Source Code Capital | "Curiosity" Is Ethan Yuanming Hu's Mentor at Taichi Graphics

#Source Code Capital

Ethan Yuanming Hu skipped the gauntlet of China's college entrance exams in his second year of high school, securing early admission to Tsinghua's prestigious Yao Class. From there he went to MIT, where he wrapped up his master's thesis in a single year before pressing on to his PhD. In 2019, still a doctoral student, he led the development of a programming language called "Taiji." After completing his doctorate, he returned to China and founded Taichi Graphics. He once went viral on Zhihu with a Frozen-like effect rendered in just 99 lines of code. Today we trace his six major "sharp turns" — the pivotal choices that shaped his path.

Livestream Excerpts

Source Code Capital Xingchen Zhang: What is the Taiji programming language?

Ethan Yuanming Hu: It depends on who you ask. For Python developers, Taiji is a tool that makes Python programs run faster. For university students, it's a tool that makes learning computer graphics algorithms and data-computing programs much easier. For researchers, it's an accelerator for scientific progress. For engineers working in AR, VR, short-video, or live-streaming, it's a system that lets you run dazzling physics effects on mobile devices.

Source Code Capital Xingchen Zhang: How is the open-source community for Taiji developing?

Ethan Yuanming Hu: We feel pretty good about it. User growth has stayed roughly 3.5x annually, which is fairly rapid for an open-source tool. Along the way, we've realized that building tools or products really comes down to communicating with users. As our user base has grown, many have reached out proactively — to exchange ideas, or to vent frustrations — and that's driven the community forward. As a founder, you have to maintain direct, first-hand contact with users. You have to actively ask what their biggest pain points are, what needs your product is meeting, and what can still be improved. Open source is done better overseas; some engineers will even fix bugs for you. Open source is a culture of openness and sharing, and we've genuinely enjoyed being part of it. We're especially hopeful that Taiji will become standard infrastructure for parallel computing within five years.

Source Code Capital Xingchen Zhang: You also have the Taiji Kaiwu cloud platform. How is that developing?

Ethan Yuanming Hu: User feedback on the Taiji Kaiwu cloud platform has been solid. The development workload is substantial and takes considerable time. What we discovered along the way is that in the graphics software industry, simply building an MVP to test product-market fit doesn't really work. The Lean Startup emphasizes rushing out a minimum viable product for user testing, but if you actually do just that — build the bare minimum — you often learn very little. You need to invest the effort to build what I'd call a "minimal awesome product," something users genuinely love.

Source Code Capital Xingchen Zhang: You've made six sharp turns in your entrepreneurial journey. The first was pivoting to physics simulation. What happened there?

Ethan Yuanming Hu: During my undergraduate internship at Microsoft Research Asia, I worked on deep learning, which was white-hot at the time. But I'd wanted since childhood to research how to simulate the world, so during my PhD I decided to return to that original passion. Even now, my work sometimes touches on deep learning, so my path has been this spiral of circling back and moving upward.

Source Code Capital Xingchen Zhang: If the first turn was sticking with a childhood passion, the second was choosing entrepreneurship after your PhD. Why?

Ethan Yuanming Hu: Three factors, roughly. First, some academic drift as my doctorate was winding down. Second, I've always loved exploration. Third, I wanted to give it a shot — and there were investors pulling me "into the water." Looking back, I actually enjoy struggling in the water. No regrets.

My late-PhD uncertainty was about what to do after graduating. But I could feel that I wanted to do something closer to everyday life, something that created convenience for more people and put my research into wider use. That something belonged in industry. At the time, I was watching videos of successful Chinese entrepreneurs in the US, wondering what building a company would feel like, what succeeding would feel like. In academia, your trajectory is visible — no thrill, no nervousness. But I wanted adventure. I wanted to try building something myself.

It so happened that my Zhihu post about the 99-line Frozen effect caught some investors' attention. They filled my head with business knowledge. Our Taiji open-source community was also developing well. Adobe and Unity were doing great at the time. So I chose entrepreneurship.

Source Code Capital Xingchen Zhang: Entrepreneurship is hard. Did you think about downside scenarios? Did you worry, feel anxious, or afraid?

