Photo: Visual China
Source: 21tech
Author: Yang Qingqing
Editor: Li Qingyu
If AI unicorns are accelerating their entry into the capital market in 2020, then in 2021, the listing journey of AI companies can be described as a lot.
At noon on December 13, SenseTime announced that the global offering and listing will be delayed, and it is expected to publish a supplementary prospectus. At the same time, SenseTime said in the announcement that the company is still committed to completing the global offering and listing as soon as possible, and all the application shares will be refunded to all applicants without interest.
Previously, SenseTime was originally expected to announce the final offer price, placement results and other information on December 16, and officially listed on the Hong Kong stock market on December 17, thus officially winning the title of "AI first share" in the first IPO.
SenseTime has suspended its listing, and the fuse is mainly "external factors". On the 10th, the US Treasury Department suddenly announced that SenseTime would be included in the investment blacklist. That was the time for SenseTime to price its listing on the Hong Kong Stock Exchange.
However, looking at the other "small partners" of SenseTime, this year's listing road is also quite a lot of twists and turns. In August and September this year, Yuncong Technology and Megvii Technology submitted their registrations on the Science and Technology Innovation Board, but as of now, the two companies have not been listed for a long time. Since the termination of the listing review of the Science and Technology Innovation Board at the end of June this year, YITU Technology has not taken any new actions.
In addition to the "FOUR LITTLE DRAGONS OF AI", the list of AI companies listed this year is still very long: Yunzhisheng, Yuntian Lifei, Geling Deep Pupil, Haitian Ruisheng, Hesai Technology, UBTECH... For AI unicorns, there is no listing dream in 2021.
Listing "closed door soup"
For the artificial intelligence industry, the obvious phenomenon in 2020 is that related unicorns are trying to accelerate their entry into the capital market.
Whether it is SenseTime, Megvii, Edu Technology or Yuncong Technology, the domestic "AI Four Little Dragons" listing news is continuous. At the same time, artificial intelligence companies, including Haitian AAC, Yunzhisheng, Yuntian Lifei and other artificial intelligence companies, have submitted IPO applications for the Science and Technology Innovation Board; companies such as SPI Chi and Fourth Paradigm have also disclosed their willingness or plans to list.
At that time, the AI unicorn "2021 listing" seemed to be a foregone conclusion, and the suspense only remained on who would sprint to become the "first AI stock". However, the situation is changing, and at the end of 2021, no AI company has successfully listed.
On the one hand, the listing encountered obstacles, on the other hand, looking at several AI companies that sprinted for IPOs, it is still difficult to escape the fate of huge losses. According to the data of various prospectuses, from 2018 to 2020, Megvii Technology has a cumulative loss of 12.77 billion yuan, a cumulative loss of 2.6 billion yuan from Cloud Technology, and a cumulative loss of 6.118 billion yuan in the first half of 2018 to 2020. From 2017 to the first half of 2020, Yunzhisheng's cumulative loss reached 833 million yuan.
SenseTime, which is at the head of the industry, has accumulated losses of 24.2 billion yuan from 2018 to the first half of 2021. SenseTime mentioned in the prospectus that since the AI industry is still in its infancy and requires sustained large-scale investment, it may not be able to achieve or maintain profitability in the future.
Behind the high losses is the high R&D investment of AI companies. According to SenseTime's prospectus, in 2018, 2019, 2020 and the first half of 2021, its R&D expenditure was 848 million yuan, 1.916 billion yuan, 2.453 billion yuan and 1.77 billion yuan respectively, and the cumulative R&D expenditure in three and a half years was nearly 7 billion yuan.
The time cost and capital cost required by artificial intelligence itself are relatively high, and considering the rapid pace of the industry and the high cost of hardware, many AI companies have a high proportion of research and development expenses in revenue.
According to the prospectus data, in the first half of 2021, senseTime's R&D expenses revenue ratio was as high as 107.3%. Other AI companies also maintained a high proportion of R&D investment, with Megvii Technology's R&D expenses accounting for 71.87% in 2020, Yuncong Technology accounting for 75.59%, and in the first half of 2020, the figure of Yitu Technology was 100.1%.
