeed China Partners are also among Yushu's backers, with holdings ranging from 4.9493 million to 19.8491 million shares.

Embodied intelligence is also a strategically supported industry for local governments. The Beijing Robotics Industry Development Fund, managed by Shoucheng Holdings, is Beijing's municipal-level special robotics industry state-owned fund. As of before the listing, the fund held 13.928 million shares, representing 3.44% of post-issuance shares. Based on the 61 billion yuan offering valuation, the stake is worth approximately 2.101 billion yuan, making it Yushu's largest state-owned single shareholder.

"The original purpose of our fund was Beijing's plan to build a humanoid robot industry cluster. Establishing an industry fund is a capital-driven approach to support local hard-tech enterprises," Kang Yu said. "We are optimistic about the long-term prospects of physical AI, focusing on investable and deployable hardware bodies and key algorithms, and prioritizing teams that can deeply integrate with Shoucheng Holdings' own scenarios. At the same time, our investment scope is not limited to Beijing—we invest in any company valuable to the domestic embodied intelligence supply chain, and investing in Yushu follows this approach."

"For Shoucheng Holdings, a portfolio company's listing does not necessarily mean an exit point. If we judge that the company's long-term growth space remains enormous, we will choose to hold it long-term," Kang Yu added. "Our fund's positioning is long-term patient capital, accompanying enterprises throughout their growth journey while leveraging the group's offline industrial resources to fully empower portfolio companies."

The China Internet Investment Fund, a national-level internet industry state-owned fund, is also a shareholder, holding 7.6767 million shares. Additionally, China Mobile's industry fund subsidiary Zhongyi Hechuang, Haidian State-Owned Assets' Zhongguancun Science City, Shanghai State-Owned Assets' Shanghai Science and Technology Innovation, and Xiamen State-Owned Assets' Junshi Venture Capital hold 2.1792 million, 3.3886 million, 1.3637 million, and 0.6702 million shares respectively.

Notably, entering the IPO stage, Yushu has gained a significant new shareholder. Liang Wenfeng, founder of leading domestic large model company DeepSeek, participated in Yushu's strategic placement, receiving 933,400 shares. In the offline institutional placement, Liang further increased his position, with his quantitative institutions Ningbo High-Flyer Quant and Zhejiang Jiuzhang Asset Management jointly receiving 258,200 shares. As of before the listing, Liang holds a total of 1.1916 million shares through three entities.

The "marriage" between two Hangzhou "Six Little Dragons" and leading domestic AI companies has drawn market attention. At the pre-listing investor communication session, Wang Xingxing responded that DeepSeek's participation in Yushu's strategic placement was based on a signed Strategic Cooperation Memorandum. According to Yushu's latest announcement, the two companies will cooperate in three areas: joint R&D for general artificial intelligence, deep cooperation on high-performance general-purpose robots, and deep collaboration on AI large models.

This deep binding, extending from capital linkage to business cooperation, directly targets Yushu's core pain point in robot development. "Yushu Technology and DeepSeek are both located in Hangzhou, which gives them a natural geographical advantage. Yushu may leverage DeepSeek's large model capabilities to accelerate breakthroughs in the core technical bottlenecks of the embodied brain," an industry insider said.

From a performance perspective, Yushu Technology is the first humanoid robot company in China to achieve profitability. According to prospectus data, from 2023 to 2025, Yushu generated revenues of 159 million yuan, 393 million yuan, and 1.699 billion yuan respectively, with a three-year compound annual growth rate of 226.78%. Its 2025 revenue growth rate reached 332.64%. In 2024, Yushu achieved its first net profit turnaround in the industry. From 2023 to 2025, its non-GAAP net profit was -18 million yuan, 78 million yuan, and 591 million yuan respectively, with non-GAAP net profit growth of 674.29% in 2025.

Other industry peers have also maintained high revenue growth rates, though most remain loss-making. Hong Kong-listed UBTech Robotics (9880.HK) achieved 2025 revenue of 2.001 billion yuan, up 53.33% year-on-year, but recorded a net loss attributable to shareholders of 703 million yuan, narrowing by 37.42%. CloudMinds achieved 2025 revenue of 337 million yuan, up 227% year-on-year, and became the second profitable humanoid robot company in 2025, with net profit attributable to shareholders of 28.7 million yuan. Leju Intelligent achieved 2025 revenue of 258 million yuan, up 365.2%, with a net loss of 70 million yuan. Zhiyuan Robotics achieved 2025 revenue of 1.05 billion yuan, up 16.5 times year-on-year, with a net loss of 83.54 million yuan.

