For three months, Huawei's talented teenager created a unicorn and was robbed crazy by VC

Guide

THECAPITAL

It has only been half a year since Zhihuijun left Huawei.

This article is 4216 words, about 6 minutes

Author | Hyacinth Edit | My people

Source | Rongzhong Finance

（ID：thecapital）

I heard that investors don't like entrepreneurs from big factories, so they have to see if they are good enough, which makes Li Yanhong fancy the "Huawei genius boy" Zhihuijun.

Recently, Zhihuijun's entrepreneurial project Zhiyuan Robot was exposed to have completed the third round of financing, and the Tianyancha App shows that the new shareholders in this round include Li Yanhong's Sanya Baichuan Zhixin Private Equity Investment Fund Partnership (Limited Partnership), as well as first-line VCs such as CDH, Jingwei, and Gaorong, with the registered capital increasing to about 48.4702 million yuan.

Zhiyuan Robot's affiliated company is named Shanghai Zhiyuan New Technology Co., Ltd., established on February 27, 2023 in Shanghai Lingang New Area, the legal representative is Shu Yuanchun, and its business scope includes the research and development of intelligent robots, artificial intelligence theory and algorithm software development, artificial intelligence hardware sales, electronic product sales, etc.

It is reported that the valuation of the current round of Zhiyuan robot market is more than one billion US dollars, which means that the company established only 3 months ago has officially entered the ranks of unicorns, and really responded to the sentence, "someone was born in Rome".

01

Dive into AI robots

At the end of last year, Zhihuijun, a blogger who had joined Huawei as a "genius boy", said on Weibo that he had left Huawei and would start a new career to do "more challenging things", and this more challenging thing is AI robots.

Robot is the product of "AI + mechanical + electronic" common progress, since the birth of ChatGPT, AI technology has undergone qualitative changes, which has brought unlimited imagination to robots, and the combination of general robots and large language model technology has become an important opportunity for robot entrepreneurship.

Large language models (LLM) refer to deep learning models trained using large amounts of text data to generate natural language text or understand the meaning of language text. Large language models can handle a variety of natural language tasks, such as text classification, question answering, dialogue, etc., which is an important path to artificial intelligence.

At present, an important channel to understand Zhiyuan robot is the content released by Zhihuijun on station B. In April, Zhihuijun released his latest homemade robot "neZHa" on station B, which is the first video he released after leaving Huawei, mainly explaining how to design robots and some of the innovations he made, and he also predicted that the robot will be officially released in a few months.

Next, Zhiyuan Robot began to recruit intensively, on April 5 about Zhiyuan Robot recruitment information posted on a number of public accounts, the recruitment information introduction, Zhiyuan Robot (Agibot) is a company focusing on the development of general humanoid robots and embodied intelligence, committed to the deep integration of AI and robots, long-term investment to create competitive robot products and ecology. The company's vision is to create the next generation of mass-produced robot products with high intelligence and practical value that can serve the human workforce in the real world.

When describing the attractiveness of the company, the job posting wrote that the exciting direction: artificial intelligence is ushering in a new round of industrial upheaval, accelerating in the direction of AGI holy grail, LLM+Robotics combination has become an inevitable trend, robots will usher in the inflection point of application scenarios in the near future. Top expert team: At present, the team has a large number of industry bulls, domestic and foreign first-line manufacturers and scientific research institutes related to high-level experts in related fields, can work with excellent people. Competitive salary: technology competition is talent competition, we provide excellent people with more than the first-line Internet manufacturers of generous monthly salary + year-end + options, more excellent people can be interviewed, salary is not capped!

From the recruitment advertisement of Zhiyuan Robot, I believe that many readers can also feel the passion of this group of entrepreneurs and their firmness in the direction of AI robots.

From the perspective of specific recruitment positions, the positions recruited by Zhiyuan Robot include large model LLM algorithm engineer, robot motion control algorithm engineer, hardware engineer, C++ software development engineer, embedded development engineer and other 5 positions, of which algorithm engineers require master's degree or above, and the other three positions require more than 3 years of work experience.

Regarding Zhihuijun's AI robot entrepreneurship, there is an article in Zhihu about "How to treat Zhihuijun's entrepreneurial Zhiyuan robot, what is its prospect?" In the question section, many professionals expressed their views on this matter from the perspective of attitude and professionalism.

