Image source @Visual China Article 丨 Intelligent Relativity (ID: aixdlun), author 丨 Yixi On May 17, the “Guangdong Province Social Credit Regulations” was issued, which clearly prohibits the collection of natural person information such as diseases and biometrics. Earlier, Tianjin, Hangzhou, Nanjing and many other places began to legislate to strictly control facial recognition, and even required the removal of facial recognition equipment such as sales offices and properties. Since the second half of 2020, protests about the mandatory use of face recognition have become increasingly louder. This year’s CCTV March 15 party broke out that 2,353 companies across the country, including Kohler, BMW, MINISO, etc., installed millions of face recognition cameras, suspected of collecting hundreds of millions of users’ facial information without the user’s knowledge. A wave of public opinion crisis involving personal privacy issues. Artificial intelligence image recognition scenes including face recognition are the main track of the AI Four Little Dragons. As early as 2014, Megvii successfully seized the opportunity of Alipay to find a technology provider for “face payment” and found the first commercial landing scene. Coincidentally, Yitu, Yuncong, and SenseTime also started from “swiping their faces” and opened up the market in the fields of security and finance. Because of their outstanding performance, they were collectively referred to as the “AI Four Dragons” in the domestic artificial intelligence field. Nowadays, the business scene of face recognition is facing the risk of rectification. In addition to the increasing resistance of policies and public opinion, the AI Four Little Dragons, at the same time, are facing problems such as long-term huge losses, exodus of executives, and repeated obstacles to listing. It was also exposed to the spotlight. On May 20th, the four AI dragons are so big that they will even be on the hot list. It is not difficult to find that the four little dragons that have been surrounded by capital have fallen into the altar in the past few years, and the life of AI companies is actually not as good as before. Today, the AI industry has entered the deep water zone, and how to find a more suitable path forward in the future has become the focus of the entire industry. Capital goes backward, market goes forward Since the beginning of this year, the unicorns that once vie for the “first share of artificial intelligence” have smashed their IPOs. On March 11, Yitu Technology suddenly applied for suspension of IPO. Just one day later, Megvii Technology was exposed after the Hong Kong stock IPO stranded and moved to the Science and Technology Innovation Board. Recently, Yuncong Technology’s IPO was also suspended due to “the financial information recorded in the issuance and listing application documents has passed its validity period”. SenseTime has not disclosed its listing plan, and companies such as Yunzhisheng and Hesai Technology have also terminated their listings. In the past few years, there was often a saying: The next decade is the decade of AI. In the eyes of a large number of investors, artificial intelligence is the fourth productivity revolution after steam engines, internal combustion engines and the Internet. At that time, the financing of the AI track was like a competitive game, and the financing amount was refreshed over and over again. Public data shows that in 2017, the total number of domestic AI companies reached more than 600, and the financing scale of Chinese AI companies reached a peak of 148.5 billion yuan in 2018. According to the “Intelligent Theory of Relativity”, Yitu Technology has undergone 10 financings from September 2012 to June 2020. Strategic investors include Sequoia Capital China, Hillhouse Capital, etc.; SenseTime Technology November 2014 As of September 2018, it has experienced 9 financings. Investors include SoftBank Vision Fund, Hopu Investment, Silver Lake Capital, Alibaba, Temasek Temasek, IDG Capital, etc.; Yuncong Technology from April 2015 to May 2020 A total of 10 rounds of financing have been experienced. Investors include China Net Investment, Shanghai Guosheng Group, Nansha Financial Holdings, Industrial and Commercial Bank of China (5.200, -0.04, -0.76%), ZhongAn Capital, Shunwei Capital, etc.; Megvii Technology has started from 2011 From November to May 2019, there have been 7 financings. Investors include ICBC Asset Management Co., Ltd., Abu Dhabi Investment Authority, Bank of China Investment, Macquarie Capital, Alibaba, Ant Financial, Lenovo Ventures, etc. However, under high valuation and high investment, the four little dragons are far from profitable. According to the prospectus, in 2017, 2018, and the first half of 2019, the amount of cash flow from Megvii Technology’s operating activities was -150 million yuan, -720 million yuan and -680 million yuan, respectively. The prospectus disclosed by Yitu also showed a huge loss. The net loss in 2019 was 3.647 billion yuan, and the net profit in the first half of 2010 was still a loss of 1.303 billion yuan. Yuncong’s prospectus disclosed that the adjusted net loss in 2017, 2018, and 2019 were net losses of -106.3163 million yuan, -180.6752 million yuan, and -1708.0161 million yuan, respectively. In the first half of 2020, the net loss reached -286.2022 million yuan. In the financing