In October this year, Face ++ completed financing of C + up to 460 million U.S. dollars, which not only surpassed the 410 million U.S. dollars of round B financing, but also refreshed the global AI financing record. At 11 In January, Cloud Technology completed RMB500 million of round B financing, together with the previous government support of RMB2 billion from cloud technology from Guangzhou Municipal Government. Cloud received a total of 2.5 billion yuan of development funds from science and technology.
The above is just the financing of the three head unicorn companies in the field of computer vision. If you count other players in the field of computer vision, as well as the financing of various startup companies in the field of automatic driving, AI chip, smart hardware and robotics, 2017 In the whole year, the amount of financing in the AI field will exceed 20 billion yuan.
China AI company's valuation is too high?
After obtaining the above financing, the valuation of domestic AI companies has become staggering.With computer vision leader Shang-Tang technology as an example, Alibaba invested in the rumor of 15 billion yuan, the valuation of Shang Tang Technology reached 30 billion US dollars, equivalent to 19.6 billion yuan, and Kuang Shi, cloud from the valuation has already surpassed 10 billion.After experiencing the investment boom from 2016 to 2017, the Chinese AI company's valuation rose like a rocket, And even compared to bitcoin is not inferior, so that investors lament AI project is too expensive at the same time, many people in the industry began to question the valuation of AI in China is too high.
Of course, if you take the secondary market is one of the few similar targets, but also the Chinese AI sector 'Big Brother' Branch News (58.820, -0.82, -1.37%) to compare, the valuation is no big deal. The P / E ratio has not fallen below 100 times since 2015, reaching an astonishing 177 times now. In the first three quarters of 2017, the share capital of iFLYCOM is only 172 million yuan. At present, the market capitalization is more than 85 billion yuan.
For the current valuation of AI companies in the country, we may only say that artificial intelligence is so important to the future that capital and investors are willing to pay so much premium to get 'ticket.' And most worth considering In the next two or three years, how many AI companies with high valuations should develop to support their valuation over the long term? To do this, it is necessary to establish a maturity model framework for observing and tracking the number of " Potential giants of the road 'company's development path.
AI Enterprise Maturity Model
Historically, research in the field of artificial intelligence has experienced two troughs.Throughout this wave of artificial intelligence, the essential cause of the wave of artificial intelligence is the concentrated outburst of deep neural network technology that has accumulated more than 30 years tirelessly, especially in the field of computer vision Great progress has been made. AlphaGo success to a large extent artificial intelligence technology once again returned to the public horizons, which in turn promoted the pursuit of capital for outstanding projects in artificial intelligence, but also makes the original has been fragmented into various self-study Once again, the world of artificial intelligence has returned to artificial intelligence under the banner of artificial intelligence.
After experiencing the second artificial intelligence trough in the 1980s and 1990s (the disintegration of the expert system), there are only a few researchers in sub-disciplines such as robotics, machine learning and cognitive science which still insist on the field of artificial intelligence. Which outstanding scientists even more rare.Artificial intelligence scientists who persevered these batches in recent years has been domestic BAT, FLAG and other giants meltdown, of course, then these elite scientists in chasing the capital to re-start have set up their own artificial Intelligent companies, such as former Google star scientist Li Zhifei set out to ask questions, in addition to Wu Nanda Baidu side, the three deep learning laboratory core figures Yu Kai, Huang Chang and Yu Tienan founder of Horizon Robot; Baidu driverless General Manager Wang Jin and chief scientist Han Xu founded King Chi technology.
In the face of the throes of curiosity into the field of artificial intelligence into the capital of all walks of life, outstanding scientists and their projects have become extremely scarce, the domestic leader in artificial intelligence almost ten fingers can count on.In the review of the past two years in the field of artificial intelligence Financing events, we can clearly observe that the investment logic of capital can be said that is not complicated - people and teams, and venture capitalists have become the most mainstream form of artificial intelligence venture capital in the dividend period of this capital , Artificial intelligence scientists have become the most profitable group.
However, the dividend period of this wave of capital has basically belonged to the past tense.
For example, the leader Shangtang Technology has supported its valuation of nearly 20 billion yuan in the past year or two. To some extent, Professor Tang Xiao-Ou and the rhetoric expressed by Shang-Tang Science and Technology themselves: "Proficient in deep learning People have basically read PHD, China in this respect now a total of one or two hundred people, and Shang Tang swept 120. 'Shang and future Shang to continue to support its valuation, in addition to people and the team must be Commercialization of its AI products In order to better observe and track the current evolution of AI companies in the future, this paper proposes an AI maturity model.
Stage one: basic technical service providers
As the AI application scene is not yet mature and market validation, any AI companies tend to engage in the accumulation of basic technologies, both for domestic and foreign AI start-ups are the same, the most typical case is DeepMind. Currently Most domestic AI start-ups belong to or are ready to go beyond this stage. The accumulation phase of basic technologies is characterized by the competition for talent and the development of AI-like technologies and algorithms similar to those in the laboratory. Its core drivers are the team and the people.
At this stage, most AI companies are keen to send papers at various top-level conferences and participate in international AI competitions. Due to the lack of commercialization of technologies, AI technologies are often provided only through project-based programs Services, that is, selling rough models and selling rough algorithms, such as face recognition technology services, basic language recognition services, knowledge mapping in the financial field, etc. However, people and algorithms as the core competence of enterprises are unsustainable, especially in current deep learning The area of algorithm bonus period becomes shorter and shorter, and talent gap is gradually being filled.
Stage two: the overall solution provider
For the underlying AI service provider, one obvious fact is that the single point technology itself can not form a complete application and product, such as a narrow face recognition technology that needs to be integrated with other businesses or products such as technology + camera , As a smart video surveillance device, or combined with traditional payment products in the password / phone verification code coupled with a layer of face recognition verification, similar to the ATM machine to increase face recognition in order to form a particular scene with commercial value application.
The underlying AI technology service providers need to evolve toward holistic solution providers that form the core competency of their own holistic solutions from both deep-sectoral industry scenarios and data operations perspectives, with core drivers evolving from humans to scenarios and data at this stage - focused Depth of the breakdown of scenes and data and the overall solution behind them.
AI enterprise team from science and technology, with its CAS background, the main attack face recognition and anti-fraud in the financial scene, has taken over, including CCB, ABC, Bank, China Merchants Bank and other 50 clients.
Stage three: AI product phase
AI technology alone and the overall solution itself is difficult to achieve a great AI company, because even the best technology and solutions can not escape as a subsidiary of other people's product positioning, AI companies to go further, the product Is a hard to bypass the road, which we can fly in the ITC and Baidu took a few years ago detours, see more clearly ... For the current valuation of the domestic AI companies, the future can not do without the launch of the market On the influential and sticky industrial / consumer AI products, simple technology and the overall solution is easy to touch the ceiling.
Stage four: collaborative eco-builders
What is Collaborative Ecology? Take Amazon Echo as an example, as the number of applications (Alexa's Skill) at the start of the Amazon Echo roll out is appalling, then when Echo comes out After the explosive growth of volume, Amazon has attracted a large number of developers into the Alexa ecosystem. Currently, Alexa already has more than 10,000 Skill. The key driving force behind this is Amazon's powerful cloud computing capabilities - through Amazon Voice System, and ASK (Amazon Skills Kit) are open to build an ecosystem with low barriers to development and even developers that do not even need any technology for speech recognition. AVS will solve all the problems of speech recognition and semantic processing This Echo-like collaborative ecosystem, which is able to power a large number of participants around the AI product center, is a source of future profits for the business.