Summary of Key Points
The industry of embodied intelligence (represented by humanoid robots) is entering a phase of "real implementation": giants from both domestic and international markets (such as Tesla, Xiaomi, Huawei, Tencent, etc.) are making strategic moves, but their approaches differ. Some are directly involved in building the robots, while others are creating technical platforms. Mass production faces two major challenges: a lack of an industrial chain and insufficient data processing capabilities. 2026 is considered the "year of delivery," marking a shift from conceptual presentations (PPTs) to practical application in real-world scenarios. Capital is flowing towards leading companies, making it more difficult for smaller players to secure funding. The future competition will focus on execution and the ability to create a sustainable business model.
1. Giants' Strategic Positions: Building Robots or Creating Platforms
In the realm of embodied intelligence, giants have chosen two different paths:
- Directly building robots: Tesla and Xiaomi are examples. It took three years for Tesla's Optimus to move from laboratory development to production lines, but Elon Musk admitted that mass production is the most challenging aspect of the company's history due to the need to develop all new components from scratch (such as joints and dexterous hands), similar to building a completely new automobile manufacturing line from zero. Xiaomi has placed robots in their factories for training; after four months, the success rate of assembling nuts was close to that of skilled workers, although the actual operation speed is still slower (the video was played at triple speed), indicating that further refinement is needed.
- Creating platforms: Huawei and Tencent are taking a more supportive role. Huawei Cloud has launched the CloudRobo platform in collaboration with Yijiahe, a specialized robotics company. Yijiahe provides real data from power and transportation scenarios, which Huawei uses to train robots using large models, essentially laying the foundation for all robot manufacturers. Tencent has upgraded its "full-stack embodied intelligence solution," offering end-to-end support from cloud infrastructure to application development to help companies quickly set up robotic systems.
These two approaches reflect differences in resources: those with manufacturing and supply chain capabilities (Tesla, Xiaomi) build the robots themselves, while those with technical and platform advantages (Huawei, Tencent) focus on building ecosystems.
2. Two Major Barriers to Mass Production: Lack of an Industrial Chain and Insufficient Data
To move from the laboratory to the factory, two hurdles must be overcome:
- Lack of an industrial chain: Tesla's experience is a prime example. They had to develop all core components like joints, sensors, and dexterous hands themselves, as there were no existing suppliers. This is akin to trying to make a popular milk tea drink without the necessary cups and straws, requiring the establishment of a new production line with high costs and long timelines.
- Insufficient data and processing capabilities: Robots need to learn from real-world experiences to function effectively. Xiaomi's robots in factories are collecting data on tasks like screwing nuts and handling magnetic interference, but they still rely on manual assistance (such as moving boxes to align with clips), indicating that the range of covered scenarios is limited. Moreover, collecting data in the physical world is extremely costly (e.g., complex factory environments). Companies that can generate data efficiently (using simulations) and process it quickly (with large models) will have a competitive advantage.
3. Capital's Preference for Leaders: Smaller Players Face Financing Difficulties
The enthusiasm for embodied intelligence is evident in funding trends: domestic financing in the first half of 2026 increased by five times, reaching 93.5 billion yuan, with eight companies valued over 20 billion yuan. However, capital is not distributed evenly; it favors leading firms (Tesla, Xiaomi, Huawei). Smaller startups are finding it increasingly difficult to secure funding.
Why? Investors no longer focus on how impressive a PPT presentation is but on whether a product can be mass-produced and has practical value. For example, Xiaomi's robots' ability to work in factories and Tesla's automotive manufacturing infrastructure demonstrate tangible capabilities that attract investment. Smaller companies without core technologies or relevant use cases struggle to obtain funding.
4. 2026: The Year of Delivery - A Transition from Dreams to Reality
2026 is dubbed the "year of embodied intelligence delivery," meaning that companies must now present products that can be delivered and continuously used in real-world scenarios.
- Overcoming practical challenges: Robots must not only perform tasks but also be cost-effective. If they are more expensive than hiring workers, they won't be adopted by businesses.
- Adapting to real-world conditions: Robots need to handle dirty, messy, and irregular environments (e.g., factories and homes).
- Establishing standards: Governments are starting to set industry norms, with leading companies contributing to the development of safety and efficiency standards. The industry will move from unregulated growth to a more regulated environment.
5. An Industry Turning Point: From Technological Showoff to Value Creation
The focus in embodied intelligence is shifting from who has the most advanced technology to who can create real value. Previously, companies competed on the complexity of robot capabilities; now, the focus is on how robots can reduce costs and improve efficiency for businesses. For instance, Xiaomi's robots reducing errors by replacing manual tasks represent tangible value. If robots are only visually appealing but not practical, even the most advanced technologies will be ineffective.
In the next two to three years, the industry landscape will solidify: cross-industry giants (like Tesla and BYD) will leverage their manufacturing strengths, while traditional firms (Huawei and Tencent) will strengthen their positions with platforms and data. Smaller players will either find niche markets (e.g., healthcare, elderly care) or be marginalized.
In conclusion: Embodied intelligence is no longer just a science fiction concept; it's becoming a practical tool for factories and homes. Those who can solve the challenges of mass production and implementation will gain access to the next wave of productivity.