"Science and Technology Innovation Board Daily" (reporter Ao Jin) news, recently, cambrian CTO Liang Jun's departure, triggered the outside world's attention and concern about the current development of the domestic AI industry.
In the capital market, as soon as the announcement of the departure of the core technology member came out, the Cambrian stock price opened the next day and plunged, with a one-day decline of 18.38%. As of press time, cambrian stock price was reported at 65.22 yuan / share, compared with the grand scene when it first landed on the science and technology innovation board, it once rose to 300 yuan / share, and the market value has shrunk significantly.
In fact, not only the AI chip industry involving the basic layer of the artificial intelligence technology system has been frequently questioned in recent times, but the entire domestic AI field has been re-examined by the capital market and ordinary investors in the news of the bloody IPO of major manufacturers.
Dr. Zhang Jun, chairman of CEIBS Capital and former vice president of Huawei, said in an interview with the Science and Technology Innovation Board Daily that whether it is the drastic adjustment of the company's internal team or the poor performance of the stock price in the secondary market, these are not the unique status quo of a single company in the AI field, but the problems faced by the entire AI field. He judged that in the future, a large number of AI companies will fall in huge losses because they do not have real core technologies and large-scale application scenarios.
During the interview, he repeatedly stressed that AI companies must do three things to survive the next life-and-death period: "The first is to calm down and not to be carried away by the advocacy of capital; then to continue independent research and development and find their own technical control points; the third is to find commercialization scenarios other than security that can be massive." ”
At the same time, Zhang Jun believes that capital, as one of the important external influencing factors for the development of the AI industry, should return to calm and "stop the hype game of drumming and passing flowers." ”
AI companies lack core technologies and high-volume application scenarios
Cambrian's halo of "the first share of AI chips" was listed on the Science and Technology Innovation Board on July 20, 2020. On the same day, the Cambrian opening price reached 250 yuan / share, nearly 300% higher than the issue price of 64.39 yuan / share, and the market value exceeded 100 billion yuan. But since then, Cambrian has opened high and gone low, and the stock price has gradually declined, and now the stock price is almost the same as the issue price.
At the level of the company's operation, Cambrian has still not achieved profitability, and since its establishment in 2016, the company's net profit attributable to the company has continued to be in a state of loss. According to the performance express, the total operating income of Cambrian in 2021 was 721 million yuan, an increase of 57.12% year-on-year. But at the same time, the company's losses are also expanding, and in 2021, the company's year-on-year losses have expanded by 94.98%. As for the reasons for the loss, Cambrian gave the explanation that it was due to the increase in share payments and sales expenses in research and development expenses and management expenses.
According to the Southwest Securities Research Report, from the perspective of the artificial intelligence industry chain, Cambrian is at the basic level of the industrial chain structure, and is committed to building various intelligent cloud servers, intelligent terminals and intelligent robot core processor chips. In the basic layer of AI chip segmentation, the mainstream artificial intelligence chip is divided into GPU, FPGA and ASIC three, the current GPU is in the dominant position of artificial intelligence chips, FPGA and ASIC chips are developing rapidly. Cambrian AI chips belong to ASIC chips, and in the research and development of this part of the chip, the technical gap between domestic companies and foreign giants is small.
Zhang Jun said in an interview that from a technical point of view, in general, the current technical threshold of domestic AI chips is low. "It's more of an image recognition or image processing chip that lacks real technical content."
He further said that the current AI chip can even be said to be a pseudo-concept or gimmick. "AI is a technology, there can be many application scenarios, can not say that the application of AI algorithms of the chip is 'AI chip'." The classification of chips can be classified from materials, structures, and application scenarios. ”
Wei Shaojun, director of the Institute of Microelectronics at Tsinghua University and vice chairman of the China Semiconductor Industry Association, has also made a similar statement, "The current AI chip is not really AI." He believes that the real AI chip should make a breakthrough from the architectural aspect, and the current popular architecture of the market manufacturers includes CPU + SW, CPU + GPU, CPU + FPGA, etc., "but these are not the ideal architecture for AI." ”
Dr. Zhang Jun further said that in fact, the current domestic artificial intelligence industry as a whole lacks real independent technological innovation. "Many companies in the field of AI, whether it is the basic level, or the technical level and the application level, are closely focused on the security field, but the technology involved in artificial intelligence for security is relatively low-end, and it is already a red sea in the market, and companies can only spell out competitive advantages by fighting price wars, so the result is that the company's gross profit is very low, on the one hand, it is low gross profit, on the other hand, it is high research and development costs, and higher technical personnel salaries, which leads to AI companies can not make ends meet, continuous losses."
