A few days ago, under global human scrutiny, what was expected to be a highly technological World Humanoid Robot Championship instead produced numerous memorable moments: a robot participating in long-distance running knocked down a staff member and continued running as if nothing happened; during a boxing match, while staff weren't paying attention, a robot circled behind and started throwing punches; when descending stairs, a robot got stuck and even lost its head in the fall...

But don't think these humanoid robot athletes were just there for comic relief. From humanoid robots performing traditional Chinese dances on the Spring Festival Gala stage earlier this year to riding bicycles, charging themselves, switching freely between wheels and feet, and delivering impressive performances at various competitions - these robots may already represent the highest level of domestic robotics in China.

Since the beginning of this year, A-share robot concept stocks have risen nearly 60%. UBTECH ROBOTICS, Unitree Robotics, and Zhiyuan Robotics have successively secured major commercial contracts, all indicating that China's embodied intelligence robots are experiencing a rapid breakthrough phase in both technology and capital markets over the past six months.

**Capital Pursuit: Embodied Intelligence "Red Hot" - Robot Concept Stocks Up Nearly 60% in Half Year**

Embodied intelligence refers to intelligent agents with physical carriers that accumulate knowledge and skills through perception, control, and autonomous learning in a series of interactions, forming the ability of intelligent agents to influence the physical world. As humans moved from accessing ChatGPT through the internet, passing through the AI smartphone trend, the most cutting-edge physical carriers for embodied intelligence today are undoubtedly robots.

It is precisely because of embodied intelligence that emotionless machines have transformed into intelligent agents that can listen, move, and express themselves, allowing capital markets to see opportunities in the secondary market.

On December 29, 2023, UBTECH ROBOTICS listed on the main board of the Hong Kong Stock Exchange. After listing, it created a single-day surge myth of 88% in March of the following year, with stock prices once touching a high of HK$328 per share. However, the stock price subsequently fluctuated downward, touching a low of HK$40.8 per share in January this year.

High-tech sectors all experience cyclicality, and UBTECH ROBOTICS has noticed this. In its 2025 financial budget report released in April this year, it revealed plans to increase R&D expenses to RMB 540 million, compared to RMB 478 million and RMB 490 million in the past two years. Meanwhile, management expenses will be further compressed to RMB 350 million.

In A-shares, the robot concept has risen 58.55% in eight months from less than 4,000 points in January this year, reaching a new high of 6,273.29 points on August 20th. Among the top three constituent stocks by market value, Foxconn Industrial Internet's total market value approaches one trillion yuan, with stock prices rising 2.51 times from this year's first-half lows. Midea Group, ranked second by market value, maintains stock prices around 70 yuan with annual R&D investment around 15 billion yuan, the highest among the top three. Luxshare Precision, ranked third by market value, has also returned to the 40 yuan range after experiencing volatile markets earlier this year.

Notably, ChiU Composite, a STAR Market listed company, gained 10 consecutive daily limit increases after Zhiyuan Robotics became a major shareholder, setting a new record for consecutive "20cm" limit-ups in A-shares.

In the primary market, unlisted star companies have also gained capital favor. In June this year, Unitree Robotics completed Series C financing, with industry giants including China Mobile, Tencent, and Alibaba participating, bringing the company's post-financing valuation to 12 billion yuan. Its founder revealed that the company expects 2025 revenue to exceed 1 billion yuan and has maintained profitability for five consecutive years since 2020, with gross margins consistently above 50%.

The robot championship is not only a arena for technological prowess but also a barometer for capital markets. Data has proven that the robotics industry is welcoming a new round of rapid development. In the robot mass production year, more robots are landing in factories, beginning new "evolution" training journeys to prepare for further volume expansion.

**Scale Expansion: Nearly 900,000 Related Companies - Concepts Converting to Productivity**

"Embodied intelligence" was written into the government work report for the first time this year, listed alongside 6G and quantum technology as future industry tracks, indicating that China, as the world's largest robot market and manufacturing center, is welcoming a new round of high-speed growth.

