Meet the Four Post‑90s Chinese AI Pioneers Sitting Beside Jensen Huang at GTC

Observing the AI industry, a more insightful indicator than financial reports is who sits beside Jensen Huang at Nvidia's GTC. The four young Chinese faces who appeared at the 2025 Beijing thank‑you dinner and later at GTC 2026 map directly to the hottest AI tracks of the following year.

By 2026 the list expanded, and four individuals stand out. They align with Huang's repeated bets on reasoning large models, embodied intelligence, and brain‑computer interfaces, and share the traits of being young, Chinese, and turning academic papers into companies.

1. Yang Zhilin: The Sole Independent Large‑Model Representative on the GTC Main Stage

Yang Zhilin secured a seat no other independent startup obtained at GTC 2026, joining a roster dominated by AI leaders from Tesla, DeepMind, Cursor, and Runway. In 2025 his startup Moonlight Darkside closed a $500 million Series C round, reaching a post‑money valuation of $4.3 billion, with IDG Capital leading and Alibaba & Tencent over‑subscribing.

Later that year Kimi K2.5 achieved state‑of‑the‑art performance on Hugging Face, topping the leaderboard. At GTC Yang disclosed Kimi’s roadmap, emphasizing three resonant dimensions: token efficiency, long context, and agent swarms. He argued that scaling now requires simultaneous gains in compute efficiency, long‑range memory, and automated collaboration.

Yang’s background includes topping his class at Tsinghua Computer Science, completing a four‑year PhD at Carnegie Mellon under former Apple AI head Salakhutdinov and Google’s Cohen, and authoring seminal papers such as Transformer‑XL and XLNet. He has worked at FAIR and Google Brain and collaborated with a Turing Award winner, positioning him as a classic technical prodigy.

Beyond the standard profile, Yang founded a rock band and named his company after Pink Floyd’s “The Dark Side of the Moon.” He describes his team as “stubborn AGI purists,” viewing rock and innovation as the same disruptive force.

2. Wang Xingxing: From Huang’s Dinner Table to the GTC Stage

Wang Xingxing represents “embodied AI.” At the 2025 Beijing thank‑you dinner he posted a brief photo with Huang, symbolizing the rise of Chinese AI and robotics firms. His hometown of Ningbo shares Huang’s ancestral roots in Zhejiang.

In 2026 GTC Taipei, Huang co‑announced a humanoid robot reference design based on Nvidia’s Isaac H2 Plus platform, highlighting the Chinese robotics company Yushu (宇树) as a core partner in the Isaac ecosystem.

Wang’s journey began in 2009 at Zhejiang Science & Technology University, where he built a bipedal robot as a freshman. During his master’s at Shanghai University he created a low‑cost quadruped robot XDog, gaining early recognition. Yushu later produced high‑volume quadruped robots and a series of humanoids (H1, G1, H2) that appeared on Chinese New Year TV specials.

He coined the term “humanoid robot large model,” arguing that large language models alone are insufficient for robots; visual, joint‑command, and lidar data, together with imitation‑learning training, must be integrated. He identified data scarcity as the main bottleneck and advocated building affordable, functional bodies first to generate the necessary data.

3. Wang He: The Scholar‑Entrepreneur Putting Robots to Work in Factories

Wang He focuses on the “brain” and “work” aspects of embodied intelligence. At the same Beijing dinner he sat beside Huang as CTO of Galaxy General (银河通用), a showcase of Nvidia Isaac applications.

His résumé includes a physics competition gold medal that earned him admission to Tsinghua Electronics, work on robotic arms and visual navigation, a PhD at Stanford under ACM Fellow Leonidas Guibas, and the NOCS model that reshaped grasp generalization (ICCV 2023 Best Paper candidate). He later returned to China as a professor at Peking University and director of embodied‑intelligence labs at Zhiyuan.

In 2023 he co‑founded Galaxy General, choosing a wheeled chassis over legs to prioritize practical manipulation over flashy mobility. The company’s robots now operate on automotive assembly lines, 24‑hour retail warehouses, and a 24‑hour smart medical warehouse, leading to a $300 million Series C round in December 2025 that pushed valuation to $3 billion.

Wang balances academic teaching (courses on computer vision and embodied intelligence) with running a multi‑billion‑dollar company, turning his research on generalization into a commercial moat.

4. Zhu Yixin: Re‑Rendering Vision Directly into the Brain

Zhu Yixin tackles the extreme question of enabling a blind person to “see” by bypassing the eyes and feeding visual information straight to the brain. At GTC 2026’s China AI Networking Day he discussed his work with Nvidia’s senior director of network technology.

He co‑founded NeuroVision, a brain‑computer‑interface startup focusing on invasive visual cortex implants. Their “neural graphics card” uses a multi‑channel high‑process‑node neural stimulation chip to rewrite visual cortex activity, effectively rendering external scenes into neural signals. The company recently closed a seed round of nearly ¥100 million.

The venture’s challenges stem from three sources: scarce clinical visual‑cortex data, state‑of‑the‑art decoding and reconstruction algorithms, and industrial‑grade chip fabrication. Zhu’s expertise in visual cognition and neural decoding bridges these gaps.

Conclusion: The Logic Behind Huang’s Invitations

Collectively, the four figures illustrate a clear trajectory in Huang’s focus: Yang Zhilin pushes reasoning large models deeper, Wang Xingxing adds a physical body, Wang He makes that body productive, and Zhu Yixin attempts to feed intelligence back into the biological brain. This chain—from silicon‑based reasoning to carbon‑based perception—mirrors Nvidia’s shift from a pure chip maker to an “AI factory” and physical‑AI infrastructure provider.

The growing presence of young Chinese innovators at Huang’s side signals both the centrality of this generation in global AI narratives and the strategic weight of Huang’s invitations as early bets on future industry directions.