Zhang Jun said that as early as his undergraduate studies in 1981, image recognition technology related to security has made certain developments. "At that time, there was no special artificial intelligence major in the university, and the major I read was called pattern recognition and intelligent control, and the foothold of this technology was actually image recognition, and the recognition involving three dimensions was computer vision." Jun Zhang is a postdoctoral fellow at Shanghai Jiao Tong University, where his undergraduate and postdoctoral careers include automatic control, mechanical engineering, artificial intelligence, and communications and information systems.
The lack of independent research and development technology, and the lack of application scenarios other than security and feasibility on a large scale, are the two major problems facing the domestic AI field in Zhang Jun's view.
Capital speculation spawned an industry bubble
After retiring from Huawei, Zhang Jun founded CEIBS Capital in 2015 and currently serves as chairman of CEIBS Capital.
According to the data, CEIBS Capital focuses on strategic investment in hard technology fields such as artificial intelligence, semiconductors/integrated circuits, Internet of Things, and 5G. Correspondingly, CEIBS Capital's partners are more industry experts with technical backgrounds. It is said that at present, CEIBS Capital has 18 industrial technology expert partners, who have work experience in The World's Top 500 companies such as Huawei, Qualcomm, Intel, and Foxconn. CeIBS Capital and partners have invested in companies including chip design company Suzhou Naxinwei, Shenzhen Hongyuan Electric Appliances and participating in C*Core Technology.
After joining the venture capital industry, Zhang Jun looked at the AI industry with an extra layer of capital perspective. He believes that from the current situation, capital has fueled the waves, which has actually spawned a bubble in the domestic AI field. "Many technologies or concepts in the field of AI can catch fire, largely because some speculators feel that artificial intelligence gimmicks are big, so they hype up." The hype of large funds has led to impetuousness and blind expansion in the industry. ”
According to Southwest Securities Statistics, during the period from 2014 to 2018, start-ups in the AI field sprung up like mushrooms, and investment and financing events reached a historical peak in 2018, reaching 400 times. Since 2020, the average single financing amount of AI companies has jumped from about 100 million yuan to 330 million yuan, and the single financing amount of many enterprises has exceeded 3 billion yuan.
Zhang Jun said bluntly that the phenomenon of large institutions raising the valuation of AI companies in the primary market is already an open secret in the industry. "In the early round of corporate financing, some investors and even investment institutions, joint enterprises, after doing a round of financing, immediately opened the second round, and advocated that the valuation of the second round of companies has risen many times, and the investment window is small, creating an illusion that investment is indispensable." The final result is that the institutions that make the game cash out early, and the institutions that join later, including the shareholders who participated in the investment after the company was listed on the secondary market, became the people who bore the losses in this drumming game. The inningsman makes a lot of money, and he floods the earth. ”
Zhang Jun further said that the reason why this phenomenon of "making the game" can occur in the domestic primary market has a lot to do with the professional background of most investors in the current industry. "There are many VC/PE investors in Silicon Valley who are industry experts who have been working in related technology fields for many years and have gone through several investment cycles. However, 99% of China's current VC/PE partners and decision makers are actually from finance, investment banks, brokers, accountants and even lawyer backgrounds, and the proportion of young people is also relatively large, so they may not understand the value of some technology industry companies, more just look at financial statements, but the company's statements in the early and medium terms do not have much to see, which is easy for investors to follow the trend, chase hot spots, and look forward to the head institution. ”
Capital continues to increase investment to blow up the valuation bubble, and ai enterprises that actually lack technical control points and application scenarios at the operational level have been pushed to the cusp of the storm by capital, "so we will see that after many AI companies are listed, there is an inversion of the valuation of the primary and secondary markets." Zhang Jun believes that there is already a big bubble in the entire AI industry, "The next bubble bursts the company's market value falls, then the worst result is to scatter the birds and beasts." ”
At present, there are already many AI companies in the industry who have left the technical team, "The technical giants in the industry, of course, hope that they can make a difference on a big platform, but when a ship has begun to falter, it is also common sense to jump ashore." ”
The reporter of "Science and Technology Innovation Board Daily" noted that in addition to the departure of Liang Jun, a core core technology member of Cambrian, in 2019, Yuntian Lifei also experienced the resignation of co-founder Tian Dihong as CTO of the company.
