In November 2017, Yinqi, founder of Kuangshi technology, said that in the next two years, companies with pure artificial intelligence will be listed, "I hope we will be the first". The previous month, Kuangshi technology completed round C financing, with a financing amount of US $460 million, breaking the highest single financing record in the field of artificial intelligence in the world. The primary market showed extraordinary enthusiasm for artificial intelligence. Since May of that year, Yitu technology and Shangtang technology have successively completed large-scale financing. When the "x-based" face recognition algorithm was first introduced by Apple's "vision based" mobile phone, it was believed that the "x-based" face recognition algorithm was introduced into the future. At that time, artificial intelligence was popular in the market, just like the concept of meta universe not long ago. However, it has been two years since Kuangshi technology really applied for Hong Kong stock IPO, during which Yinqi was constantly asked about the listing. Since then, Yitu technology and Yuncong technology have successively impacted the science and innovation board. However, Shangtang technology, which was the latest to apply for listing, took the lead. At the end of 2021, Shangtang technology was listed in Hong Kong stock market. After only four days of listing, its share price more than doubled, with a market value of more than 272 billion Hong Kong dollars. Since then, the share price began to decline. As of the latest closing day, the market value of Shangtang technology has shrunk by more than HK $100 billion compared with the highest point. This is very much like the trend of AI startups, opening high and going low. In short, the secret war of AI four dragons on the "first share" has come to an end, and it seems that there is an answer to who first listed on the scientific innovation board. On April 6, the CSRC agreed to apply for registration from cloud technology. Pressure comes to both sides of Yitu and Kuang Shi. Is listing so important? Very important. Since 2020, among the four AI dragons, only Yitu technology has made one financing. You know, they have a soft hand in financing in the hot period of AI venture capital. But more important than listing is when to "stop bleeding"? The "Ai four dragons" were exposed in the process of impacting the IPO, and the amount of losses can be called a "miserable conference": does AI really have a "money path"? This is the self redemption that the AI four dragons must complete. 01 AI tide rise and fall For the AI four dragons, 2011 is a very important time point. In October this year, Yinqi, who graduated from the "Yao class" of Tsinghua University, together with Tang Wenbin and Yang Mu, founded Kuangshi technology in Beijing. In the same year, Zhu long, who studied at New York University in the United States, had the idea of returning home to start a business. The next year, he founded Yitu technology with his classmate Lin Chenxi. Shangtang technology was established relatively late, but it has accumulated quite deep in the field of technical research. From 2011 to 2013, in the field of computer vision, there were a total of 29 articles on deep learning in the world, of which 14 were from Tang Xiaoou's team. In 2014, Shangtang technology was established. Also in 2011, Zhou Xi, an alumnus of Tang Xiaoou's University of science and technology and founder of Yuncong technology, returned from the United States and entered the Chongqing Branch of the Chinese Academy of Sciences. Four years later, he founded Yuncong technology. So far, the "Ai four dragons" have gathered at the track of computer vision, that is, the popular understanding of "face recognition". Computer vision was also one of the most popular AI tracks at that time, and the players were far more than the "four dragons". However, an insider from the AI four dragons once told leopard transformation that he did not recognize the so-called "Ai four dragons" and believed that it was a forced binding of some companies to attract capital and public attention. At first, the "four little dragons" were not well known to the outside world, and AI was a relatively unfamiliar concept. Until the 2016 go world champion Li Shishi lost the "man-machine war" with alphago, China set off an AI boom overnight. Tang Xiaoou said bluntly, "after this game of chess, artificial intelligence doesn't need our explanation. Everyone suddenly understands that artificial intelligence is like this." Capital then began to embrace AI crazily. "After alphago, we started financing in March. We can add $100 million in April (valuation), another $100 million in May and another $100 million in June. I didn't do anything, so I can." Zhu long described the pursuit of AI by capital at that time. According to the statistics of the blue book on the development of artificial intelligence industry in 2021, there have been 2048 equity investments in the field of artificial intelligence before IPO in the past nine years, with an investment amount of 480 billion yuan, and the peak of investment is from 2015 to 2018. It can also be seen from the financing process of the AI four dragons that since 2016, the overall financing has occurred frequently and the amount is huge. Among them, Shangtang technology is highly "favored" by capital, and the cumulative financing scale exceeds the sum of the other three. The valuation has also gone up. From the public data, the four companies have carried out more than 30 rounds of financing, and the pre IPO valuation has reached about 140 billion. Shangtang technology, known as the "financing machine", conducted 12 rounds of financing with a total amount of US $5.2 billion before listing. The last round was valued at US $13 billion, nearly 60 times higher than the valuation in 2014. After completing the financing of US $750 million in May 2019, Kuangshi technology has no action, and the company's valuation is about US $4090 million. Cloud technology completed RMB 1.8 billion financing in May 2020. According to the data disclosed in the prospectus, the company's valuation after this round of financing is about RMB 13.6 billion. Yitu technology carried out the 11th round of financing in 2022, with a valuation of more than 20 billion yuan. No tuyere never dies. In fact, after 2019, the enthusiasm of capital for AI has obviously subsided. Southwest Securities research report pointed out that the number of investment and financing events in the AI field peaked in 2018, and the number of investment and financing has decreased significantly since 2019. Among them, the proportion of investment and financing after round C is more than 50%, and the round moves back. After 2020, among the four AI dragons, only Yitu technology has completed one financing. A middle-level person from the AI four dragons told leopard change that "(the money in the primary market) can't be melted". In 2020, Li Kaifu talked about AI startups on a public occasion and said that the era when three AI experts can value 700 million and deceive investors by AI concept has passed. An investor