Summary of Key Points
Yushu and Zhiyuan are two leading companies in the field of embodied intelligence (robots that can perceive, move, and perform tasks like humans) in China, yet they have taken completely different approaches: Yushu, similar to Li Auto, focuses on creating exceptional products (robot bodies + motion control) to capture the market and has subsequently listed on the A-share stock market after achieving profitability; Zhiyuan, akin to NIO, relies on a platform ecosystem (large models + comprehensive industrial chain layout) to secure its future prospects. Although Zhiyuan has not yet turned a profit, it is growing at an impressive rate and plans to list on the Hong Kong Stock Exchange. Both companies are now seeking to complement each other's strengths (Yushu is developing large models, while Zhiyuan is leveraging its motion control capabilities for sales). They both face common challenges such as talent mobility, technology integration, and difficulties in finding practical applications for their robots.
I. Origins: The Product Kingdom of a Wild Genius vs. the Ecological Empire of a Established Player
Yushu can be seen as the personal project of its founder, Wang Xingxing, who holds three roles (Chairman, General Manager, CTO) and controls 65% of the voting rights through A and B shares. The company has a simple structure with only three core researchers and a team of 175 people, focusing on developing leading motion control technology (even American laboratories purchase their robots). Just as DJI perfected drones, Yushu targets the market with its dual approach of quadrupedal and humanoid robots, following the same strategy as Li Auto, which achieved success with a single product.
Zhiyuan, on the other hand, is part of Huawei’s ecosystem. Founded by Deng Taihua (former president of Huawei's Ascend business unit), it brings together four partners with Huawei backgrounds, along with popular blogger Zhihui Jun (with AI expertise). The team has a more extensive scope, having spun off five subsidiaries in areas such as dexterous hands and data collection, and has invested in 52 industrial chain projects with Hillhouse Capital. Similar to NIO’s battery swapping and user community initiatives, Zhiyuan aims to create a network that covers the entire industry.
II. Business Strategies: Profit through Sales vs. Building an Industry-Wide Network
Faced with the issue that robots are mainly used in exhibitions and laboratories, Yushu has chosen a practical approach: it reduced the price of its humanoid robot G1 from 99,000 to 85,000 yuan, and the R1 model to below 30,000 yuan. Its quadrupedal Go series is also priced in the tens of thousands of yuan range. The company even made an appearance on the Spring Festival Gala to increase its visibility and has become a popular choice for rental businesses.
Zhiyuan, however, is betting on the future. Its subsidiary Qingtian Rent operates robot rentals in 13 countries, while Mifeng Technology collects data on robot movements. Zhiding Robot has already started making profits in commercial cleaning applications. The company has also established an industrial fund to invest in the entire chain, aiming to position itself strategically when the integration of “embodied intelligence” and GPT technologies becomes a reality.
III. Capital Paths: A-share Market for Profitability vs. Hong Kong Stock Market for Growth Potential
Yushu’s financial performance fits the typical profile of a product-oriented company: revenue grew from 159 million yuan in 2023 to 1.699 billion yuan in 2025 (a compound annual growth rate of 226%), with gross margin increasing from 44% to 60%. In 2025, its net profit after deducting non-recurring expenses was 591 million yuan (after a loss of 18 million yuan in previous years). As such, Yushu chose the A-share Science and Technology Innovation Board, which requires profitability and technological barriers. Its initial public offering resulted in a share price of 845 yuan on the first day of trading (with an issue price of 150.8 yuan), generating a profit for early investors.
Zhiyuan, being a fast-growing company, had revenue of only 300,000 yuan in 2023 but aims to reach 1.05 billion yuan by 2025 and over 1 billion yuan in the first quarter of 2026 (with an annual target of 4 billion yuan). Despite still being in the red, it plans to list on the Hong Kong Stock Exchange under Chapter 18C, which allows unprofitable tech companies to go public. The company expects a valuation of HK$40-500 million, reflecting the capital market’s interest in its innovative ecosystem.
IV. Common Challenges and Areas of Convergence
- Complementing Each Other’s Weaknesses: Yushu is investing 2 billion yuan from its IPO funds in developing large models and open-sourcing WMA and VLA models, testing tasks like organizing meetings autonomously; Zhiyuan is also selling robots, with a particular focus on their motion control capabilities that are valuable for performances (it shipped 8,400 units in the first half of 2026, accounting for 44% of global sales).
- Talent Mobility: Both companies are facing challenges with employee turnover. Yushu’s two equity incentive programs affected 9 out of 23 employees; Zhiyuan lost its chief scientist, Dr. Luo Jianlan (a former researcher at DeepMind). Talents from Li Auto, Huawei, and NIO are moving to the embodied intelligence sector, where salaries tend to increase significantly.
- Technology Integration: The integration of large models (the “brain”) and motion control systems (the “small brain”) is still in its early stages. There isn’t much difference between the technologies used by different companies, but training data remains a key factor, with larger firms having an advantage.
- Finding Practical Applications: Robots are mainly used in exhibitions and research labs; there is a need to develop real consumer or industrial use cases for them, similar to the long journey from early electric vehicle development to widespread adoption.
Conclusion
The competition between Yushu and Zhiyuan boils down to two different visions: one focuses on refining a single technology to perfection, while the other aims to create a comprehensive industry-wide network. It’s too early to determine a winner, but their converging strategies and shared challenges highlight the critical transition of the embodied intelligence industry from experimental development to practical commercialization. The company that first successfully integrates robots into everyday life could become the next big player in this field, similar to DJI or Tesla.