plan, SenseTime’s debt ratio in 2019 is also as high as 102.35%. “Intelligent Relativity” sees, The slow progress of commercialization and the amazing ability to burn money has made the business models and liquidity capabilities of artificial intelligence companies such as the Four Dragons become the focus of market doubts. The long-term investment return period and high investment in the artificial intelligence industry have also been widely known. Since 2020, the number and scale of financing events have dropped significantly. According to CB Insights, the overall number of AI financing transactions in the first quarter of 2020 dropped from 542 in the previous quarter to 506, while the number of pending transactions in the third quarter of 2019 was 660. Yitu Technology recently disclosed that the financing amount was 30 million U.S. dollars invested by the Runcheng Industrial Pilot Fund in March 2020, which is less than one-sixth of the previous highest amount in the C++ round. Even SenseTime, which is known as a financing machine, has not had new capital to settle in for two years. The wave of artificial intelligence has gradually faded. It is becoming more and more difficult for AI companies to raise funds in the primary market, and after several rounds of large-scale financing, the primary market has been unable to afford a company of this size. In order to raise funds to promote business development, the four dragons have successively embarked on IPOs. road. In addition, besides capital, the market is also evolving . In the last wave of capital dividend period, AI application scenarios are not yet mature and have not been verified by the market too much. At that time, AI companies were in the stage of basic technology accumulation, and the core driving factors were teams and talents. Nowadays, the development of the AI industry has entered a new stage. Only simple technical business can not bring enough room for imagination. “This year, we really have rarely looked at pure AI projects. We are more concerned about the ability of AI to integrate with industries such as finance, logistics, and medical care, how many compound talents the company has, and how deep the understanding of the industry is.” “In addition to focusing on technology and products, investors also value the competitiveness of the entire business environment, including operations, customer relations, and manufacturing. In the development of the AI industry, scenarios are more important than technology.” A long-term investment in hard technology The man said to the reporter. Capital goes backward, market goes forward . The AI industry has long been not limited to the research and development and implementation of the algorithm level, but has developed in many aspects such as chips, industry solutions, and ecology. In the process of technology upgrades, model upgrades, and method upgrades, a number of new AI companies will surely emerge, and of course, a number of companies will gradually become obsolete. From +AI to AI+, the industry logic has completely changed After five or six years of development, AI has now been applied in many vertical fields, with more mature fields including home furnishing, finance, transportation, and medical care. By combining with many vertical fields, artificial intelligence technology can empower the industry through two aspects: on the one hand, increase production efficiency, reduce costs and increase efficiency, that is, “+AI”; on the other hand, create new demand and growth points, that is, “AI+ “. In terms of “+AI”, whether the AI subdivision track can be commercialized, similar to all 2B industries, depends on how much industrial efficiency is improved, whether it brings increased revenue, or how many pain points AI technology solves pain. In terms of commercialization, there is a very straightforward logic for deciding whether to pay: Does this AI product help it increase productivity and effectively solve a certain problem? How much cost has been saved? How much income has been improved? Therefore, unlike in the past, using a single point of technology as the direction of force, seizing a certain scenario and continuing to burn money to obtain, process, and train data is different. AI companies need to provide customers with more and more precise services, which requires the team to have The ability to combine cutting-edge technology and business logic meets the needs of companies such as reducing costs and increasing efficiency to the greatest extent. In addition, In today’s fierce homogeneity competition, companies are also required to have excellent B-end channel operation capabilities. “The current trend is to make integrated and large-scale solutions. AI is only one of the capabilities integrated.” An industry insider introduced to reporters. At present, the growth of AI companies’ valuations driven by technology alone is very limited. In the final analysis, the transformation that AI companies face is from a technical service provider to an overall solution provider. The core driving factors are gradually changing from team and algorithm technology to data and scenarios. For AI companies whose entrepreneurial teams are mainly scientists In other words, the ability to acquire customers, service capabilities, and bargaining power are all new challenges. In addition, in the view of Ma Yunlong, an associate