However, entering 2026, Yushu, as a leading company, has seen its revenue growth moderate to some extent. The prospectus shows that Yushu achieved first-quarter 2026 revenue of 423 million yuan, up 68.49% year-on-year. The company expects first-half 2026 revenue of 1.052 billion to 1.128 billion yuan, representing year-on-year growth of 35.62% to 45.41%. In terms of profitability, due to increases in period expenses such as R&D and selling expenses, Yushu's non-GAAP net profit for Q1 2026 fell to 40.2536 million yuan from 84.8365 million yuan in the same period last year, down 52.55%. The company expects H1 2026 non-GAAP net profit of approximately 236 million to 283 million yuan, down approximately 21.97% to 6.43% year-on-year, narrowing the decline compared to Q1.

Yushu noted in its prospectus that its rapid business growth is closely tied to the global surge in embodied intelligence industry enthusiasm since early 2025. As the base grows to a higher level, industry heat has gradually moderated, market competition has intensified, and the company's growth rate has slowed relatively. Yushu is expanding its robots' application scope to broaden revenue sources. According to its response to the second round of inquiries on the STAR Market, its humanoid robots serve three major application scenarios: scientific research and education, commercial consumer use, and industry applications, with scientific research and education being the largest revenue source.

In 2023, all of Yushu's humanoid robot revenue came from scientific research and education, totaling 2.9671 million yuan. In 2024, commercial consumer revenue began to emerge, with scientific research/education and commercial consumer revenue at 92.2552 million yuan and 14.6425 million yuan respectively, accounting for 86.3% and 13.7%. From January to September 2025, Yushu added industry applications focused on enterprise guided tours, intelligent manufacturing, and intelligent inspection. During this period, scientific research/education, commercial consumer, and industry application revenues were 438 million yuan, 103.5 million yuan, and 53.6 million yuan respectively, representing 73.6%, 17.39%, and 9.01% of the total.

From a revenue source perspective, Yushu's humanoid robot application scenarios still primarily rely on universities, research institutes, and AI laboratories purchasing humanoid hardware platforms for training and validating robot and AI algorithms, dependent on their research funding. Market-oriented orders from ToC commercial consumers and ToB industry applications remain relatively small, particularly orders involving real industrial scenario deployment. Its quadrupeds, with more mature technology, have a higher proportion of market-oriented orders. From January to September 2025, its quadruped robots generated 42.3%, 31.58%, and 26.12% of revenue from commercial consumer, scientific research/education, and industry applications respectively. Nevertheless, whether for humanoid or quadruped robots, revenue from real industrial scenario industry applications remains the smallest share.

In fact, across the industry, current humanoid robot shipments are mainly concentrated in specific scenarios. "Scientific research-oriented products, exemplified by Yushu, mainly meet laboratory low-cost platform construction and academic research needs. Function-oriented products for real industry needs, such as material handling and low-precision operations, are expected to enter small-batch delivery this year, but full entry into factories and households will still take time," an investor focused on the embodied intelligence industry said.

"Currently, what can be commercialized in the industry are closed-loop vertical scenario orders, which can scale up quickly with lightweight models in the short term, but single scenarios have growth ceilings. For example, commercial performance humanoid robots have high short-term heat, but after supply increases, demand faces growth bottlenecks. Companies need to continuously expand into new vertical scenarios while intensifying large model R&D to adapt to more scenarios and build second and third growth curves," Ding Jian said.

In his view, scenarios such as showroom commercial performances, factory material handling, express sorting, and power special inspection may have short-term scaling potential for humanoid robot companies. Additionally, sub-segment scenarios like UBTech's simulation humanoid robots and Wonan Robotics' tennis robots also have scaling potential.

Li Xuenan, a finance professor at Cheung Kong Graduate School of Business and director of its Industrial Policy Research Department, proposed several dimensions for judging commercialization progress: the proportion of industry client revenue, real operating time per robot, failure and human intervention situations, customer repurchase rates, after-sales maintenance costs, and how much cost robots save for customers. "Only after these data gradually emerge will people know whether robots are a novel tech product or a new production tool," she said.

Kang Yu stated that commercialization landing will happen naturally and cannot be forced prematurely. "When models and hardware capabilities iterate to the corresponding stage, application scenarios will naturally emerge. We cannot force immature projects to land just to meet capital market expectations."

