Summary of Key Points
Over the past 20 months, the field of embodied intelligence (robots that can perceive and act like humans) has experienced a rapid shift through 15 different focal areas of industry interest, known as "narratives." These have evolved from robot performances showcasing impressive skills at the beginning of 2025 (such as performing the Yangko dance on Chinese New Year's Eve TV shows and generating daily rents in the tens of thousands of yuan), to focusing on fundraising and securing orders in the middle of the year (where a PPT was enough to attract investment, leading to questions about the actual number of orders), to a race for mass production by the end of the year (with domestic shipments accounting for 84.7% of the global total). In 2026, the focus has shifted to "technical capabilities" (such as physical AI and world models), "data collection" (with data acquisition companies raising 17 billion yuan in funding), "commercial validation" (with robots being deployed in logistics and automotive applications), and most recently, to the "IPO rush" (with Yuzhu becoming the first company in the sector to list on the A-share market). Behind this rapid change is a combination of technological breakthroughs (large models that enable robots to understand human language), a surge in capital investment (over 90 billion yuan in the first half of 2026), and simultaneous experimentation in various use cases (industrial, retail, and logistics sectors). However, there is also a contradiction between capital's desire to quickly realize profits and the industry's need for time to mature—essentially, the decade-long evolution of traditional industries has been compressed into just 20 months.
Detailed Analysis
The Rapid Succession of Narratives: From "Yangko Dance" to "IPO"
The focus in the embodied intelligence sector has been changing as quickly as flipping through pages of a book:
- Early 2025: The stage of showing off technical capabilities. Robots attracted attention through performances; Yuzhu robots performed the Yangko dance on Chinese New Year's Eve TV shows, and robots were rented out in shopping malls and tourist attractions for promotional purposes (with daily rents in the tens of thousands of yuan). However, this was soon criticized as "PPT-based fundraising" (i.e., obtaining funding without a physical product).
- Mid-2025: The stage of practical implementation. Investors began to ask about the presence of actual orders. Zhiyuan and Yuzhu won contracts worth 120 million yuan from China Mobile, while UbiSelect secured a 250 million yuan contract for the world's largest humanoid robot project, marking a shift from robots being viewed as performers to tools for practical work.
- End of 2025: The stage of mass production. Orders drove the need for mass production; Zhiyuan produced 5,000 robots, and Yuzhu sold 5,500 units, with domestic shipments accounting for 84% of the global total. Chinese manufacturers now dominate the top six in the global market.
- Early 2026: The focus shifted to more advanced technologies. The focus moved from the physical design of the robots to their "brains." Jeff Bezos initiated a project on physical AI, and Sam Altman declared the arrival of "ChatGPT for physical AI." National funds invested 2.5 billion yuan in Galaxy General, and UbiSelect offered annual salaries of 124 million yuan to attract top scientists.
- Mid-2026: The emphasis on data and commercialization. It became clear that robots' "brains" needed more data, leading data acquisition companies to raise 17 billion yuan in funding. Xingdongjiyuan's logistics robots were put into regular use by SF Express, proving their profitability.
- End of 2026: The rush for IPOs. Yuzhu went public on the STAR Market with a market value of 444.9 billion yuan, and Meikamander underwent an IPO review on the Hong Kong Stock Exchange. Multiple companies were in line to go public—just one and a half years after the start of this trajectory.
Key Point: These narratives are not mutually exclusive but rather build upon each other. For example, robots still perform, but now it's more about using them for practical tasks, with mass production becoming the new norm.
Why Such a Rapid Pace?
The reason for this rapid progress is the combination of three factors:
- Technological Advancements: Large models have enabled robots to understand human commands, and world models have helped them understand physical principles. Decisions that used to take years for traditional robots to make can now be made in just two years.
- Capital Momentum: In the first half of 2026, 93.5 billion yuan was raised in funding, three times the amount invested in the entire year of 2025. Investors are pressuring companies to innovate quickly; otherwise, they risk losing their investment.
- Support from Supply Chains and Use Cases: Robot prices have dropped significantly, from 600,000 yuan per unit in 2023 to 160,000 yuan in 2025, with some models even available for as little as 9,998 yuan (nearly reaching consumer-grade prices). Simultaneous experimentation in various industries has helped identify which applications are profitable.
The Contradiction between Capital and Industry
The core issue is the gap between capital's expectations and the industry's actual progress:
- Capital Expectations: Yuzhu's revenue in 2025 was 1.7 billion yuan, but its market value upon listing was 444.9 billion yuan (a price-earnings ratio of 260 times), reflecting investors' belief that Yuzhu would become the "Apple of the robotics industry" in ten years.
- Industry Reality: Companies are still struggling to achieve monthly shipments of over a thousand units and to prove that robots can save costs for their customers.
- Consequence: Without immediate financial results, companies rely on new narratives (such as the importance of data or commercial validation) to meet investors' expectations, leading to frequent changes in industry focus.
Evolution in Thinking: From "Building Robots" to "Making Robots Useful"
Each shift in narrative reflects a deeper understanding of what is truly important in this industry:
- Initial Phase: Competition focused on who had the most impressive hardware.
- Mid-Phase: The focus shifted to securing orders and practical implementation.
- Later Phase: The emphasis was on the technical capabilities of the robots' "brains."
- Current Phase: The ultimate goal is to determine who can generate profits and create a sustainable business model.
Conclusion: The frequent changes in the embodied intelligence sector are not a sign of confusion but rather a result of the industry accelerating its development. The decade-long process of trial and error, validation, and improvement has been compressed into 20 months, with each step bringing us closer to the goal of integrating robots into our daily lives.
Final Remark
Embodied intelligence is like a rapidly growing teenager: Capital is pushing it to mature faster, technology is enabling it to make leaps forward, and various use cases are providing opportunities for rapid experimentation. Although the pace may seem intense, each shift in narrative represents another step towards its full potential and maturity.