According to statistics released by Morgan Stanley in June this year, China's robot market reached $47 billion in 2024, accounting for 40% of the global total. The analysis firm predicts this figure will grow to $108 billion by 2028, with a compound annual growth rate of 23%.

The report states that drones, mobile robots, and collaborative robots will become major growth engines, with compound annual growth rates of 20%, 35%, and 46% respectively. China's manufacturing robot density has jumped from 97 units per 10,000 workers in 2017 to 470 units per 10,000 workers in 2023, nearly a 5-fold increase.

Behind this is a complete robot industry chain already constructed in China. According to Tianyancha Professional Edition data, China currently has over 891,000 robot-related companies in business and continued operations. Among these, approximately 91,000 new related companies have been registered so far in 2025. From enterprise registration trends, robot-related company registrations have shown year-over-year growth over the past five years.

Unlike past development paths for LED and new energy vehicle industries, embodied intelligence's uniqueness lies in requiring both hardware breakthroughs and software revolution, meaning its future is not entirely determined by capital but by the speed of technological iteration. Therefore, more enterprises means more refined industrial division of labor, benefiting overall market expansion and economies of scale.

Consequently, we see that whenever DeepSeek, iFlytek intelligent voice packs, and other large models and applications upgrade, robot "brain power" leaps forward; whenever companies like Zhongda Lide and Orbbec break through technical barriers in domestic component substitution, costs drop significantly.

Notably, in many regions, the robotics industry is a key investment attraction project. In areas like the Pearl River Delta and Yangtze River Delta, companies at all industry chain levels are accelerating clustering, making the physical distance from core component production to use almost negligible. This makes every test and component replacement extremely fast in iteration speed and trial-and-error costs.

With confidence built on China's industrial foundation, embodied intelligence has rapidly transformed from an emerging concept into tangible and touchable productivity.

**Industry Pain Points: Commercial Monetization Still a Bottleneck - Key is Solving Standardization Issues**

Since embodied intelligence robots are so popular, can we buy them for home use? Of course. A few days ago, embodied intelligence robot manufacturers brought their products to e-commerce platforms for sale, priced from 13,000 to 450,000 yuan.

These listed products include humanoid and quadruped robot products, focusing on entertainment performances, education and research, and industrial-grade applications.

In the business world, any capital investment pursues profit maximization. But the biggest problem currently plaguing China's embodied intelligence robot sector is commercialization. After all, when you calm down, you realize that spending tens of thousands of yuan on a robot that can only clumsily fold clothes, or a robot dog that can do backflips but struggles to bring you a cup of water, makes you ultimately save the money.

However, in B2B scenarios, leading companies including UBTECH ROBOTICS, Unitree Robotics, and Zhiyuan Robotics have recently secured large orders. In June this year, Unitree Robotics and Zhiyuan Robotics jointly won China Mobile's subsidiary's 2025-2027 humanoid biped robot OEM service procurement project, with a total package value of 124 million yuan, setting a new record for the largest single order for domestic humanoid robot companies at the time.

But this is ultimately only in highly standardized scenarios like industrial manufacturing and exhibitions. To deeply cultivate segmented markets and form "small but beautiful" ecological layouts requires further expansion and breakthrough.

After all, humanoid robots are highly complex systems, and solving "hand" and "brain" coordination requires considerable time. The technical principle is that achieving large-scale matching of hardware and algorithm design requires stable and low-cost combination of components from different suppliers, and forcing supply chain maturity requires time.

Additionally, due to the current diversity of humanoid robot types, manufacturers are all collecting data, but data cannot be shared, resulting in insufficient resource utilization efficiency.

