Summary of Key Points
On the eve of the 2026 World Artificial Intelligence Conference, the focus of competition in the field of embodied intelligence (robots that can perceive, act, and think like humans) has shifted from building the physical robots themselves to developing the underlying infrastructure (data, computing power, development platforms, etc.). Huawei Cloud has chosen not to build robots directly but instead uses its CloudRobo platform for embodied intelligence development. Following a partnership model similar to that of SAIC Motors in the automotive industry—where multiple companies collaborate on research and development, mass production, and promotion—the company aims to address current industry challenges such as insufficient data, a disconnect between simulation and reality, and fragmented development processes. Huawei Cloud is prioritizing high-value sectors like electricity (which involve hazardous tasks and face difficulties in recruiting workers) and is committed to building an open ecosystem, emphasizing that the industry is still in its early stages and requires sustained investment.
1. Shift in Competition: From “Who Has the Coolest Robot” to “Who Controls the Infrastructure?”
Last year, companies were competing to see whose robots were the most flexible or whose large models were the most intelligent. This year, the focus has shifted to the underlying technologies that power these robots. Why? Because having a robot alone is not enough; what really matters is the infrastructure that enables it to function effectively in real-world scenarios—high-quality data (similar to the extensive practice materials needed for human learning), sufficient computing power, platforms for quickly deploying models onto robots, and a system for continuous improvement. Without these essential components, robots can only be impressive in the lab but cannot be used in factories or power grids.
2. Industry Challenges: Insufficient Data, Inaccurate Simulation, and Dispersed Development
Despite the growing interest in embodied intelligence, there are several major barriers to its widespread commercialization:
- Lack of Data: There is a shortage of high-quality, multi-modal data (e.g., comprehensive data on how robots see, hear, and touch things), similar to students lacking textbooks.
- Disconnection between Simulation and Reality: The simulation environments in labs do not reflect the real world accurately, causing robots to struggle when applied in practical situations.
- Fragmented Development Processes: Steps such as model training, data management, and motion control are often carried out separately, leading to inefficiencies.
- Mismatch in Industry Roles: Robot manufacturers understand mechanics but lack algorithms, while AI companies specialize in intelligence but do not know how to build robots; no single entity can achieve commercial success on its own.
3. Huawei Cloud’s Approach: Building the “Android System for Robots”
Huawei Cloud is not involved in building robots themselves but aims to become a platform provider:
- Developing a Platform: It has launched the CloudRobo platform, which integrates various capabilities such as data management, model training, and cloud-edge collaboration (allowing seamless data exchange between robots and the cloud), thus lowering the barriers for developers and speeding up the development process.
- Partnership Model: Huawei Cloud collaborates with companies like Yijiahe (specializing in robotics for the electricity and healthcare industries). Yijiahe provides industry expertise, use cases, and products, while Huawei Cloud contributes its platform, computing power, and model technology. Together, they develop prototypes, verify their effectiveness, and promote mass production—similar to how Huawei supports SAIC Motors in manufacturing cars, providing comprehensive support rather than just selling components.
4. Why the Electricity Industry as the First Pilot?
Huawei Cloud has chosen the electricity industry for its pilot project for several reasons:
- High Value: The electricity sector involves many tasks that are difficult or undesirable for humans to perform (e.g., working with live power lines, conducting long-term inspections). These applications generate immediate economic value and make it easier to establish a viable business model.
- High Degree of Standardization: Inspection tasks in the electricity industry are relatively consistent (e.g., checking utility poles and equipment), making them suitable for testing technologies and business models before scaling to other sectors like logistics and healthcare.
5. The Industry Is Still in Its Early Stages: Patience Required for Ecosystem Development
Huawei Cloud believes that embodied intelligence is still in its infancy:
- Open Ecosystem: The CloudRobo platform is open, allowing partners to contribute their data, models, and tools to build a robust ecosystem together.
- Long-Term Vision: Expecting breakthrough products after just one month of collaboration is unrealistic; it requires a long-term investment to establish a foundation that enables more companies to utilize this technology effectively and drive industry growth.
In summary, the embodied intelligence sector is moving from conceptual demonstrations to practical applications. Huawei Cloud’s strategy is to focus on building the infrastructure rather than competing for attention by creating robots. By collaborating openly with industry partners, it aims to solve real-world problems and gradually expand the technology’s impact, much like building a highway that allows more “robot cars” to operate efficiently.