Summary in Plain Language
A startup called Feimo Yi, which has only been around for half a year and has not even revealed what its first product looks like, has just secured nearly $100 million (about 700 million RMB) in seed funding from leading industry players such as IDG, Didi, and Hesai Technology. The founder comes from a background in advanced autonomous driving, and the team plans to develop a “physical intelligent entity” that can understand vague human commands and complete tasks on its own in the real world. Currently, all the capital in the AI industry is flowing into this area, with investments happening at a much earlier stage—right after the company is registered. However, it’s still unclear whether Feimo Yi will be able to deliver a functional product.
---
Detailed Explanation
1. How Unusual is This Round of Funding?
It’s almost like getting a startup license without having produced any product yet and receiving a startup capital of 700 million RMB. What is the typical seed round like in the startup world? For a regular business like a bubble tea shop, the startup capital is only in the hundreds of thousands of RMB. Even in the tech sector, securing several million RMB in the seed round is considered quite good, and for traditional hardware startups, the seed round usually doesn’t exceed 50 million RMB. Feimo Yi has shattered this record, raising nearly 10 times the average funding amount for early AI projects in China in the first half of 2026 (73 million RMB). What’s even more surprising is that the company’s registered capital is only 1 million RMB, meaning that the capital was injected just after the registration process. The reason investors are so willing to invest is twofold: on one hand, the founding team has experience in the entire process from technology development to mass production, which has proven their capability; on the other hand, the investors include not only financial institutions but also industry players like Didi (with its transportation resources) and Hesai (a supplier of core robot sensors), who can provide tangible resources for the project’s implementation.
2. A Team from Autonomous Driving Moving to Robotics: It’s Not a Cross-Country Transition, but a Technical Shortcut
Many people might wonder why a company that used to make autonomous vehicles would suddenly switch to robotics. The underlying technologies are essentially the same. An autonomous vehicle is a large robot that needs to navigate the road, understand its surroundings, predict the behavior of pedestrians and other vehicles, plan its route, and respond to unexpected situations without human instruction. This technical expertise can significantly speed up the development process for a robot that can move and perform tasks. However, the challenges for robots are much greater than those for autonomous vehicles, as they need to handle various environments and user commands.
3. Feimo Yi’s Goal: Creating a Robotic Entity That Can Complete Tasks
The robots we see on the market today are typically limited to performing a single action (e.g., a robot that screws screws in a factory). Even the popular humanoid robots that can dance, do stunts, or deliver drinks have pre-programmed routines and struggle to adapt to new situations. Feimo Yi aims to create a robot that can understand natural language commands and complete multiple tasks independently. They have divided their capabilities into three levels: the first level is to perform individual actions (e.g., picking up a cup), the second is to perform a series of actions (e.g., moving a cup from a coffee table to a dining table), and the third is to achieve the ultimate goal (e.g., making tea for you). Such a robot would need to be able to adjust its plan dynamically, even if the tea pot is in a different location or the cup has been moved elsewhere.
4. Capital’s Rush into This Field
The reason for the capital rush is that AI has reached a critical point where it needs to move from the virtual world to the real world. In the past, AI applications like ChatGPT, image generation, and video editing remained limited to servers and screens, providing text, images, or videos. However, the current market is highly competitive, with large models requiring more and more funding but offering fewer new applications. “Physical intelligent entities” represent the next massive market opportunity. If AI can truly help with real-world tasks, the market size could be several times larger than all current internet businesses combined. Investors are willing to invest early, hoping to secure a position in future leading companies. Feimo Yi is not the only example; four early AI startups in the first half of 2026 already received over 70 million RMB in seed funding.
5. The 70-Million RMB Funding is Just the Starting Point
The 70-million RMB funding is just an entry ticket. The real challenge lies ahead. Feimo Yi has chosen a “full-stack self-development” approach, meaning they will develop everything from the AI software to the mechanical hardware, the training data, and the systems for interacting with users and the environment. While this approach ensures better integration, it also means higher costs and greater engineering complexity. Additionally, there’s a significant risk: the success rate of complex tasks decreases significantly as the number of steps increases. Even if each small action is highly reliable, the overall success rate for a series of tasks can be low. This is why robot demonstrations often look smooth but often fail in real-world use.
6. The Future of Robotics
The ability of robots to understand natural language commands and complete tasks independently represents a major breakthrough. Currently, no mature product on the market can do this. The capital rush reflects the realization that AI needs to become more practical and integrated into our daily lives.