In Wang Xingxing's view, the humanoid and quadruped robot market Yushu operates in has enormous market space and application prospects. In the short to medium term, its main applications will be in scientific research and cultural performance fields. As model generalization capabilities improve, in the medium to long term, the goal for humanoid robots is to enter factory workshops and ordinary households.

As the first humanoid robot stock on A-shares, establishing a reasonable valuation framework for an embodied intelligence track with vast imagination but difficult commercialization is a major focus for capital market investors. Yushu's subscription at a 61 billion yuan valuation and 219 times P/E has sparked a valuation debate over whether it is a "hardware company" or an "AI company."

The core of the dispute: if valued as a traditional hardware manufacturer, the 61 billion yuan valuation and 219 times P/E are significantly elevated—the average P/E for the general equipment industry it belongs to is 38.56 times, making Yushu's offering P/E about 5.7 times the industry average. If valued under an AI technology and embodied intelligence platform narrative, capital market pricing that focuses on long-term growth ceilings can tolerate higher valuations.

At the IPO investor communication session on August 7, an investor asked whether the company's offering valuation was too high. Wang Xingxing responded: "Based on the audited 2025 non-GAAP net profit attributable to shareholders of 591 million yuan, the P/E ratio corresponding to the offering price is 92.92 times." Under STAR Market disclosure requirements, IPO companies must disclose the diluted P/E ratio based on the lower of non-GAAP and GAAP net profit. In 2025, Yushu recorded 349 million yuan in share-based compensation expenses as non-recurring items, and its GAAP net profit attributable to shareholders was 278 million yuan. Based on this, Yushu's offering P/E is 219.23 times.

Gao Guolei, chairman of Shanghai Zhanghe Investment, said that Yushu's offering valuation certainly contains bubbles, but as long as market enthusiasm for the "first humanoid robot stock" persists, the valuation can hold. Whether it can sustain until commercial performance materializes remains uncertain.

Ding Jian believes that the combined market cap of all listed humanoid robot companies is currently less than 400 billion yuan, extremely low compared to overall AI hardware market cap. In the long run, the embodied intelligence track has quadrillion-level space, and leading companies valued at trillions or even tens of trillions of yuan could emerge, giving humanoid robot assets enormous potential for increased weighting in the secondary market. In Ding's view, the domestic pricing logic for robot body companies has been unreasonable for a long time. With Yushu and subsequent humanoid companies listing, domestic embodied intelligence companies will regain autonomous pricing power in the capital market, and the entire humanoid robot track could undergo systematic revaluation.

As the "first humanoid robot stock on A-shares," how should a reasonable valuation system be established for Yushu? The market has offered different pricing models and results. Li Xuenan provided an analytical framework: Yushu's valuation contains two layers—one is hardware operating value, and the other is AI and platform value still in growth. "The capital market is currently willing to give Yushu a relatively high valuation, essentially reflecting expectations for this future value."

"Yushu's current valuation is underpinned by hardware operating results, with AI and embodied intelligence capabilities providing future options," Li summarized. "The core of Yushu's valuation is whether it can, in the coming years, convert today's 'future price' paid by the capital market into customers, orders, data, profits, and cash flow one by one."

Ding Jian believes that compared to new energy vehicles, the smartphone industry provides a better analogy for explaining the valuation logic of the embodied intelligence industry. Topsperity Securities' valuation model calculates hardware and software separately: assuming cumulative global humanoid shipments of 15 million units by 2035 with 5 million added that year, an average hardware price of 100,000 yuan per unit, and 2 times P/S benchmarked against Apple for hardware; for software, referencing Tesla's FSD monthly subscription of $500, 30 times P/ARR; also referencing the 20% market share of leading global smartphone companies and software market penetration, under neutral assumptions, the 2035 global leader's hardware valuation would be 200 billion yuan, with combined hardware and software valuation of 470 billion yuan. Discounted at 7% to 2026, the combined valuation corresponds to approximately 255.6 billion yuan. Under optimistic scenarios, the 2026 industry leader's discounted valuation approaches 400 billion yuan. Adjustments to model parameters such as hardware P/S multiples and software subscription pricing would cause enormous fluctuations in results.

Currently, institutions' valuations for Yushu also differ significantly. CCB International believes the market generally expects Yushu's market cap to be revalued to between 60 billion and 100 billion yuan after listing. Based on brand premium and peer company valuations, it forecasts a 2026 target P/S of 32 times, corresponding to a market cap of 109 billion yuan. Some market institutions provide optimistic expectations of 200 billion to 300 billion yuan based on long-term industry space.