The Reply Opinions of Issuers and Sponsoring Institutions issued before the listing of Yuntian Lifei in 2021 disclosed that "with the development of the company, the two (referring to The founders of Yuntian Lifei, Chen Ning and Tian Dihong), gradually had major differences in the future development direction of the company, and Tian Dihong decided to stop working with the issuer in September 2018." On September 4, 2018, the board of directors of Yuntian Limited deliberated and passed the proposal for Tian Dihong to resign as CTO of the company. ”
The AI industry ushered in a major reshuffle
Zhang Jun concluded that the speculation of the concept of capital creation, the emergence of bubbles, coupled with the lack of technical control points and application scenarios, several major factors combined, resulting in the current domestic AI companies are difficult to ride the dilemma.
To solve these problems, he believes, "First of all, AI companies should calm their minds, do not blindly expand because of the flickering of capital and short-term profits, continue independent research and development with a peaceful attitude, find their own control points, and application scenarios other than security that can be scaled up." He further said that autonomous driving may be one of the feasible application scenarios. But the relevant companies developing in this direction, how far to go remains to be seen, there are too many variables involved, and the technology is constantly exploring and developing. ”
At the same time, he believes that in this process, capital should also play a positive role in guiding and promoting, "capital, as an investment institution, can no longer continue to maliciously speculate." Zhang Jun pays attention to the hard technology companies in the artificial intelligence and 5G track, and he said that when evaluating investment targets, he attaches the most importance to the personality of entrepreneurs and the company's core technology.
"Character first, martial arts second." Investing in a company often occurs in this situation, A round of a company may do a product, B round may be changed to another product, but as long as the entrepreneur is determined and reliable, no matter which product, it may eventually stand out. ”
In terms of core technology, "if there is no threshold at all, it means that anyone can do it, such as sharing bicycles, community group buying, this single model innovation is not enough, but also needs to have a strong ability to innovate in technology; finally, there is also a strong ability to expand the market, just like the two ends of a dumbbell." He took Hua as an example, "Huawei's strong place lies in the dual grasp of sales and technology." The outside world only sees that Huawei is strong in technology, but in fact, people in Huawei's product line run around the world most of the year, interviewing various government people and corporate CEOs, essentially for market sales, and finally made top sales. ”
At present, the leading enterprises in the FIELD of AI have begun to explore new application scenarios. After the sale of the previous core medical business, Yitu aimed at the new scenario of autonomous driving; the cloud was positioned as a human-machine collaborative solution provider, and in the process of continuous deep cultivation in the financial field, it also added investment in robot and Internet of Things technology research to try to build a more standardized AI product; Megvii cut into AIoT and regarded the logistics business as the future growth point; SenseTime proposed the "1+1+X" platform strategy, of which 1 represents research and development and technology industrialization, and X represents the empowerment of hundreds of industries Horizon is making efforts in the field of autonomous driving.
Zhang Jun estimates that in the next period of time, there may be a group of enterprises in the AI industry, because they cannot make breakthroughs in technology and sales and fall on the road of entrepreneurship, "the great reshuffle is coming, and living is the first pursuit and the highest pursuit." But the lessons of the past, the teacher of the future, there will certainly be a batch of enterprises born in the future. Especially in the context of today's world situation, 'domestic substitution' has actually encountered a once-in-a-lifetime development opportunity."
![The general melted the silver into a large ball, and the thief could only see and take it away, and the general was overjoyed, and found the plan[fig]](data:image/gif;base64,R0lGODlhAQABAIAAAP///wAAACwAAAAAAQABAAACAkQBADs=)