researcher at Tongji University, AI China and Taiwan will be a major trend in the future. He believes: “AI Zhongtai integrates a variety of complex algorithms in the digital transformation of cities, improves the adaptability of the product, and is closer to the needs of users and the market.” According to “Intelligent Relativity”, AI Zhongtai is the strong point of companies such as Shangtang, Yitu, Baidu, and Alibaba. Companies do not need to have too high AI development capabilities, they can quickly empower these AI capabilities in their own enterprise systems on the basis of AI mid-stage. In terms of AI+, Baidu and Taobao have used AI to do intelligent applications around their search business. HKUST iFLYTEK is the dominant player in the voice field, and other Internet giants also use social interaction, e-commerce, online car-hailing, and information recommendation. And other business scenarios let AI technology land. For a long time, the amount of data is large and small, and it is considered to be the technical barrier that constitutes the core of AI companies, which means that Internet giants with digital capabilities will have more advantages. In addition to their own advantages in data and scenarios, digital giants in related fields will achieve breakthroughs at a lower cost, and their applications will be more refined and smooth. In the landing stage, how to enter the industry to compete with industry players and win market share is a huge challenge facing AI companies. At present, the commercialization of AI is highly concentrated. According to the “China Artificial Intelligence Industry Research Report 2020” released by iiMedia, security and finance are the top two areas with the largest market share of AI empowered real economy, accounting for 53.8% and 15.8% respectively, totaling nearly 70%. Taking Megvii Technology as an example, the prospectus shows that the urban Internet of Things business, which focuses on security, contributed 73.2% of Megvii’s revenue. However, both areas are facing fierce competition. In the field of security business, compared with the old brands such as Dahua, Yushi, Hikvision, etc., the four little dragons are hardly competitive in terms of brand, channel, and bargaining space. In addition to price factors in the security field, hardware capabilities are more important. For example, camera sensors are involved in the upstream and downstream of the industry chain, and it is very difficult to integrate well. The AI four dragons can only cut in from software capabilities, and software is difficult to make a profit. It is understood that Hikvision has also built its own algorithm team, with densely packed cameras on the road outside the Hikvision Park, collecting test data by itself. In the financial field, the “AI Four Little Dragons” were eaten by giants. For example, Ant Financial released the “Dragonfly”, which is a facial payment hardware, and WeChat Pay followed by the similar product “Frog”. The algorithms are all from Ali and Tencent’s own algorithm teams. . It is worth noting that although the Four Little Dragons also have a layout in terms of hardware, the competition they face in the security and financial fields is still mainly competition in visual image-related businesses. The previous four little dragons have always kept AI services at a level that is relatively easy to implement, such as voice, image recognition, etc. . However, after two years of development, due to factors such as algorithm bottlenecks, homogenization, difficulty in landing scenarios, and policies, the computer vision field lacks stamina. As early as 2019, after an AI technology giant resigned from an AI startup company, he said: “A few years ago, a set of facial recognition algorithms sold for tens of millions, but now it’s only worth 400,000 yuan.” After the giant’s exit, the image The threshold of recognition algorithms is getting lower and lower, and face recognition algorithms are not worth anymore. “At present, except for some security vision companies with G-end resources in China, vision companies are basically unprofitable.” Chen Pengren, investment director of Qingtong Capital, said in an interview with the media before. At present, the fruits of the AI shallow water area have been picked, and facing a larger deep water area, AI companies are actively looking for more commercial scenarios, tapping deeper industry needs, and achieving self-made blood through differentiated development paths. Just as Megvii CEO Yin Qi said, after an early outbreak of five or six years, the AI industry has now entered a “deep water zone”, and AI companies that do not bring real value will be eliminated. In the deep-water area where the AI industry has landed, whoever can first realize the closed loop of value and the industry landed will have the right to survive, and the market left them with little time left. How will the unicorns evolve? With changes in the market environment and changes in models and strategies, the “AI Four Dragons” are changing from a narrowly defined face recognition company to a more comprehensive company. “Intelligent Relativity” understands that Megvii cuts into the robot AIoT field from a single point of computer vision, and regards logistics business as a future growth point. In terms of specific business, it is divided into personal Internet of things, urban Internet