The "Didi Chuxing" of the Robotics Industry? Unraveling the $500 Million Valuation Myth of Mecka AI
Hello everyone, I'm your financial journalist. Today, we're talking about a story that sounds a bit magical: a company founded just two years ago by people with backgrounds in food and beverage payment systems and cryptocurrency has managed to secure a $500 million valuation led by Sequoia Capital, all by having people wear sensors while performing everyday tasks.
It's like in the internet era, when everyone was still debating whose phone had the bigger screen, someone suddenly came up with a "phone repair crowdsourcing platform" that was valued even higher than the phone manufacturers themselves.
So, what exactly is going on behind this? Why has capital suddenly become so eager to invest in companies that "feed data to robots"? Let's break it down in plain language.
Summary of the Core Points: Who is making the money? And why?
In one sentence:
Mecka AI is a company that provides "teaching materials" for robots. It doesn't build robots or write code; instead, it pays people to wear sensors and record their actions while doing household chores, repairing cars, or making coffee. It then packages these videos and motion data and sells it to robot companies, teaching them how to interact with the physical world like humans.
Key Highlights:
1. Rapid Valuation Growth: The company has doubled its valuation to $500 million in just three months, with Sequoia Capital joining the investment.
2. Diverse Team Background: The founders come from various fields (former food and beverage finance, cryptocurrency industry executives) and have no prior experience with robotics, yet they identified the biggest pain point in the industry—lack of data.
3. Business Model: By using a crowdsourcing approach (similar to Didi Chuxing or Meituan's delivery services), Mecka reduces the cost of data collection significantly.
4. Industry Trend: AI investment is shifting from the "textual world" (large language models) to the "physical world" (embodied intelligence/robots), and data is the fuel for AI in this new realm.
---
In-Depth Analysis: Understanding This "Data Revolution" from Five Dimensions
1. Pain Point Insight: The "Illiteracy" of Robots
Why do we need Mecka? Because robots are currently "highly ambitious but illiterate."
In the past few years, the biggest success in AI has been with large language models (LLMs) like ChatGPT, which learned to write poetry, code, and chat by reading vast amounts of text data from the internet. However, robots are different. They need to understand the physical world:
- How much force should be used to pick up an egg without breaking it?
- What is the direction and speed of the water flow when a coffee cup is spilled?
- How much friction is there when tightening a screw?
This kind of knowledge doesn't exist in text data, and even videos don't capture it fully. Robots need "action data" from a first-person perspective that includes touch, force, and spatial relationships.
The current situation is that most robot companies use remote operation in laboratories, where engineers control the robots from a distance. This is like asking a college student to copy books by hand in a library—it's inefficient, costly, and limited to a very specific environment (the laboratory).
Mecka's founders realized that the problem lies not with the robots' hardware or algorithms but with the lack of suitable "teaching materials." They realized that the best teachers for robots are ordinary people.
2. Business Model: Turning Data Collection into a Crowdsourcing Exercise
How does Mecka do this? It turns what used to be engineering tasks into everyday activities.
Mecka's core business is very practical:
1. Recruit Ordinary People: They find people like housewives, car repairmen, and baristas.
2. Wear Devices: These people wear simple sensors and smartphones.
3. Record Daily Activities: They perform their usual tasks—making coffee, repairing cars, folding clothes.
4. Data Packaging: The company collects these videos, motion trajectories, and sensor data, cleans and annotates it, and sells it to robot companies as training materials.
What makes this approach effective?
- Low Cost: Remote operation in laboratories requires expensive robots, professional engineers, and specialized facilities. Mecka only needs smartphones and a few sensors, allowing ordinary people to contribute in their spare time.
- Scalability: Laboratories can collect a few hundred pieces of data a day, while Mecka can mobilize tens of thousands of people simultaneously, exponentially increasing the amount of data.
