Summary of Key Points
In 2025, the financing for embodied intelligence and humanoid robotics was booming (over 300 financings, totaling over 38 billion yuan, involving more than 230 companies), but the actual implementation results were dismal: there were very few cases where robots could operate stably on production lines and customers were willing to renew their contracts. The industry has shown clear differentiation—leading companies have entered the early stages of commercialization, while most firms are still at the demo or pilot phase, facing issues such as unstable long-term operations, failure to scale up effectively, and unfeasible return on investment (ROI). HuSiu will host a closed-door meeting on July 8th, inviting industry experts from various fields (technology, end-users, investors, etc.) to focus on the gap between making robots functional and profitable. The meeting will delve into key issues such as technical shortcomings, business models, and capital evaluation.
Detailed Analysis
1. Behind the Surface of the Boom: Most Robots Remain at the “Sample Stage,” with Few Being Profitable
Last year, everyone rushed into the process robotics market, claiming their robots would be used in factories. However, a year later, the reality became clear:
- Demo ≠ Production Line: Many robots work well in laboratories but fail in real factories due to factors like component tolerances, environmental disturbances (dust, temperature changes), and differences in material properties (different processing methods for metals and plastics), leading to unstable performance.
- Pilot ≠ Scaling: What works on a small scale fails when applied across multiple production lines, as factory layouts and process requirements vary significantly.
- Deployment ≠ Profitability: Even if the system is installed, the payback period is too long (e.g., 5 years), which factories are unwilling to accept.
Now, the industry is separating into two groups: leading companies with actual production line experience are exploring commercialization, while most firms are still relying on demos to showcase their capabilities.
2. Technical Challenges: VLA Alone Is Not Enough; Multiple Abilities Are Needed
VLA (Vision-Language-Action models) were highly sought after last year because they can understand human commands and generate actions. However, in factories, this approach is insufficient:
- Factories Care About Practicality: They require precision within 0.1 millimeters and a quick payback period (e.g., 2 years), which VLA models cannot meet.
- The Physical World Is Complex: Robots need to account for errors in components, chaotic environments, and material variations; they must not only “see” and “think” but also understand physical principles and control movements in milliseconds.
- Industry Consensus: The solution is to enhance VLA by integrating it with a “world model” (that understands physical laws) and a “control system” for real-time operations, creating a complete working system suitable for factories.
3. End-User Requirements: Robots Must Meet Strict Standards to Stay in Use
Zhou Yurong from Siemens Digital Manufacturing is a key figure in factory acceptance, having deployed AI-powered systems globally. He highlights the essential criteria for robots to be accepted:
- Stability: The robot must not fail frequently, as it affects production efficiency.
- Performance Metrics: Production rates, quality standards (product qualification), and the need for minimal human intervention must meet factory requirements.
- Feasible ROI: The payback period should be reasonable (2–3 years or less) to ensure customer satisfaction.
Robots that can only demonstrate their capabilities will not be adopted due to failing these criteria.
4. Capital Perspective: Which Robot Companies Will Survive?
Mao Yuhao from Qingliu Capital, who has invested in StarDong纪元, offers insights from a capital standpoint:
- The Bubble Has Burst: The era of financing based on promises is over; now, companies must prove their ability to scale and generate profits across multiple factories.
- Time for Listing Is Not Yet: Most firms have not reached profitability, so listing is still far off.
- Future Paths: Leading companies may go public independently, while smaller ones will either be acquired by larger players or traditional manufacturers or disappear.
- China-US Gap: The U.S. excels in algorithmic research, while China has advantages in supply chain (hardware manufacturing) and application scenarios (many factories), but there is still a gap overall.
5. The Value of the Closed-Door Meeting: Addressing the Profitability Challenge and Providing Direction for the Industry
The meeting will focus on four critical issues:
1. Barriers to Implementation: What prevents robots from moving from demos to production lines? Which metrics indicate a robot’s suitability for practical use?
2. Software-Hardware Approaches: Should software or hardware be developed first, and what pitfalls have been encountered along the way?
3. Technical Roadmaps: Should VLA models be upgraded, and how should they be integrated with other systems?
4. **Listing and Mergers: Who Will Make It to the Market? How big is the gap between China and the U.S. in terms of technology and supply chains?
Participants will gain insights into real-world cases, investment strategies, and industry connections, helping to address the question of how robots can transform from costly investments into profitable tools.
The core goal of this meeting is to turn robots from laboratory curiosities into practical solutions for factories—after all, factory owners care only about whether they can save money and increase profits, not about the sophistication of the technology. If you are a robotics entrepreneur, investor, or industry observer, this closed-door meeting is worth attending.