Ten "Genius Youth" who left Huawei

When gathered, they are the "Genius Teens"; when dispersed, they are stars in the sky.

Saying "Genius Teens" is quite apt. This talent program launched by Huawei during its most difficult times was once highly prominent, attracting top - tier talents with high - profile recruitment, and luring the most talented and capable young Chinese people to Songshan Lake, the location of Huawei's headquarters.

However, people come and go like flowing water. Especially in the new cycle of the AI technology explosion, the "Genius Teens" have scattered across the industry, becoming stars in the Chinese AI wave.

Some have started their own businesses, while others have returned to universities.

Just a simple count reveals 10 Huawei Genius Teens in the industry.

When gathered, they are the "Genius Teens"; when dispersed, they are stars in the sky

According to public information, currently, the geniuses have two notable career directions - the entrepreneurship group and the university group.

△

Let's start with the entrepreneurship group.

"ZhiHuijun" Peng Zhihui

Let's first look at Peng Zhihui. He joined Huawei through the "Genius Teen Program" in 2020 and was responsible for the full - stack R & D of Huawei's Ascend computing product line.

During his tenure at Huawei, Peng Zhihui was praised by Ren Zhengfei in a speech, who said he was "the driving force for Huawei's innovation."

However, at the end of 2022, Peng Zhihui chose to resign from his job with an annual salary of 2 million yuan and became the co - founder and CTO of ZhiYuan Robotics.

Besides Peng Zhihui, there are also many former Huawei executives in the founder team of ZhiYuan Robotics.

1. Deng Taihua, the chairman and CEO of ZhiYuan, once served as the vice - president of Huawei Technologies Co., Ltd. and the former president of the computing product line.

2. Qiu Heng, the COO of ZhiYuan, once served as the CMO of Huawei's China enterprise business.

3. Jiang Qingsong, the vice - president of marketing at ZhiYuan, once served as the head of the product management department of Huawei's P & S solutions.

Now, ZhiYuan Robotics has built three robot families: "YuanZheng, JingLing, LingXi" and jointly won a 124 million - yuan procurement order for humanoid robots from China Mobile with Unitree Technology in the first half of this year. The company expects to ship thousands of units this year.

The capital market is also very optimistic about the prospects of ZhiYuan Robotics. As of now, ZhiYuan Robotics has completed 11 rounds of financing, with 3 rounds this year alone.

In March 2025, ZhiYuan Robotics received a new round of financing led by Tencent; in May 2025, JD.com and the Shanghai Embodied Intelligence Fund participated in its new round of financing, and many old shareholders such as SAIC increased their investment; on August 1, 2025, ZhiYuan Robotics announced a new round of financing, jointly led by LG Electronics and Korea Investment Group.

So, what makes Peng Zhihui, the CTO of ZhiYuan Robotics, different?

Peng Zhihui graduated from the University of Electronic Science and Technology of China for both his undergraduate and master's degrees. After graduation in 2018, he joined the I Laboratory of OPPO Research Institute as an algorithm engineer.

Besides being good at algorithms, Zhihui also has some talent in self - media.

His Bilibili account "ZhiHuijun" has posted several demonstration videos of self - made robots - a mini robotic arm that can sew grapes, a bicycle equipped with autonomous driving, and a coin - sized mini Linux computer - all of which have received over one million views.

Especially the video of the autonomous bicycle, which has received over 5 million views on Bilibili. Even "the father of Java," James Gosling, and Ren Zhengfei gave it a thumbs - up!

Ren Zhengfei was deeply impressed by his video and used it as a case in a discussion:

For the 2012 Laboratory, the company has never imposed too many restrictions on you. For example, someone is researching the autonomous driving of bicycles, and the company has never restricted him. Do we want to produce bicycles? No. This is a "scalpel" he has mastered, which may play a role and generate great commercial value in the future.

James Gosling also couldn't hide his admiration for this invention on Twitter:

Although I can't understand Chinese, I can understand engineering. The autonomous bicycle is really well - made.

As a member of the "Genius Teen Program", ZhiHuijun is the most well - known one among the public. After all, not everyone can become one of Bilibili's Top 100 UPers, especially through hardcore innovation.

Chen Yuanpei

Next is Chen Yuanpei, a post - 00s generation. He passed the assessment of Huawei's "Genius Teen Program", but there is no information indicating his work content at Huawei.

In 2024, Chen Yuanpei co - founded LingChu Intelligence with Wang Qibin and Chai Xiaojie.

LingChu Intelligence is committed to creating a "thinking hand". As a co - founder, Chen Yuanpei is mainly responsible for technology R & D and related planning on the scenario side. He focuses on the coordination between commercialization and technology, aiming to create a commercially closed - loop project.

Currently, LingChu Intelligence has launched core models such as Psi - R0, Psi - R1, enabling robots to autonomously reason and perform dexterous operation tasks in an open environment.

You may not be familiar with just the models, but the mahjong robot that was surrounded by people at this year's WAIC is from LingChu Intelligence.

In November 2024, LingChu Intelligence received angel - round financing led by Hillhouse Capital and BlueRun Ventures. It is worth mentioning that ZhiYuan Robotics founded by ZhiHuijun is also a shareholder. After that, LingChu Intelligence joined NVIDIA's Inception Program and became a part of Huang's circle of friends.