From a sentimental point of view, many people have a supportive attitude towards this matter, and one of the most liked answers is: I very much hope that our country can also have a robot company, or laboratory, or research institute comparable to Boston Dynamics. Since 2020, I have personally worked hard with this goal in mind. Seeing that Zhihui is doing this thing, I really feel a lot of emotion, first, I envy him for having such resources, second, I am moved by our naïve idealists, and third, I admire his courage to not break through. Regardless of the outcome, I wish you a good time!

Some people also think that Zhihuijun's steps are too big, and some people say that the ambition of the team can be seen from the recruitment information, including the operating system, motion control algorithm, and AI direction. In fact, each of these directions is a small industry, and it is understandable if it is a strategic plan for initial financing of entrepreneurship. But if you're really going to get started, it's better to focus on something.

From a professional point of view, everyone expressed relatively consistent concerns, all believing that the human and financial resources required for the research and development of AI robots are incalculable, and it is very difficult to overcome any link such as LLM and operating system, and considering the popularity of the market, it is difficult to land in the short term. Some people said that doing LLM is an obvious way to catch the trend, and it feels very unsteady. There are still many unsolved problems with humanoid robots, from perception to control of these problems it takes a lot of money and time to do, and it is not the turn to do LLM, Musk's humanoid robots now have no sound, let alone Zhihuijun.

To sum up, Zhihuijun's choice of AI robot entrepreneurship is exciting, the difficulty is unimaginable, this is a game of genius.

02

Wild Iron Man

There is a biography about Musk, Chinese translated as "Silicon Valley Iron Man", and Zhihuijun is nicknamed "Wild Iron Man" by netizens, what is going on?

Zhihuijun, formerly known as Peng Zhihui, was born in 1993 in an ordinary working-class family in Ji'an, Jiangxi Province, and he has been known as "someone else's child" since he was a child, and has won many competitions in physics in middle school.

In 2015, he received his undergraduate degree in biomedical engineering from the University of Electronic Science and Technology of China, and because he had been teaching himself computer knowledge, he switched to information and communication systems as a graduate student.

After graduating with a master's degree in 2018, he went to the OPPO Research Institute, when he got two offers for both hardware and algorithm jobs. After joining OPPO Research Institute, he became an algorithm engineer in the AI Lab.

In 2020, after seven rounds of interviews, including three rounds of technical interviews, expert interviews, supervisor interviews, technical vice president interviews, and president interviews, Zhihuijun became a member of Huawei's "Genius Youth" team, with an annual salary of more than 1 million yuan.

Huawei's "Genius Youth" program is a project launched by Huawei founder Ren Zhengfei in 2019 to attract top talents with top challenges and top salaries. Ren Zhengfei once mentioned that "genius boys" should "be like 'loaches', drilling into our organization and activating our team." ”

In 2019, Ren Zhengfei issued an email from the president's office, announcing the implementation of annual salary management for the 8 "talented teenagers" that year. The highest annual salary is 1.82 million to 2.01 million yuan; It was followed by 1.405 million to 1.565 million yuan, and finally 896,000 to 1.008 million yuan.

From 2020 to 2021, Huawei recruited about 26,000 fresh graduates, of which more than 300 were "talented teenagers" defined by Huawei.

On April 25, 2022, Huawei issued a global "Genius Youth" convocation order, at which Huawei introduced that the "Genius Youth" recruited by Huawei are not limited to education or school, and as long as they have relevant knowledge and technical capabilities in mathematics, computers, physics, materials, chips, intelligent manufacturing, chemistry, and other fields, they will have the opportunity to join.

Ren Zhengfei's selection criteria for talented teenagers are to allow some 'weird' students to grow, and the most important thing is to see his problem-solving ability, not his comprehensive quality.

After joining Huawei, Zhihuijun became a member of the Ascend division of the computing product line, and the external title given to him by Huawei was Ascend AI edge computing expert.

However, compared with the unreachable of genius, the most "close to the people" side of Zhihuijun is his UP main identity on station B, which currently has 2.43 million fans and a total of 59 videos, most of which are robot-related.

According to himself, he initially only used station B as a video storage server, but unexpectedly gained a group of stable fans, so he began to transform into an up master.

In 2021, Zhihuijun's self-driving bicycle became popular online. The bike is equipped with lidar, which can go on the road on its own, avoid obstacles, and maintain balance, and the final hardware cost is less than 10,000 yuan. In the end, the video had nearly 5 million views. The self-driving bike was also mentioned by Ren.