Notably, during the IPO inquiry phase, Yushu's lead underwriter CITIC Securities provided a reasonable valuation range of 50.6 billion to 55.9 billion yuan for 6 to 12 months after listing—nearly 10% below its 61 billion yuan offering market cap. Previously, the market estimated Yushu's offering price at approximately 104 yuan per share based on fundraising amounts, but the actual offering price came in 45% higher. These contradictory data points reflect pricing divergence between the primary and secondary markets, and between professional institutions and market sentiment.

Whether embodied intelligence companies like Yushu can overcome their core risks is a key factor affecting the industry's valuation changes. The market generally believes the robot industry is at a critical inflection point from "Demo to Product," with commercial deployment being a major core risk. Gao Guolei believes that for Yushu to generate truly commercialized revenue, third parties could provide physical simulation training systems, industrial enterprises could provide deployment scenarios, and together they could build mature, reliable robots suited to specific scenarios. Only through deployment in concrete application scenarios will Yushu's orders become sustainable.

Pressure from competitors also cannot be ignored. Yushu is currently in the industry's first tier, but the landscape is far from solidified. Domestic and international competitors such as Tesla, Zhiyuan, and UBTech, as well as cross-industry players like Xpeng and Xiaomi, could all threaten Yushu's market position. Additionally, another major risk Yushu faces is what it candidly acknowledged in its prospectus: the problem of "heavy body, light brain." Yushu is clearly aware of this shortfall. In its IPO fundraising projects, it plans to invest nearly half of the proceeds into embodied brain development and has introduced DeepSeek as a strategic placement investor, seen as a critical step in filling the "brain" gap.

"Different segments of the embodied intelligence supply chain have different valuation logic. For a body company like Yushu, which has already achieved profitability leveraging the domestic supply chain, its listing will make it a universal valuation benchmark for the entire industry," Kang Yu said. "In the next year or two, multiple embodied intelligence companies will enter the capital market. Post-listing stock price fluctuations will directly reflect capital market expectations, testing companies and founding teams' mentality more than primary market valuations, and their long-term comprehensive strength will ultimately be reflected in market cap."

On the industry bottleneck front, overcoming technical, cost, commercialization, and regulatory standardization barriers to truly bring robots into millions of households and realize the quadrillion-level track potential is the core concern for the entire embodied intelligence supply chain and investors. Kang Yu said that for the embodied intelligence supply chain, vertical scenario application deployment will be rapid in the short term, but the popularization of full-scenario general-purpose humanoid robots still has a long development cycle.

At the technical level, what is the biggest bottleneck to achieving general-purpose embodied intelligence? Ding Jian identified physical interaction data collection and standardization challenges, embodied-specific large model training iteration efficiency, and whole-machine hardware iteration speed as the three core constraints. He emphasized that data is currently the industry's most critical bottleneck. Compared to autonomous driving and text large models, embodied intelligence requires real physical action interaction to generate effective training data. Currently, the industry lacks unified data standards, with severe data silos, directly constraining embodied large model iteration speed.

Kang Yu believes physical AI must combine real machine physical interaction data to significantly improve model performance. The industry currently has two approaches to data collection: one centered on human collection, the other on robot bodies collecting physical interaction data. Miao Tianyi, executive partner at Puzhuo Capital, said: "Simply excelling at motion control is no longer sufficient to build long-term barriers; the AI brain is the core of next-generation product iteration. The deep integration of software and hardware is the biggest technical bottleneck at this stage. Hardware bodies and AI large models cannot simply be stitched together; they require deep joint optimization."

As technology iterates, how will humanoid robots land in real scenarios? PricewaterhouseCoopers believes that considering the dynamic changes in value and cost, China's intelligent robot large-scale application will evolve in three waves: the first wave from 2026 to 2028, landing first in high-risk heavy physical labor and standardized commercial service scenarios; the second wave from 2029 to 2032, gradually penetrating repetitive service scenarios such as supermarkets and retail, though large-scale substitution still requires simultaneous maturation of technology and regulatory systems; the third wave after 2032, with the pace of substitution in semi-skilled manufacturing scenarios depending on technological breakthroughs and cost reduction, with no clear timeline in the short term.

Li Xuenan predicts that embodied intelligence will first land in scenarios with clear tasks, controllable environments, and calculable ROI, such as manufacturing, warehousing and logistics, energy inspection, and hazardous chemical operations. In contrast, home environments are highly non-standard, and combined with triple barriers of safety, cost, and generalization capability, large-scale household use will require a longer cycle.