"Most Chinese intelligent enterprises are facing the problem of difficult to land research and development results, high research and development costs, and difficult to achieve profitability," the research team of the Leopard Research Institute analyzed, "although the gross profit margin of artificial intelligence enterprises remains within the higher range of 40% to 60%, how to achieve a breakthrough in profit zero is still the main contradiction of Chinese intelligent enterprises." ”
For the current situation of AI enterprise losses, Huang Wei, CEO of Yunzhisheng, previously said in an exclusive interview with the 21st Century Business Herald reporter that he hoped that the outside world would have more patience with the industry.
"We can't emphasize Sino-US technology competition on the one hand, and ignore the long-term investment of the United States in this field on the other hand," Huang Wei pointed out, "At present, none of the pure AI companies in the United States are profitable. I think it's a paradox. ”
In Huang Wei's view, for the current AI company, the word loss is reasonable. But this does not mean that enterprises can endlessly "burn money", Huang Wei believes, "it is indeed possible to pay attention to whether the loss of enterprises is narrowed, or whether the corresponding scientific research investment has generated corresponding commercial returns." ”
Explore technology landings
Although there are many twists and turns in the listing of AI companies, on the other hand, the artificial intelligence industry is still developing, and the exploration of industrial landing has never stopped.
According to the statistics of the China New Generation Artificial Intelligence Development Strategy Research Institute, as of 2020, there are 2205 artificial intelligence companies in China. In contrast, the number of artificial intelligence companies in China in 2019 was only 797. In terms of industrial scale, according to the calculation of the Digital Research Center of the Chinese Academy of Information and Communications Technology, the scale of the Chinese intelligent industry in 2020 will be 303.1 billion yuan, an increase of 15.1% year-on-year, and the growth rate will exceed the world.
According to the "China New Generation Artificial Intelligence Technology Industry Development Report (2021)", by the end of 2020, the layout of Chinese intelligent enterprises focuses on the application layer and technology layer. Among them, the number of artificial intelligence enterprises at the application layer accounted for the highest proportion, reaching 84.05%; Followed by the number of technology layer enterprises, accounting for 13.65%; The number of basic layer enterprises accounted for the lowest, at 2.30%.
This also means that the development of the Chinese intelligent technology industry is mainly driven by application demand. In the process of boosting the application of artificial intelligence technology, platform enterprises have played an important role.
Recently, Pengcheng Lab and Baidu jointly held a press conference to officially release the Pengcheng-Baidu Wenxin (model version number: ERNIE 3.0 Titan) jointly developed by the two parties. According to reports, the scale of the model parameters reached 260 billion, which is currently the world's largest Chinese monomer model, and the world's first 100 billion large model for knowledge enhancement.
In an interview with the media, including 21st Century Business Herald, Wu Tian, vice president of Baidu Group, said that there are two major difficulties in the application of AI technology in the industry. First of all, data annotation is very expensive, and even in specific scenarios can not complete large batches of annotation, on the other hand, in the industry application, the industry's own knowledge accumulation is crucial. The introduction of pre-trained large models can ensure the technical foundation without a large number of data annotations, and the superimposed industry "basic skills" can achieve the rapid landing of AI technology.
iFLYTEK also announced in October this year that it will upgrade the AI open platform to version 2.0, and the first batch of three-party collaborative sharing and co-creation will be realized in 18 industries, including education, agriculture, electricity, finance, tourism, transportation, medical care, recruitment, etc. In the future, iFLYTEK will also invest 500 million yuan in R&D funds and 1 million developers every year to carry out technology empowerment and cooperation and co-creation, and comprehensively tap the value application of the artificial intelligence industry.
"In the past, we were more about providing individual AI technologies to developers and training them so that they could apply the technology to more applications," wu Xiaoru, president of iFLYTEK, explained in an interview with the media, including the 21st Century Business Herald, "With the advancement of industrial digitalization, many traditional industries also need artificial intelligence technology, but the problems that need to be solved in various industries are more complex, and the experience and knowledge of industry experts need to be precipitated for a long time. So we upgraded the AI open platform to 2.0. ”
If this year's industry theme is an unexpected storm that has painted more different colors on the overly sunny "expectations" of many companies, looking forward to the upcoming 2022, I believe that the artificial intelligence industry is more stable and far-reaching after experiencing the baptism of wind and rain, and the trend of China's AI companies leading the world will not change.