Ethan Yuanming Hu: I did. I spent a long time deciding. Academia would have been more secure. My family is in academia too; staying at a US university as a professor would have met their expectations. But I kept coming back to what I truly wanted inside. If I flipped a coin — heads, stay in the US for academia; tails, return to China to start a company — while the coin was still in the air, I found myself hoping for tails. Since that was the case, why fear? I'd been at MIT for three and a half years and gotten a lot done. When I told my advisor I wanted to return to China to start a company, he was incredibly supportive. He really thought from my perspective, and I'm deeply grateful for his encouragement.

Source Code Capital Xingchen Zhang: How did you meet your co-founder?

Ethan Yuanming Hu: Kuang Ye and I met on GitHub. Before deciding to start a company together, we'd never met offline. I was in Boston, he was in Tokyo — thirteen hours apart. It was pretty remarkable. It happened through the Taiji open-source community. I sent him a message saying some investors had approached me and I thought Taiji might be viable as a company, asking if he had any interest. Five minutes later he said, let's talk. We talked, and things just moved forward. Finding the right co-founder is exceptionally lucky.

Source Code Capital Xingchen Zhang: Your third sharp turn was choosing to start your company in China. Based on your observations, how does entrepreneurship in the US differ from entrepreneurship in China?

Ethan Yuanming Hu: American universities have a very strong entrepreneurial culture, especially Stanford near Silicon Valley. In Boston, there are lots of biotech startups. American entrepreneurship celebrates small teams, emphasizing individuals, efficiency, and innovation — Instagram had only about ten people before being acquired. China's advantage is its rare single large market. Second, Chinese teams execute strongly; with proper organization, overall output is very high. That's an Eastern characteristic.

Source Code Capital Xingchen Zhang: Entrepreneurship brings all kinds of problems. First is the identity shift — from PhD to CEO. What changes did that bring? That would be your fourth sharp turn.

Ethan Yuanming Hu: First, in academia I faced peers with equivalent expertise. Building products, I face the general public, who don't yet know what your product is. You need to help users quickly understand who you are, and your product needs to save users time and provide convenience.

Second, in academia you're more often a solo operator, whereas entrepreneurship means uniting a group of people to do one thing well. Without leveraging team strength, success is impossible. Different colleagues need different management and communication approaches — some need you to check progress, some need timely encouragement, others need space. The CEO serves as a servant and supporter. These things you learn by doing. From PhD graduation to becoming a CEO, I stepped on plenty of landmines in people management. Really learning by doing.

Source Code Capital Xingchen Zhang: In that process of stepping on landmines, do you have any experience in identifying and using talent?

Ethan Yuanming Hu: Early-stage startup hiring must be "rather short than loose." I stumbled here — not setting proper standards, relaxing my grip and hiring people who turned out to be a bad fit, which was bad for both sides. So hiring standards must be set high.

For people, first is self-drive — they must want to do things well and deliver on commitments. I also look at whether they care about external evaluation; if they don't care at all, they're hard to motivate. Second is optimism and confidence. Startup development changes by the minute, with roller-coaster states. You need to face change positively. Third is whether you'd want to work with them — finding people you like and can collaborate with. It sounds intangible, but it matters. Like attracts like.

Source Code Capital Xingchen Zhang: Your team is now over 40 people. How do you drive internal innovation and creativity?

Ethan Yuanming Hu: I think it's often "from the masses, to the masses." The typical decision process: you have an idea, consult around and find it viable, explain it clearly to everyone, then execute. The earlier the company, the more it depends on the founder's conviction and creativity. Early on there are so many uncertainties; only the founder has a clear picture of what needs to be done. The hard part is describing that clearly to others. Founders must have firm conviction, know clearly what they want to do, and use that conviction to drive the team in the same direction. I haven't experienced larger organizations, but I believe at that stage, collective creation becomes necessary.

Source Code Capital Xingchen Zhang: Organizational growth is closely tied to personal growth — actually your fifth sharp turn. What's your view on personal growth? Any particularly striking moments?

Ethan Yuanming Hu: Since starting my company, such moments have happened many times. When my team grew from the initial 3–4 people to a dozen or twenty was one. Expansion happened so fast, and we hit many problems. At that point the company had moved beyond linear development, with different projects each having leads. If a project lead wasn't suitable, or had problems, or was inefficient — these all existed. I realized insight into people really matters, truly achieving "knowing people" and "using people." So sometimes I feel my learning speed isn't fast enough; otherwise our company's overall development would probably be faster. There's a Stanford professor who specifically teaches entrepreneurship; when he later started his own company, he made all the same mistakes he should have known to avoid. Theory and practice really are different.