"By comparison, overseas companies like Tesla and Figure AI pursue a high-profile approach, directly investing heavily in general-purpose embodied large models with the goal of achieving full-scenario general-purpose humanoid robots in one step, making short-term deployment difficult. Domestic companies adopt a fast-follow approach: first landing vertical scenarios to generate cash flow, then iterating into general-purpose humanoid platforms. This path better aligns with objective industry development laws," Ding Jian said.

In his view, the richness of domestic industrial scenarios far exceeds that of overseas markets. Industrial, exhibition, education, and special operations tracks can be deployed simultaneously, allowing faster data accumulation. As vertical scenario orders continue to scale, the market will re-recognize the rationality of this development path.

"Currently, both domestic and overseas embodied intelligence model vendors have not reached maturity. The industry is still in its early development stage. The domestic core advantage is the abundance of offline physical scenarios, enabling rapid accumulation of physical interaction data, while local governments and capital markets will support the track with policy resources," Kang Yu said.

As the industry moves toward a more mature future, establishing standardized evaluation criteria becomes the next focus. "In the coming years, the biggest change in the embodied intelligence robot industry will occur in evaluation standards," Li Xuenan said. In June 2026, the Ministry of Industry and Information Technology and the State-owned Assets Supervision and Administration Commission launched the 2026 Humanoid Robot and Embodied Intelligence Real-Scene Training Special Action, promoting robots into real production and living environments, building a closed loop of "real-scene training - data accumulation - product iteration - large-scale deployment," and pushing key products to begin "operation mode" in representative scenarios by year-end. Li believes this action indicates that industry evaluation standards are shifting.

Currently, the embodied intelligence industry is still in a multi-point flowering pattern. Yushu is the first to list on A-shares, capturing a leading position to some extent, but the industry landscape is far from formed. Who will ultimately become the leading enterprise in the trillion-yuan embodied intelligence track? Industry observers believe the biggest variable may come from Tesla. Its Optimus robot, though not yet commercially sold externally, could become a formidable competitor to Yushu given its technical strength and talent pool. Among domestic companies, Zhiyuan Robotics, valued at over 20 billion yuan, is considered ahead of Yushu in "brain" level technical accumulation. UBTech has already listed on the Hong Kong Stock Exchange, and its Walker S series has entered automotive manufacturing and logistics factory scenarios. Xpeng and Xiaomi have also entered the robot track cross-industry.

"Yushu is in the domestic first tier for hardware and motion control capabilities, with clear first-mover advantages in commercialization and capital market progress, and prominent brand and talent aggregation effects. But the industry is still in a state of diverse development, with Yushu, Zhiyuan, and UBTech each having advantages in different sub-segments," Ding Jian said.

Regarding the possible long-term evolution of the embodied intelligence landscape, he said two types of ecosystems may form: one benchmarking Apple—closed-source integrated hardware vendors that use self-developed whole machines plus proprietary large models, charging through differentiated skill subscriptions; the other benchmarking Android—open-source humanoid brain platforms where third-party hardware vendors can access general-purpose foundation models and win through differentiated hardware. Both paths will coexist long-term, with neither achieving absolute dominance.

For Chinese embodied intelligence robots at the critical juncture of transitioning from "technology vision" to "industrial reality," how should they build core competitiveness? Institutions believe China's intelligent robot industry possesses three strategic assets: the world's most complete industrial supporting system lowering entry barriers; policy and capital dual empowerment supporting large-scale investment; and massive manufacturing scenarios and data providing fertile ground for technological upgrades and commercialization.

Miao Tianyi believes companies should first hone their hard skills internally before supplementing software collaboration, externally use scenarios to build barriers, while properly planning patents, using capital rationally, focusing on R&D and production lines, and balancing technical ideals with business reality in a long-cycle industry.

Li Xuenan believes China's robot industry's previous advantage was its ability to build complex machines and continuously drive down prices. The next stage of competition is about getting robots into real work sites—factories, warehouses, power grids, and mines—where they genuinely work daily. China's complete and massive manufacturing base itself is a globally scarce embodied intelligence training ground—data generated in operations, products iterated in scenarios, and costs reduced through application. Once the "scenario - data - product - cost" loop runs smoothly, China's manufacturing advantage today will ultimately grow into an industrial moat in the embodied intelligence era.