In 2004, an Embodied AI CEO Goes Full-Stack Open Source; Xiaomi and Galbot Bet Tens of Millions of Dollars
The Appeal of an All-Gen Z Founding Team
"The Appeal of an All-Gen Z Founding Team" By Zhiyan Chen
AnYong Waves has learned exclusively that RoboParty, a full-stack open-source bipedal humanoid robot company founded this April, has completed its seed round. The round was jointly led by Matrix Partners China, Xiaomi Strategic Investment, GalaxyBot, and L2F Light Source Entrepreneurs Fund. To date, RoboParty has raised tens of millions of dollars.
RoboParty founder Yi Huang was born in 2004, making him one of the youngest humanoid robot entrepreneurs on the market.
At the end of 2023, while still a freshman at Harbin Institute of Technology, Huang and several roommates built a walking bipedal robot called AlexBot in their dormitory in just a few dozen days, with only 15,000 RMB in funding. They documented the entire R&D process and shared all materials openly on Zhihu, GitHub, and Feishu. In February 2025, with sponsorship from Fourier Intelligence, they completed an upgraded version, AlexBot Mini. Shortly after, Huang graduated from college one year early and founded RoboParty, focusing on a single direction: full-stack open-source bipedal humanoid robots.
Fitting the archetype of the "boy genius" founder was merely Huang's ticket into the orbit of top-tier VCs. But it is the "full-stack open-source" strategy in the humanoid robot space that may be the core reason RoboParty has attracted capital in 2025, even as the embodied intelligence "battlefield" grows increasingly crowded.
Huang has written specifically to explain his rationale for open-sourcing. He believes that "if closed systems are a symbol of power, then open source is a faith in civilization." In his view, the advantages brought by "open-source development"—reducing collaboration friction, enabling deep cooperation, increasing technical transparency—far outweigh the potential downside of being copied by competitors. "There are also benefits like greater safety, reduced trade friction, and improved robot accessibility."
RoboParty is currently the only domestic manufacturer building full-stack open-source bipedal humanoid robots, and is now raising its angel round. Just months after its founding, industry giants and star financial institutions have already begun assembling behind it. Is this the demonstration effect of DeepSeek's global explosion earlier this year? Or is it another inflated capital bubble, swollen by the spreading FOMO in embodied intelligence?
With curiosity and skepticism, on a stormy late-summer afternoon in Shanghai, AnYong Waves met Huang at RoboParty's office in Zhangjiang Robot Valley.
Still groggy from sleeping in the office the night before, he began introducing ATOM, the team's latest development and "China's first fully open-source humanoid robot," as soon as we walked in. Several desks held disassembled humanoid robots in various states. Some workstations sat empty, as if waiting for more young people to join the adventure.
Huang shared with us his experiences founding the company and raising funds, his understanding of open source, and his views on the current state and future of the humanoid robot industry. Somewhat unexpectedly, we found in this 21-year-old founder a grasp of the business world that transcended his age, and differed from the typical geek profile—along with reflections on technology, society, and endgames.
The following conversation has been edited by AnYong Waves—
Part 1
The Only Answer to the Ultimate Order
AnYong: RoboParty's most distinctive label is "fully open source." How did this idea first come about?
Huang: It started from user demand. I had this idea when I built the first-generation robot at school. I open-sourced AlexBot, and it attracted many users who wanted to do secondary development. Two years later, people are still contacting me wanting to buy it.
This made me realize that open source has a powerful vitality and a social lever for technology democratization.
For humanoid robots to achieve ultimate "usability," relying on a few companies working behind closed doors is very difficult. It requires bringing together developers across the entire industry, allowing robots to learn and evolve like humans in various scenarios. Open source is the fastest path to that ultimate order.
Like the Android system—it provided an underlying platform, countless developers built apps on top of it, and that's how we got today's thriving mobile ecosystem.
AnYong: So you're not just trying to sell robots, you want to become a standard-setter for the industry?
Huang: I think there are two ways to set standards. One is through extremely high market share, the other is through open source. I do hope that through each generation of open-source product iteration, we can define the industry's norms. For example, how many degrees of freedom should a humanoid robot have? What kind of motors should core joints use? When the entire industry starts using your standards, your platform, you can build a powerful ecosystem. I call it the "Radish Chain."
At that point, we're no longer just a hardware company, but a platform company.
AnYong: Could this just be wishful thinking on your part?
Huang: This is the general trend. Even if it's not RoboParty, there will definitely be some other "platform." I'm also continuously developing my ability to quantitatively assess enterprise development, to see clearly the relationship between the world five or ten years from now and RoboParty.
AnYong: Is the possibility of becoming a platform the reason you've attracted so much capital?
Huang: Possibly. But at our very early stage, I feel that investors are more betting on the potential of our team. I follow the stock market myself, and my observation is that "the secondary market looks at the thing, the primary market looks at the people."
AnYong: It's usually assumed that open source means giving up technical moats. Isn't that commercially risky?
Huang: This is precisely our moat. For a startup, opening up all technology seems risky, but it can exchange for the power of the entire developer community—a power that no closed company can match.
Moreover, from the perspective of overseas expansion potential, a completely open-source, transparent platform is more easily accepted by international markets. It can avoid many potential regulatory and geopolitical risks, and ensure future ethics and safety.
AnYong: Why did you take money from industrial players like Xiaomi and GalaxyBot at such an early stage?
Huang: The quality of money is different. For me, investors choosing us and us choosing investors is not entirely a rational, quantifiable decision. It also involves a lot of irrational judgment, or what I'd call a matter of taste.
I chose strategic investors because I believe they can bring value beyond just money. Xiaomi, for example, has extremely strong supply chain capabilities and hardware manufacturing experience. We are currently the only robot company that Xiaomi Strategic Investment has invested in.