- Real-World Relevance: Laboratory data is often standardized and perfect, but the real world is messy (coffee might spill, screws might slip). Crowdsourced data includes these imperfections, making it more realistic for real-world applications.
3. Team Background: Why "Outsiders" Can Succeed?
Their lack of robotics experience is actually an advantage.
Mecka's founders—Josh Gao and Mogen Cheng from the food and beverage finance sector, Jason Chong from the cryptocurrency industry, and Duy Nguyen from operations—didn't have any background in robotics. This outsider perspective helped them avoid technical biases:
- Technical Thinking: "We need more advanced sensors or more complex algorithms for data collection."
- Business Thinking: "We need to obtain data the cheapest, fastest, and on the largest scale."
Instead of focusing on building better robots, they focused on the infrastructure of data collection. They saw that the best data source is not engineers but ordinary people.
4. Competitive Landscape: The Race Is Intense
How competitive is this field? More so than you might think.
- International Competitors: UC Berkeley's XDOF, for example, focuses on remote operation and is valued at $1.2 billion, 2.5 times more than Mecka. This shows capital's optimism, but XDOF's approach is more specialized.
- Domestic Players: Companies like Luming Robotics and Mifeng Technology are also working on wearable data collection; Pasini has a foothold in tactile data.
Mecka's Differentiation:
- Cost Advantage: The crowdsourcing model is naturally cheaper than laboratory-based methods. If Mecka can provide data of 80% quality at 20% of the cost, it has significant pricing power.
- Platformization: Mecka is not just selling data; it's also building a data infrastructure. Its Egoverse dataset serves both as a technical showcase and a customer acquisition tool. Developers can use this dataset to test their algorithms, making them dependent on Mecka's platform.
Risks:
The biggest challenge with crowdsourced data is noise—the data from non-professional annotators may be inconsistent. If Mecka can't clean and standardize the data effectively, its cost advantage could turn into a disadvantage.
5. Capital Logic: From "Selling Water" to "Selling Shovels"
Why did Sequoia Capital invest? Because this is the "oil" of the second half of the AI revolution.
Looking back at the internet:
- PC Era: Those who controlled the operating system (Windows) won.
- Mobile Internet Era: Those who controlled app stores or traffic channels won.
- AI Model Era: Those who controlled computing power (NVIDIA) and data (OpenAI's datasets) won.
Now, AI is moving from the virtual world to the physical world. The benefits of large language models have peaked, and the next big opportunity lies in embodied intelligence (robots). For robots to become practical, what's most lacking is not hardware (which can be mass-produced) but high-quality, large-scale, and diverse physical interaction data.
Mecka plays the role of the "data oil company" in the robotics era. It doesn't manufacture robots but provides the essential "fuel" for their operation.
Sequoia Capital's investment strategy in this area is clear:
- Investing in the Data Layer: Mecka AI (data collection), XDOF (remote operation data).
- Investing in the Complete Robot Layer: Companies like Wujie Dongli, Zi Bi Xing Robo, and Iti Zhihang.
This is a "full-chain" investment strategy. Capital believes that a large portion of future profits in the robotics industry will go to companies that control core data and infrastructure. Just as in the early internet, companies that built servers didn't necessarily outperform those that built social networks, but those that built data centers became stable.
In summary:
Mecka AI's story is about using internet principles (crowdsourcing, platformization) to reinvent traditional industrial processes (data collection). The company bets on:
1. The widespread adoption of robots.
2. The fact that data is the biggest bottleneck for the practical implementation of robots.
3. The ability of the crowdsourcing model to solve cost and scalability issues.
If these assumptions hold true, Mecka could become the "Didi" or "Amazon AWS" of the robotics industry. However, if the quality of crowdsourced data doesn't meet expectations or if the adoption of robots slows down, the $500 million valuation might turn out to be a bubble.
Regardless, the trend is clear: the battlefield of AI is shifting from the cloud to the physical world around us. And data is the gold in this new territory.