E06. Zhang Wei: Forget the Leaves, Flourish in Growth

🎙️ Episode Introduction

This episode features Professor Wei Zhang, founder of LimX Dynamics, an Oasis Capital portfolio company. We explore the three great leaps of his life: from USTC to elite American universities; from engineering training to theoretical research, and on to becoming a professor at Ohio State University, where he and his students learned from each other; and finally, his decision to leave behind tenure and return to China to start from scratch, pushing the frontiers of embodied intelligence.

We asked him: facing such momentous choices, how do you stay so certain?

His answer: see the trunk, forget the leaves. In his view, America's technological iteration moves too slowly, while China sits on the main trunk of development.

In Wei's story, we see how a scientist upholds the spirit of truth-seeking through "training core strength, not tricks"; how he maintains abundance and self-sufficiency within stability; and how, through the transformation of identity, he redefines himself and relearns on the path to becoming an entrepreneur and business leader.

👤 Guest Bio

Wei Zhang: Founder and CEO of LimX Dynamics, and current professor at Southern University of Science and Technology.

He excels at "learning math through an engineering lens" and "explaining theory through a user's perspective" — and even more so at reconstructing problems from chaos. He believes the essence of technology is serving real problems, not proving status superiority.

After graduating from USTC, he pursued advanced studies in the US, earning his PhD from Purdue University with a minor in mathematics. His research spanned control theory, stochastic dynamics, and artificial intelligence. He served as a postdoctoral researcher at UC Berkeley, then joined the faculty at Ohio State University for many years, where he received tenure as an associate professor.

In 2020, he resolutely gave up his US tenure and returned to China to found LimX Dynamics, dedicated to exploring disruptive innovation in embodied intelligence and advancing the productization of full-size general-purpose humanoid robots. He also proposed the "IDS Ecosystem Synergy Strategy" to serve innovators, developers, and system integrators, driving technological innovation, development, and broad application of embodied intelligence in research, manufacturing, commerce, home, and other domains.

🕒 Selected Timestamps

05:30 The person with the worst math scores — how did he build a lifelong learning model?

07:39 Reading one math book for a month: the confidence that comes from strength training

10:26 The closed loop of engineering and first principles: finding the "bookshelf" of a problem's essence

13:20 Moving AI game experiments to drones: early explorations at the intersection of robotics and game theory

17:24 A teaching revolution for Kalman filters: the essence is "forgetting the leaves"

23:33 Don't chase what's hot; only do what you love

31:22 The key decision function: maximize the ceiling, not the floor

35:30 The "hooligan class monitor's" intuition for diversity: people are different

38:24 Assume I already have 10 billion: the prerequisite for making choices

42:00 Forget the professor identity; become human again

47:22 The empty-cup mindset of entrepreneurship: experience is an illusion; you must relearn

53:20 Missing the timing, missing the window: how to quickly adjust strategy

56:06 The key to unlocking organizational capability: concrete, direct, real

1:04:19 The "Windows moment" for humanoid robots: the eve of an API ecosystem

1:11:24 How does a scientist become an entrepreneur? First clarify your position

1:16:25 Holding ground is not the answer; it's about continuously updating your foundation model

1:24:47 Learn to celebrate every tiny step forward

1:28:00 Behind all regret lies the process of growth

1:33:00 The founder's state is the integral of the organization

📚 References

Kalman Filter

An algorithm for state estimation in dynamic systems, foundational to robotics, navigation, and signal processing. Wei uses this as an example in teaching methodology, emphasizing that "clarity of principle matters more than memorizing formulas."

AlphaGo and Capture the Flag

The former represents a milestone in deep reinforcement learning; the latter is a DeepMind AI game experiment based on adversarial strategies. Wei's team similarly used drones to simulate game-theoretic tasks in the early days of embodied AI research.

Boston Dynamics

The globally leading robotics company, renowned for its humanoid and quadruped robots.

Foundation Model

Originally referring to AI pre-trained models (such as GPT), Wei extends this concept to personal "cognitive strength training," arguing its value lies in the sustainable ability to generate responses to complex problems.

Windows Moment

An analogy from PC history: from writing assembly to using GUIs. Wei believes humanoid robot operating systems are currently in a "DOS to Windows" transition period, laying the groundwork for a developer ecosystem.

🎵 Music

Jordan Critz - Beau Et Rapide (Piano)

🎤 Production Team

Host | Jinjian Zhang

Produced by | Oasis Capital

Editing & Production | Shendu Studio Podcast Workshop

💬 Join the Conversation

Are you building your own "Foundation Model"?

Wei says his life's work was gradually found through continuous "core strength" training. He calls it his "Foundation Model" — a capability system that can adapt, fine-tune, and generate answers across different environments.

🎧 After listening to this episode, take a moment to ask yourself: What approach are you using to build your own "Foundation Model"?

If you'd like to join our community and chat with fellow listeners, scan the QR code to add us and note "Signal & Noise Listener Group" 🩷

Disclaimer

All investment-related content in this podcast is for exchange and sharing purposes only, for reference, and does not constitute any market prediction, judgment, or investment or consulting advice. Thank you for your interest in this podcast's original content! If you wish to reprint or quote from this podcast, please indicate the source. Prior written consent from Oasis Capital is required before reprinting.

[View episode transcript on Xiaoyuzhou]