who pays attention to AI said that he has no AI investment theme now. He analyzed to leopard transformation, "AI should have become the standard configuration of most technology companies. Just like a person, you will ask him whether he can drive or swim. This is a skill, but you won't ask him whether he can breathe and drink water. AI will be such a default ability in the future." There is also a view in the market that the pressure of the United States is also an objective factor affecting capital to continue to invest in the AI four dragons. Since October 2019, the four AI dragons have successively entered the "entity list" of the US Department of Commerce. Financing in the primary market is difficult. Even with the star aura overhead, the listing expectations of the four dragons have been delayed. In August 2019, Kuangshi technology took the lead in launching the first shot of the "Ai four dragons" IPO, but after failing to impact the Hong Kong stock exchange, it moved to the science and innovation board in September the next year. In March 2021, Kuangshi technology officially submitted its listing application. One year later, it is still waiting. Almost at the same time as Kuangshi technology, Yuncong technology and Yitu technology have successively started the listing guidance of science and innovation board. The latest result is that Yitu technology withdrew its listing application at the end of June 2021. After suspension and multiple rounds of inquiries, cloud technology was finally approved for registration in April this year. On the way of the "Ai first share" dispute, Shangtang technology, which took the latest action, was the first to be listed. Since 2017, the news about Shangtang technology's IPO has been rising one after another, and it is not considered "solid" until August 2021. Compared with the other three companies, the listing process of Shangtang technology is relatively smooth, and it takes only four months from application to listing. The secondary market is also cruel. Shangtang technology has only been listed for four days, and its share price has more than doubled, with a market value of more than 272 billion Hong Kong dollars. Since then, the share price began to decline. As of the latest closing day, the market value of Shangtang technology has shrunk by more than HK $100 billion compared with the highest point. The market value of the newly listed Geling Shentong has suffered a serious decline. Investors once judged that its valuation could reach $300 billion. It broke on the first day of listing. The latest closing price fell 40% compared with the issue price, leaving only 4.2 billion yuan in market value. There is also the Cambrian "Ai chip first stock" that landed on the science and Innovation Board earlier. Its market value once hit 100 billion yuan on the first day of listing, and now it has shrunk by 80%. A similar situation also happened to hongruan technology mentioned in the prospectus of AI four dragons. Its market value was once as high as 42 billion yuan, and now there is only 9.2 billion yuan left. AI companies encounter investors' cruel voting with their feet in the secondary market, which undoubtedly casts a shadow on the future performance of Kuangshi, Yitu and cloud on the science and innovation board. 02 why doesn't capital "Ai"? There are many reasons for the cooling of AI investment boom and the difficulty of listing AI companies. In December 2020, an algorithm engineer from the AI four dragons told leopard transformation that the capital is indeed "calm" and "if you have a practical grasp and understanding of the so-called AI technology, you know that the expectations of the management or Party A are obviously too high." The engineer said that computer vision is similar to steam locomotive. It is an epoch-making technology, but we can't expect it to run 10000 times faster than carriage, maybe only 10 times faster, which is only 0.1% of the expected value, but it is actually a qualitative leap. According to the analysis of Guojin securities, the investors were too optimistic about the return cycle of the artificial intelligence industry in the early stage, and the market had doubts about the current business realization mode of entrepreneurial AI companies, which were the main reasons for the cold in the capital market in the past two years. After the disclosure of the prospectus by Shangtang technology, 24.3 billion yuan was posted on the microblog hot search in three and a half years (from 2018 to the first half of 2021). Previously, the loss data of Kuangshi, Yuncong and Yitu have triggered a great discussion on "whether AI is a good business". According to the prospectus, from 2018 to the first half of 2021, the cumulative net loss of Kuangshi technology to its parent was 14.6 billion yuan; From 2017 to the first half of 2020, the cumulative net loss attributable to parent company of Yitu technology was 7.3 billion yuan; From 2018 to 2020, the cumulative net loss of cloud from technology to parent was 2.2 billion yuan. Why do the AI four dragons lose more than one? First of all, it should be noted that the huge losses of the AI four dragons include the profits and losses from changes in the fair value of preferred shares and share-based compensation expenses. After excluding the above factors, the net losses of Shangtang, Kuangshi, Yitu and Yuncong were 3.6 billion yuan, 4.1 billion yuan, 2.3 billion yuan and 1.6 billion yuan respectively. Although the net loss has narrowed significantly compared with that before adjustment, the AI four dragons are still at a loss because the expense rate of the four companies is very high, especially R & D expenses. Some companies even have R & D investment exceeding revenue. According to the prospectus and performance report, the proportion of R & D expenses in revenue of Shangtang technology from 2019 to 2021 was 63.3%, 71.2% and 76.9% respectively, while that of Yuncong technology was 56.3%, 76.6% and 49.6% respectively in the same period. From 2018 to 2020, R & D expenses accounted for 70%, 74% and 64.4% of revenue respectively. The revenue of Yitu technology is relatively small, but the proportion of R & D expenses is the highest, reaching 146.4%, 95.7% and 91.6% respectively from 2017 to 2019. How high is this proportion? Taking the proportion of R & D expenses in the operating revenue of the science and innovation board in 2020 as an example, wind data shows that among the 413 science and Innovation Board companies, 96% of the companies' R & D accounts for less than 50%, and 75% of the companies' R & D accounts for less than 15%. In addition to R & D expenses, the artificial intelligence ecosystem needs to extend from infrastructure and technology platform to the landing scenario. How to sell and the sales expense rate behind it are also a huge test for the company. At present, the proportion of sales fee of Shangtang technology in revenue has been stable at about 15% in recent three years, which is the lowest among the four little dragons of AI. The proportion of sales expenses of Yitu technology is still the highest, from 2017 to 2017