of things and supply chain Internet of things. For example, three hardware systems of cameras, edge servers, and cloud servers have been established for the urban Internet of Things Megvii, and more than 30 cameras have been introduced. In the supply chain Internet of Things, a variety of automation equipment has been developed to perform logistics and warehousing tasks. Last year, Megvii released 7 AI+ intelligent hardware products suitable for logistics and warehousing, and announced the launch of Hetu 2.0, a software intelligent robot operating system. It is understood that Megvii’s global flexible clothing smart warehouse, which is under construction, can connect and dispatch nearly 4,000 smart logistics equipment in 10 categories, relying on the Hetu system. Family Portrait of Megvii Robot Products Similar to Megvii, Yitu Technology has also strengthened the integration of software and hardware. Since 2020, computing power has replaced the original visual positioning and transformed algorithm + chip manufacturers. Last year, Yitu Technology released the search chip, and subsequently released the rough series of servers and cutting-edge edge computing devices based on the search chip. In its prospectus, Yitu Technology not only mentioned its competitors Nvidia and Cambrian but also Said that a large part of the funds raised in the listing is also used for the research and development of chip related products. It is understood that the current computing power support of computer vision companies mainly depends on Nvidia’s high-performance GPU series chips, and AI company’s self-developed chips can not only reduce costs, but also optimize performance for application scenarios, better meet customer needs, and accelerate The promotion speed of AI products. In fact, core making is the common choice of many AI companies. Shangtang, Cambrian, and iFLYTEK have all developed chips through self-research or external cooperation. Yunzhisheng, Yuntian Lifei, and Spitz also regard AI chips as the core element to enhance future competitiveness. Comparison of Yitu’s gross profit margin level and comparable listed companies Judging from the prospectus, self-developed chips have brought new growth momentum to Yitu. In the first half of 2020, Yitu’s software and hardware portfolio business accounted for 60.78% of revenue, while in 2017, the software and hardware portfolio revenue accounted for only 10.32%, and more than half of its revenue relied on software business. The core manufacturing has also increased Yitu’s overall gross profit margin, and the overall gross profit margin of its main business has risen to 70.99%. Yuncong Technology integrates the core technology in the human-machine collaborative operating system. On the one hand, it includes a variety of human-computer collaborative application products and overall operating systems that are provided according to customer needs in different application scenarios, serving customers to improve the efficiency of single-point business. And the overall business upgrade; on the other hand, it also includes the general service platform “Qingzhou Platform”. According to the prospectus, the “Qingzhou Platform” integrates self-developed AI technology and strategic customer service experience, integrates automated algorithm training, open source standard application solutions and other general capability models, scenario applications, and industry applications. Quantitative universal service platform. It can be seen from the prospectus that after four to five years of precipitation, the human-machine collaborative operating system has contributed to an increasing proportion of revenue. In the first half of 2020, it has been close to 50%. At the same time, it also drives the growth of gross profit margin, the latest data has reached 53%. However, unlike other companies’ deep cultivation of advantageous scenarios and polishing of their main business, SenseTime proposes a “1+1+X” platform strategy, where 1 stands for R&D and technology industrialization, and X stands for empowering all industries. SenseTime has a presence in multiple vertical fields, and its business covers smart phones, Internet entertainment, automobiles, smart cities, education, healthcare, finance, real estate and other industries. Some media have reported that SenseTime has undergone four changes in its organizational structure since its establishment. In the latest adjustment, the business is divided into smart city, mobile, business insight and innovation. In addition, SenseTime has positioned itself as an “AI factory”, and in order to support the continuous operation of the entire “factory”, SenseTime has invested about 5 billion yuan in the construction of a supercomputing center and open source core algorithms. With the introduction of AI gradually entering the deep water area, the AI Four Little Dragons have fully expanded from single-point business to multi-threaded business, and have moved towards the “three forks” in terms of business strategy and play style. Nowadays, most artificial intelligence companies are facing similar difficulties as the Four Little Dragons, such as long commercial launch cycles, fierce competition, pressure on cash flow, dependence on large customers, data crisis, and so on. After removing the mythological aura, AI has a long way to go.
You must log in to post a comment.