Summary of Key Points
Huang Minshan, a 90s-born "NVIDIA heiress" (with an annual salary of 8 million yuan and a position in physical AI), made an appearance at the 2026 Beijing World Robot Conference. She visited leading Chinese robotics companies such as Yuejiang and Ubot. Behind this visit is NVIDIA's strategic focus on "physical AI" – the technology that enables robots to understand and act in the real world. Meanwhile, China's robotics industry is experiencing a surge of IPOs (with Yushu Technology just going public, and over 50 companies waiting in line). The competition has shifted from the hardware "body" to the AI "brain," and global tech giants have turned their attention to China as a testing ground for this technology.
1. Huang Minshan's Visit to Beijing: NVIDIA's Close Attention to China's Robotics Ecosystem
Huang Minshan's trip was not for sightseeing; it was a mission to assess the technological progress of Chinese robotics companies, as these firms are key partners in NVIDIA's physical AI ecosystem. She observed Yuejiang's "one brain, multiple bodies" robots playing table tennis, Guanglu Intelligent's physical simulation equipment, and Ubot's humanoid robot Walker C1. She even experienced JD.com's real-world operating robots. Every step of her visit revolved around the implementation of physical AI technologies like NVIDIA's Omniverse (a virtual training platform) and GR00T (a humanoid robot foundation model), which require Chinese companies to apply and provide feedback in real-world scenarios. This indicates that China's robotics industry has become a central battlefield in the global physical AI competition, and NVIDIA does not want to miss out.
2. From Chef to Annual Salary of 8 Million Yuan: The Power of "Unconventional Paths" in the AI Era
Huang Minshan's career path is quite unconventional: she started as a chef, then worked in brand marketing at LVMH, and finally joined NVIDIA as an intern. Her experience challenges the stereotype that one must have a formal scientific background in the tech industry. Her attention to detail (from cooking to luxury product user experience) aligns well with the precision required in developing tech products (such as the need for accurate simulation in physical AI). Starting as an intern, she diligently learned unfamiliar terms and eventually became a senior director, disproving accusations of favoritism. Jensen Huang stated, "I won't recommend someone who would bring shame to me; ability is the true currency."
3. China's Robotics Industry on the Rise: IPO Boom, but Real-World Challenges Remain
At this year's conference, nearly 400 companies showcased over 3,000 robots, ranging from dance robots to practical models capable of handling materials and sorting goods (used in logistics and automotive manufacturing). Yushu Technology's stock price soared by over 600% on its first day of trading, and more than 50 companies, including Zhiyuan and Yunshen Chu, are waiting to go public, indicating strong capital interest. However, going public is just the beginning. In the first half of the year, China exported over 40,000 robots. The real challenge is whether they can adapt to unfamiliar environments and follow instructions effectively. Only by solving these problems in real-world scenarios can robots truly establish themselves.
4. The Next Phase of Robotics: From Hardware Competition to AI Power
In the past, robots were compared based on their hardware (number of joints, stability of movement). Now, the focus is on their AI capabilities (understanding natural language and handling unknown situations). For example, as Ubot CEO Wang Xingxing mentioned, robots in the "ChatGPT era" can enter a room they've never been to and complete tasks like pouring a glass of water with just a command. This requires powerful physical AI models, such as NVIDIA's GR00T. China's companies have an advantage: they possess the world's most complete manufacturing supply chain and a vast range of industrial applications (logistics, retail, etc.) for testing, which other countries do not.
5. What is Physical AI? The Technology That Enables Robots to Understand the Real World
In simple terms, large models (like ChatGPT) enable AI to communicate and write, while physical AI allows robots to understand the real world – knowing that a glass will break if dropped, avoiding obstacles when walking, and using the right amount of force when picking up objects. NVIDIA's Omniverse platform acts as a virtual training ground for robots, allowing them to simulate tasks thousands of times before executing them in reality, thus reducing mistakes. The physical AI division led by Huang Minshan is NVIDIA's next growth engine, and robots are the carriers of this technology. This is why she is actively visiting Chinese robotics companies.
The essence of this news is that global tech giants (like NVIDIA) are betting on the future of robotics through physical AI, with China playing a central role in this competition. China offers both the capital momentum from the IPO boom and the industrial foundation needed to develop advanced robots. The next challenge is to determine which company can first equip robots with intelligent "brains."