Are Robots Also Going to Need “Brains”? Songyan Power Enters the Realm of Embodied Intelligence: A Battle Between “Brain” and “Body”
Hello everyone, I’m your financial journalist. Today, we’re talking about a very important, yet perhaps somewhat abstract, development in the robotics industry: robots are starting to develop “brains.”
Recently, a company called Songyan Power made a significant move. You might be familiar with this company, which became well-known for its humanoid robots participating in marathons, performing on the Spring Festival Gala, and starring in sketches with comedian Cai Ming. But this time, it’s not just showing off the “body” of the robots; it has also announced the release of its own “brain” – a general-purpose embodied intelligence technology brand called Scalabot, along with a series of related models.
What does this signify? It marks the transition of the robotics industry from a focus on “muscle” (hardware manufacturing) to a focus on “intelligence” (software algorithms).
Let me break down this news into five key points in plain language to help you understand what’s behind it and what it means for our future lives.
---
1. From “Showing off Muscles” to “Developing Brains”: Why Now?
Many people might wonder, “Weren’t Songyan Power quite popular before? Why are they just starting to work on developing ‘brains’ now? Isn’t it too late?”
In fact, now is the perfect time to enter this field.
- Previous Weaknesses: Songyan Power was strong in terms of the “robot body.” Its founder, Jiang Zheyuan, mentioned that when the company was founded in 2023, the robots’ motion control was not very advanced, and they couldn’t even be produced in large quantities. However, after several years of improvement, they can now manufacture robots at low cost and with high reliability. It’s like having a skilled martial artist who can fight well but doesn’t know how to strategize.
- The Industry’s “Convergence Period: Last year, the team was still figuring out the right direction for the company’s development. This year, the industry’s path has become clearer, focusing mainly on two areas: VLA (Vision-Language-Action models) and World models.
- Conclusion: Previously, the company lacked both a strong hardware foundation and a clear strategic vision. Now that the hardware is in place and the industry’s direction is clear, adding an intelligent “brain” is essential to turn their hardware advantages into real competitiveness. Otherwise, they could easily be outperformed by competitors that combine both hardware and software.
2. Hardware is No Longer Scarce; the Focus Shifts to “Intelligence”
This news highlights an important trend in the industry: the barriers to hardware production are decreasing, while the barriers to software development are increasing.
- Hardware Competition: There are many companies in China that can produce robot hardware. It’s similar to the smartphone market, where screens, batteries, and chips are readily available. Jiang Zheyuan acknowledges that the barriers to hardware production have become less significant.
- Software Determines Value: In the past, robots were judged by how fast they moved or how high they could jump. Now, the key is how “intelligent” they are. Only robots that can understand human commands and work in complex environments will have a real competitive advantage.
- Songyan Power’s Strategy: The company realized that relying solely on selling hardware to research institutions (B2B) and children’s toys (B2C, like “Xiaobumi”) limits their potential for substantial profits. To make big money, they need to create robots with general intelligence that can be used in various scenarios. Therefore, investing in a team to develop embodied intelligence models is a necessary step.
3. The Biggest Challenge: Why Are Robots Still “Stupid”? – Lack of Data
You might think that current AI chatbots are quite intelligent, so why are robots still not much better?
The main reason is lack of data.
- What is “generalization”? Simply put, it’s about whether a robot can perform tasks in unfamiliar environments. For example, if a robot learns to fold clothes at home, but can it do the same at a friend’s house with different furniture, lighting, and clothes? If not, it lacks generalization.
- Where Does Data Come From? Chatbots are trained using massive amounts of text from the internet. However, training robots requires real-world physical interaction data—how they grasp objects, step on surfaces, and handle different materials. This data can only be obtained by having robots work in real-life situations.
- Songyan Power’s Solution: They plan to have robots enter households to collect data. This isn’t for immediate profit but to feed their models. Robots work -> generate data -> the models become more intelligent -> the robots can perform more complex tasks -> more data is produced. This creates a data loop that drives continuous improvement.
4. The Home Environment as a “Gold Mine”: Robots May Enter Your Home by 2029
Why did Songyan Power choose the home environment as the first large-scale data collection site?
- The Most Complex and Valuable Environment: The home is chaotic and unpredictable, with children, pets, and various objects. If robots can survive in such an environment, it will be easier for them to work in factories and shopping malls. The data collected from homes is crucial for improving robot intelligence.
- Large Market Potential: Jiang Zheyuan predicts that by 2029, robots might become widely available in homes, ready to help with household chores. This market is much larger than the current research and education sectors.
- How Will They Do It? They won’t start by selling fully capable robots. Initially, they will deploy hundreds or even thousands of robots in households to collect data and test models. These robots may not be very intelligent at first, but they will learn from real-life experiences.
- Privacy and Security Concerns: This brings challenges regarding privacy and safety. Users may be hesitant to allow “stupid” robots in their homes. Songyan Power needs to address these trust issues.
5. Future Prospects: From “Toys” to “Labor Force”
Let’s look at Songyan Power’s current and future business strategies.
- Current Revenue Sources: The company mainly earns money by selling robots to universities and research institutions (B2B) and children’s robots (B2C, like “Xiaobumi”).
- Future Revenue Drivers: The biggest opportunity lies in home services, which have a trillion-dollar market potential. Once robots can reliably perform household chores, this will be a huge market.
- Commercial Services: Hotels, restaurants, and shopping malls are also potential areas. However, industrial and commercial applications require high precision and stability, so the company plans to start with more forgiving home environments before expanding into these fields.
In summary:
Songyan Power’s release of embodied intelligence models is not just a technical upgrade; it’s a strategic transformation from a hardware manufacturer to an intelligent services provider that integrates both hardware and software.
- For the Industry: This marks the beginning of a new phase of competition, where algorithms and data will be the decisive factors.
- For Consumers: In the short term, you might still see robots that are not very intelligent in households. But in the long run, by around 2029, we might have robots that can help with household chores.
In one sentence: The robots have developed strong “bodies,” and now their “brains” are also starting to function. The next challenge is to see which robots can survive and learn the most effectively in real-world situations.