Industrial capital brings me strategically valuable perspectives, which matter far more than a slightly higher valuation in the short term.
AnYong: Then why did I hear that you also turned down an order for hundreds of units from a listed company?
Huang: It sounded very tempting. Taking it, the company's valuation might immediately multiply several times. But the other party attached some conditions, such as exclusive distribution rights for overseas markets. After evaluation, I felt that at a stage where our production capacity, technology, and brand haven't been fully polished, prematurely exhausting ourselves for a good PR story could disrupt our rhythm. Moreover, such an order wouldn't bring real developers. It's透支企业生存的—draining the company's lifeblood. A hundred units now doesn't mean much; a hundred high-quality developers means much more.
Part 2
Entrepreneurship Is the Most Efficient Path to Self-Realization
AnYong: From college to starting a company, you and your team have developed products quite quickly. How do you do it?
Huang: It's actually ignorance is bliss, and also one of the few advantages startups have. Large enterprise armies have strong combat power but slow speed. Every task gets planned, assigned to someone, then pieced together, then goes through procurement and processing workflows—all these fragments gradually wear down people's morale. But guerrilla units can move fast with small steps, quickly adjusting direction.
Another point is that we've done complete robots full-stack ourselves. Now we're essentially continuously productizing and mass-producing to refine our product, so it might be faster.
AnYong: Many institutions prefer hardware founders with big-company experience when investing. You started your company directly after graduating early as a junior. Looking at it now, was this the right decision?
Huang: It's actually impossible to tell right now whether it was right or wrong. It's really a matter of choice, though I genuinely never considered working for someone else. For me, any decision a person makes at their current stage is to chase their own dreams. I'm in my third year now, graduating one year early—I'd say it's a golden age for entrepreneurship.
The choice to start a company also relates to my family environment. I'm from Wenzhou, grew up in Shanghai, and my parents also run their own business. This exposed me to business very early, and gave me my own views on "surplus value."
If you work in a company, the labor you put in cannot get 100% return—some portion is always taken. Entrepreneurship is like obtaining resources from society to 100% realize myself, our team's ideas. This manifestation of self-value is, for me personally, the most efficient.
AnYong: So experience isn't that important in the humanoid robot industry?
Huang: I don't know. But I've also seen some investors' shares—for example, David Zhang from Matrix Partners China. From watching his interviews, I feel experience and entrepreneurial success don't necessarily correlate.
This industry is too new. Many of the most cutting-edge advances come from universities, even from undergraduates. But for traditional hardware, supply chain veterans and big shots are still very much needed. When it comes to innovation, it's some inexperienced but fast-learning young people who can create new things. An exponentially rising curve always has an advantage over a linear straight line.
So when I hire myself, I also tend to prefer "young high-potential" people.
AnYong: Objectively speaking, are there still opportunities for brand-new teams in the humanoid robot industry today?
Huang: Right now in this industry, I feel everyone is still more or less at the same starting line. On the software side, the gap between leaders and followers is maybe six months at most. The hardware gap might be larger—one to two years—because it involves supply chain management, mass production, and testing.
Jobs also said: people who are really serious about software should make their own hardware. Apple is exactly like this.
When technical gaps don't constitute major moats, what truly determines victory is team iteration speed and talent density. You need a high-density talent team that can keep winning battles. If your team and iteration are stronger than others, you have the opportunity to catch up to the industry frontier. In traditional industries, hardware emphasized processes, software emphasized talent density. In the current environment, automating many processes while maintaining high talent density is the optimal solution.
I believe whoever can make humanoid robots "useful" fastest wins the first step of this competition. There hasn't yet emerged a product like the Macintosh in the PC era, one that could define the industry and lower the barrier for users. Everyone is still exploring, and there are many technical paths.
This is precisely the biggest window of opportunity for startups.
AnYong: What stage is RoboParty's R&D currently at?
Huang: The company is mainly working on its second-generation robot now. This generation plans to explore some new processes and materials. We've redefined the product and conducted fairly detailed interviews and research with target users.
For a consumer-facing and small-business product, there actually exists an impossible triangle: Quality, Price, Threshold. Q represents product quality—for humanoids, MTBF (Mean Time Between Failure). Industry leaders currently achieve several hundred hours, and we're gradually closing in. P is price. When you design the robot, you can calculate the approximate BOM price. It depends on which market segment you want and how you position the product, then achieving the lowest possible product price, plus profit margin, gives you the pricing. T is user threshold—how long it takes for a beginner (in robots, this means an entry-level developer) to get started with debugging, and whether resources and tutorials are comprehensive. Early Apple computers achieved that era's impossible triangle.
AnYong: Many people talk about "digital immortality"—using AI + robots to live forever. Is this what you ultimately want?
Huang: The company's vision is "to let the whole world enjoy the services our robots bring."
I definitely don't want digital immortality. What I care more about is that my ideas, my products, the models I create, can continue to exist and influence the world after I leave it.
Like Jobs—he's no longer here, but his design-driven engineering, his extreme pursuit of product, still influences the tech industry today. Like Musk going to Mars—not to escape Earth, but to ensure humanity's continued existence.
AnYong: What endgame do you see for humanoid robots?
Huang: The endgame isn't a product, but a paradigm. What I hope RoboParty brings is not just robots, but possibly a set of standards, a new company management paradigm, even a human-machine symbiosis paradigm—building dreams together with the most brilliant minds among humanity, the dreams of all humankind.
Image source | Exploded view of RoboParty Atom01 (internal codename) components, provided by the interviewee.
Recommended Reading
The flow of money, the rise and fall of people