During a recent interview at the 2026 Fund of Funds Annual Forum and the 7th Lujiang Venture Capital Forum, Dunhong Asset partner Yu Wenchao discussed the company's investment journey with Unitree Robotics and his views on the current market dynamics.

After Unitree Robotics went public on the STAR Market in August 2026, its stock price experienced a notable pullback after the first day's market cap surged past 400 billion yuan. Yu believes that embodied intelligence could represent the largest-scale intelligent terminal in human history, stating that "the market shouldn't be too harsh" and that Unitree's current valuation remains undervalued given the expansive future market potential.

Yu pointed to Unitree's successful listing as proof that embodied intelligence is not a pseudo-demand. He noted that the application scenarios predicted for quadruped robots have all been realized one by one, giving him strong conviction in the parallel path ahead for humanoid robots. The trajectory, he predicts, will unfold from research, education, and entertainment markets first, then expand into industrial scenarios, and ultimately reach households across the country.

What Makes Unitree Different

Yu first noticed Unitree during the 2021 Spring Festival Gala when robot dogs appeared on stage as backup dancers for singer Liu Dehua. Spotting the "Hangzhou Unitree Technology" label on screen, he visited the company personally and by the second half of 2021, Dunhong Asset had led the company's Series B+ round. Today, Dunhong remains one of Unitree's top ten shareholders.

What impressed Yu most was Unitree's comprehensive approach to self-developed core technology. Recalling the robotics surge around 2020, Yu noted that many companies in the sector lacked truly proprietary technology—most were simply "assembling components and integrating systems." Unitree stood apart, "already holding a large number of core technologies in its hands." He highlighted the extraordinary difficulty of the joint modules in humanoid robots, which must pack motors, controllers, reducers, and sensors into extremely tight spaces while maintaining strong performance, low power consumption, and manageable costs.

Despite the recent share price decline, Yu remains bullish, arguing that the market needs to give humanoid robots time to take root across various industries. He drew parallels with the quadruped robot market breakthrough, which also began with research and education before expanding into industrial applications like power grid inspection and emergency rescue. "The path for humanoid robots is exactly the same as for quadruped robots—first opening the market through research, education, and entertainment, then exploring industrial scenarios. Unitree has already launched collaborations with major manufacturing leaders like Great Wall Motor and Geely, but full-scale implementation simply takes time."

Yu elaborated on the market potential, stating that the research and education segment alone represents a ten-billion-yuan market. "The broader opportunities in manufacturing and commercial services could bring hundreds of billions or even trillions in market size. If embodied intelligence eventually enters every household, we could be looking at a trillion-yuan or even tens-of-trillions-yuan super terminal—equivalent to smartphones or new energy vehicles, possibly the largest intelligent terminal in human history."

Investment Philosophy: 'After Investing in the Chain Leader, You Gain a God's-Eye View'

Dunhong Asset's strategy in the embodied intelligence supply chain extends well beyond Unitree. The firm invests first in the ecosystem leader, then deploys capital into core nodes of the supply chain—a method Yu describes as gaining "a god's-eye view" after securing the chain leader position, allowing the team to identify which segments have genuine barriers to entry.

One notable example was their early bet on Dijia Robotics, a chip startup that spun off from Horizon Robotics. "Nvidia monopolizes the end-side computing chips used in humanoid robots, but domestic manufacturers can't rely on them 100%—they need domestic alternatives. When Dijia's chips were still in development but already performing well in Unitree's testing, we placed a heavy bet in the first round of financing."

Regarding current valuations in the embodied intelligence sector, Yu was direct that "there is certainly some bubble in the primary market." Some companies with strong execution have matching valuations, but others are merely riding the wave. "We're in a period where the wind is blowing—some are eagles, some are pigs. Ultimately, industrial application will separate the two."

On the timeline for large-scale industrialization, Yu believes it must be viewed sector by sector. Some industries have already taken the lead: automotive manufacturing, semiconductors, biomedicine, and display industries are seeing batch orders today. However, comprehensive market-wide adoption may still take three to five years.

The Core Bottleneck: The 'Brain'

When identifying the critical variable that will determine further deployment of humanoid robots, Yu points to the brain as the current bottleneck. "The technical path isn't unified yet, but physical AI, world models, and data collection are all iterating rapidly. In many ways, this stage resembles the GPT-1 and GPT-2 era before ChatGPT emerged. I believe within about three years, we should see a major breakthrough."

Reflecting on the Transition from Mobile Internet to Hard Tech

Dunhong Asset originally built its reputation in TMT investments. Founded in 2015 during the peak of mobile internet, the firm backed notable projects including Mango TV, NetEase Cloud Music, Ximalaya, and Qihoo 360. But when the mobile internet dividend dried up around 2018, Dunhong—like many peers—experienced a period of uncertainty.

While some investment firms pivoted to consumer brands, Yu quipped that their team "is all straight men who can't tell the difference between one milk tea and another. Investing in consumer requires a certain sensibility—we didn't have that ability, so we decisively gave up."

Instead, Dunhong turned toward hard tech, initially investing in the underlying technology providers for platform companies like Mango TV—investments such as Danghong Technology and ArcSoft. As these companies listed successfully, Dunhong's hard tech investment cycle moved into positive reinforcement.

To date, Dunhong Asset manages over 16 billion yuan in assets, and among more than 80 direct investment projects, 20% have achieved IPO listings.

Consensus vs. Non-Consensus Investing

On whether truly excellent investments come from consensus or contrarian thinking, Yu argued that both vision and resources matter. For early-stage projects, investment teams need strong judgment capabilities. "Startup teams aren't always polished at capital operations—if you discover them early enough, you have a chance to get in." For teams with star founders or established reputations, what gets tested is the firm's industrial resources.

Yu noted that many of Dunhong's limited partners are listed companies from the industrial sector. "They join our funds precisely for AI and embodied intelligence market positioning—some are even scene providers up and down the supply chain. Beyond that, we offer strong value-added services, building synergy in industrial orders, promotional resources, and more."

Yu also reflected on the pessimistic "the primary market is dead" narrative from recent years, acknowledging that the emergence of DeepSeek, the rise of Unitree, and improving secondary markets have helped restore confidence. However, he conceded that as market sentiment shifts, hard tech valuations have risen, decision-making windows have shortened, and Dunhong has become more cautious in deal-making since the second half of last year. Looking ahead, a large number of projects entering their exit phase next year presents challenges, particularly given unpredictable market conditions.

Yu summarized his outlook with cautious optimism: the embodied intelligence story is real, the chain is forming, and value creation takes time—but those who stay disciplined through market cycles will see the returns.