UBTECH ROBOTICS Chief Brand Officer Tan Min mentioned in an interview that currently, humanoid robot manufacturers mainly design themselves, find suppliers to process components according to requirements, then assemble. "Due to non-unified component standards and different designs and requirements from different humanoid robot manufacturers, suppliers find it difficult to profit from customizing small quantities of non-standardized components; even if factories can produce needed parts, final assembly still faces challenges."

Similar to many industries in early development, standardization deficiency hinders commercialization progress. Currently, humanoid robot software and hardware industry standards are unclear, the industry chain is chaotic, unable to achieve standardized production, naturally precluding large-scale application in the short term.

Orient Securities research also points out that unlike relatively standardized industrial scenarios, service sector work has high non-standard degrees, requiring high understanding and generalization capabilities from robots. Most service robot work targets are humans, requiring stricter legal and regulatory requirements and more validation.

In the future, as technology matures, costs decrease, scenarios expand, and production increases, these challenges will be gradually resolved, and embodied intelligence robots will profoundly change existing production and service models. This is a gradual process.

**Expert Commentary: Embodied Intelligence Potential Self-Evident - Key is Breaking Through Scenario Limitations**

Looking back over the past decade, the tech circle has never lacked "hot spots": VR glasses, virtual currency blockchain, metaverse... When it comes to embodied intelligence, will it suddenly "flame out" like many previously promising sectors?

In the view of Lin Yonghua, Vice President and Chief Engineer of Beijing Institute for General Artificial Intelligence, large-scale deployment of embodied intelligence robots requires greater technological breakthroughs and needs to prove important value propositions: First, whether it can be smarter than humans, completing tasks human intelligence cannot accomplish. Second, whether it can be more efficient and durable than humans. Third, whether it can complete tasks in scenarios humans find difficult to reach, such as dangerous environments. It must occupy at least one of these to truly convince the industry to deploy.

Obviously, once robots break through hard-coded logical algorithms and execution under embodied intelligence empowerment, beginning not just to handle their respective tasks but also sensing surrounding environments and making reasoning and decisions based on large models, it will undoubtedly bring a new round of productivity revolution.

However, Tencent Chief Scientist and Robotics X Lab Director Zhang Zhengyou believes that although large models have achieved breakthroughs, putting them on robots cannot immediately achieve autonomy. He believes true embodied intelligence must be able to autonomously learn and handle problems, automatically adjusting and planning for environmental changes and uncertainties.

Zhang Zhengyou mentioned four problems that must be solved for embodied intelligence breakthrough. First, current large models only include vision and hearing, lacking tactile sense; tactile sense is very important as part of robots' complex perception capabilities, enabling them to perceive and understand unpredictable, unstructured environments and objects around them. Second, powerful execution capabilities, including movement, grasping, and operation, to interact with environments and objects. Third, learning ability to learn and adapt from experience and data to better understand and respond to environmental changes. Fourth, adaptive ability to autonomously adjust actions and strategies to handle different environments and tasks.

Based on this foundation, these capabilities must be organically and efficiently integrated to truly achieve the embodied intelligence humans hope for.

However, so far, because embodied intelligence commercialization has not been prominent, the future may face cyclical bubble burst valleys. Zhu Xiaohu believes embodied intelligence valuations have decoupled from commercialization potential, namely "high market consensus but unclear commercialization."

Many facts have proven that after innovation emerges, it's often a rich ecosystem, with the inevitable result of large companies entering, making competition more intense. Like food delivery and AI glasses sectors, when large companies have all entered, it indicates this uncertain field has become very clear.

"Artificial intelligence has deeply penetrated social life with widespread voice interaction, while embodied intelligence goes further," said Li Zhijun, Professor at the School of Computer Science and Technology, Harbin Institute of Technology. Embodied intelligence represents a future frontier field full of challenges and opportunities; it combines digital intelligence with the physical world, achieving environmental perception and physical control through IoT, robotics, and other technologies. "This 'digital-physical fusion' characteristic makes it an extremely valuable track, and practitioners can actively engage in this field to seize opportunities for creating tremendous value."