Source Code Capital Xingchen Zhang: When you encounter people with very different views, internally or externally, how do you handle it?

Ethan Yuanming Hu: I believe more in human agency, in people's subjective initiative. Internally, persuading those around you and reaching team consensus takes effort and some technique. But our culture strongly emphasizes "results orientation." You'll encounter all kinds of difficulties along the way. Internally, we only recognize功劳 (merit/actual contribution), not 苦劳 (hard work without results). You must be accountable for results.

Source Code Capital Xingchen Zhang: The last sharp turn may be a question about the future. Whether in business success or product success, I particularly want to ask: First, how much do you want to win? Second, how big do you want to win?

Ethan Yuanming Hu: I definitely want to win, but not in terms of external recognition. My internal drive is wanting to do something well. If you only pursue external validation, you might get short-term recognition but long-term disapproval — that's not desirable. I just want to do what I set out to do, and succeed at it. Then there's responsibility to every person on the team. Everyone joined me in this; I want them to succeed, and I want every colleague in our team to realize their value.

How big — we've always said we want to build a company like Adobe. But gradually you realize how hard that is, so we take it step by step. Entrepreneurship has taught me not to take myself too seriously. Every day I think about many things we haven't done well, even done quite poorly. Just step by step — first make a product people love, then build a product matrix, and finally hope to achieve something like Adobe in my lifetime. Opportunity has random elements too. When it comes, seize it. Do your best and leave the rest to fate.

Source Code Capital Xingchen Zhang: What do you think is a company's lasting, real value?

Ethan Yuanming Hu: Continuously making good products that users keep recognizing. For society, products that are era-defining and drive social progress. For users, products they love that make life more convenient — that's lasting, real value. What Taichi Graphics wants to do is create something that didn't exist before, bring it into being through our team, and have people elsewhere in the world copy us. That feels worth doing. Our name "Taiji" expresses this too. I love Chinese culture, and Chinese people have tremendous creativity.

Source Code Capital Xingchen Zhang: What advice do you have for researchers with entrepreneurial ambitions, or young PhD students?

Ethan Yuanming Hu: My advice is: be self-aware. From research to entrepreneurship, from lab to users — there's a huge gap. User experience is something you must pursue. You can't be like in the lab, where if results are unsatisfying you just move on to something else or another application scenario. Building products is hard. Entrepreneurship is harder than research.

Engage with Source Code Capital

How did you learn about Source Code Capital? How did you find this episode of the "Leap of Faith" livestream? Did Xingchen Zhang's conversation with Ethan Yuanming Hu inspire you? We welcome your comments — we'll select 10 readers with the most likes to receive curated books.

Wednesday, October 26, 8:00 PM: Source Code Capital "Leap of Faith Livestream" Episode 07, featuring Longbo Chen, Founder & CEO of Aila Cloud Technology, exploring the differences between "big company entrepreneurship" and "real" entrepreneurship.

• Source Code Capital | I'm 28 this year, and I've been in power supplies for ten years and one month
• Source Code Capital | Guangmi Li of Shixiang: Choosing Entrepreneurship Was My Leap of Faith
• MaGuan | Power Laws — Understanding Complexity in a Non-Random World
• MaGuan | Source Code Capital's Dual-Carbon Investment: Exploring New Species Through Full-Industry-Chain Layout
• MaGuan | China Electric Power News Exclusive Interview with Source Code Capital on Wang Di: Industry Inflection Point Arrived, Vast Market Space Ahead
• MaGuan | Source Code Capital's Yungang Huang: How Does AI4S Land in Different Stages, Industries, and Markets?
• MaGuan | Source Code Capital's Yungang Huang: "Cross-Disciplinary Intersection" Drives Innovation in Health and Medical Industries
• MaGuan | Source Code Capital's Wang Di: The Most Needed Capability for Investing in Carbon Neutrality Is Multi-Level Cognition
• MaGuan | Source Code Capital's Wu Fan: Opportunities and Advantages of China's New Energy in a New Cycle
• MaGuan | "Carbon Neutrality" Technology Innovation in Sustainable Development

Click "Read Original" for more MaHui members