In addition, he released homemade game charging treasure, Iron Man robotic arm, modular mechanical keyboard, etc. have a good number of views, due to the video out of the circle, in 2021 he also won the "Annual Top 100 UP Master", all hardware video content, he was also nicknamed "Wild Iron Man" by netizens.

At the end of last year, the "wild iron man" who loves hardware DIY said goodbye to the "Huawei genius boy" and devoted himself to entrepreneurship, but what kind of industry is he facing?

03

Robots in the era of strong AI

Robots have entered the era of strong AI.

Since the birth of the world's first programmable robot "Unimet" in 1954, robots have undergone nearly 70 years of long exploration and have now entered the third generation of robot era.

The first generation of robots is a remote control machine, can not leave human control to move alone, this generation of robots have memory, storage capabilities, according to the corresponding procedures to repeat the operation, but the surrounding environment basically has no perception and feedback control ability, also known as teaching reproduction robots.

The second generation of robots is a sentient robot, and this kind of robot with feeling is a sense of a certain function similar to a human, such as: force, touch, hearing. The second-generation robot relies on feeling to judge the size of force and the sliding situation, and when the robot is working, it flexibly adjusts its working state according to the information obtained by the sensory organs (sensors) to ensure that the work is completed under the condition of adapting to the environment. For example, a tactile robot can easily grasp eggs, and a robot with a sense of smell can distinguish between different drinks and alcohol.

The third generation of robots is the "intelligent robot" currently under research, which is an advanced robot that uses various sensors, measuring instruments, etc. to obtain environmental information, and then uses intelligent technology to identify, understand, reason and finally make planning decisions, and can act autonomously to achieve predetermined goals. Its future direction is to be sentient, thoughtful, and able to talk to people. This generation of robots already has autonomy, with self-learning, reasoning, decision-making, planning and other capabilities.

At present, the fourth largest robot is still in the conceptual design stage, and the fourth-generation robot in the concept is an intelligent robot with learning, thinking, and emotions.

The third generation of robots in these characteristics is based on the common progress of "AI + mechanical + electronic" technology, these three cooperate with each other to jointly realize the function of the robot, these three directions also constitute the upstream of the current robot industry, of which the AI part includes computer vision, natural language processing, machine learning, planning and decision-making and other technologies; The electronic part includes sensors, actuators, control systems, chips, etc., which are responsible for connecting AI and mechanical parts to form a unified control and response system; The mechanical part includes the structure, motion devices, and transmission devices of the robot, and is responsible for executing the decisions and instructions made by AI.

Robots are used in a wide range of applications, such as home services, transportation and logistics, industrial manufacturing, specialty, medical, agriculture, adventure and scientific research, entertainment, education, etc.

Home service sweeping robots, lawn mowing robots, swimming pool cleaning robots are currently relatively mature, the growth rate is relatively fast, the penetration rate still has a large space, taking sweeping robots as an example, according to Euromonitor data, the global and continental sweeping robot retail sales in 2022 are 16.27 million units and 4.67 million units, 2010-2022 retail CAGR is 21%, 37%. According to iRobot's annual report, in 2020, the penetration rate of sweeping robots in developed countries in Europe and the United States such as the United States, Germany, and France is about 10-15%, and the penetration rate in the UK is about 1%~2%. According to CICC's calculations, the penetration rate of urban household sweeping robots in China will only be close to 10% in 2022.

At present, the more imaginative robots are emotional escort robots, home care robots, stir-fry robots, massage robots, etc., these applications or need strong AI, or need strong computing power, or need, taking emotional escort robots as an example, the core of emotional escort robots lies in human-computer interaction capabilities, before the development of AIGC, human-computer interaction ability is weak, can not provide the basic functions of emotional escort; The second is in the appearance design and mechanical movement ability, such as bionic robots will have more obvious advantages in emotional accompaniment.

The large language model represented by ChatGPT may become the "brain" of the robot, using its language understanding and generation ability and logical reasoning ability to empower the robot to understand and execute natural language instructions, thereby breaking the bottleneck of robot development and bringing more imagination space to the robot industry.

However, the entrepreneurship of AI robots can only be said to be a sea of stars, and it may not be possible to achieve it in ten or eight years, which Li Yanhong may have a deep understanding, so do other investors who want to "enter the pit" think clearly?