Who would have thought that Chen Yuanpei, who is now deeply involved in model development for LingChu Intelligence, majored in civil engineering in his undergraduate studies.

Chen Yuanpei graduated from South China University of Technology for his undergraduate degree. In his sophomore year, he participated in a RoboMaster robot competition and was hooked ever since. Moreover, in order to have more time to study robots, he even skipped civil engineering classes to teach himself about robots.

The competition trained his engineering skills. Chen Yuanpei hopes to explore more cutting - edge fields and conduct scientific research on his own. So, he went to Peking University for his master's degree and studied dexterous hands for a year under Professor Yang Yaodong. After that, Chen Yuanpei went abroad for further studies and conducted research as a visiting scholar at Stanford University under the guidance of AI godmother Fei - Fei Li.

When asked why he didn't continue his Ph.D. in the United States but chose to return to China to start a business, Chen Yuanpei gave three reasons.

1. Embodied intelligence is essentially an engineering discipline that requires sufficient resources and a strong team to do well. Universities are more about frontier exploration. To create a product that can really enter thousands of households, one has to be in the industrial circle. 2. Now is a very good era. There is no seniority - based system in AI. It is essentially technology - driven, and everyone with strong learning ability has a chance. The country also strongly supports high - tech entrepreneurship. 3. Maybe because I'm from Guangdong, I can't get used to the food abroad. I find life too painful. Haha.

It seems that Chinese food is more to the taste of Chinese young people.

Ji Yu

Let's take a look at Ji Yu. As a member of Huawei's "Genius Teens", Ji Yu joined Huawei's Hisilicon department, mainly engaged in the design and optimization of AI chip compilers and was responsible for several Ascend compiler projects.

In 2023, Ji Yu left Huawei and co - founded XingYun Integrated Circuit and served as the CEO. Most of the core team members are from Tsinghua University and top global chip companies.

Yu Hongmin, the CTO of XingYun, once served as the general person - in - charge of Huawei Hisilicon's in - vehicle Ascend chips. In addition, he also led the architecture, design, implementation, and mass - production deployment of several advanced - process data center chips.

XingYun focuses on the R & D of large - model inference chips, aiming to promote the reconstruction of computer forms in the AI era and achieve the miniaturization of large - model infrastructure.

Ji Yu believes that the implementation of large models is restricted by the cost structure of "high fixed costs + relatively high marginal costs". However, when the computing power cost is reduced to the consumer - level, AI will enter a period of ultra - high - speed explosion similar to the logic of information products.

All the dilemmas encountered in the current AI commercialization exploration can find references under the assumption that "mainframes are the carriers of the PC industry and the Internet industry".

In April 2025, XingYun launched a new large - model inference all - in - one machine, "Brown Ant", with a minimum price of 149,000 yuan. It supports running the full - blooded version of the DeepSeek R1/V3 large model, with a maximum dialogue speed of 28 tokens/s, and is compatible with mainstream models such as Alibaba's Qwen3 series.

XingYun is currently a strong competitor in the domestic AI chip market. At the end of 2024, XingYun completed angel - round and angel + round financing totaling hundreds of millions of yuan. Zhipu AI is also one of the shareholders. In August 2025, XingYun received another 16.7 million yuan in strategic financing.

Ji Yu was admitted to the Department of Physics at Tsinghua University through a competition and later obtained a Ph.D. from the Department of Computer Science at Tsinghua University. His research focuses on architecture and AI chips. During his studies, he published several top - conference papers and a paper in the main issue of the top - tier journal Nature and won the Outstanding Ph.D. Thesis Award from the China Computer Federation (CCF).

Wang Naixing

Wang Naixing is also one of Huawei's "Genius Teens" and once engaged in research related to AI computing power chips at Huawei.

Wang Naixing's motivation to start a business was literally "forced" by the United States.

In 2019, during the dark period of large - scale technological sanctions by the United States, Wang Naixing worked at Siemens EDA in the United States, which made him deeply feel the suffocation of being "choked" by others from a second - person perspective.

Therefore, Wang Naixing went to Songshan Lake, the most intense battlefield in China at that time. During his four - year tenure at Huawei, Wang Naixing focused on industrial software R & D, aiming to solve the "choke - point" problem.

At the end of 2022, the GPT era arrived. Wang Naixing keenly noticed the strategic high - ground of high - performance computing power optimization - traditional EDA manufacturers only focus on the peak performance of chips, while ignoring the health management and energy - efficiency coordination of chips throughout their entire lifecycle.

Around 2023, the rise of large models promoted the explosion of the AI industry. The systematic optimization of computing power infrastructure became the strategic high - ground for global technology companies - traditional EDA and server manufacturers only focus on the peak performance of chips, while ignoring the health management and energy - efficiency coordination of chips throughout their entire lifecycle.

Wang Naixing believes that chip full - lifecycle management (SLM) is the key to solving this pain point. So he left Huawei and founded BoSiXinYu as the CEO.

The core R & D team of BoSiXinYu is from Huawei Hisilicon and the Institute of Computing Technology of the Chinese Academy of Sciences. More than 85% of the company's R & D team members have post