Author: Maguan summer
Editor: Kang Xiao
In 2017, the rise of artificial intelligence is in full swing. Entrepreneurs are carnival in the capital boom, and they are rumored by the revolutionary changes that AI may bring to various industries. However, in the past 2018, the blockchain wave and shared bicycles. The ups and downs of the people have once eclipsed the artificial intelligence and the stars are no longer.
This is not a bad thing. It is far from the smoldering bubble. After the arrival of the capital winter, the AI industry is more down-to-earth. Whether it is a technology giant or a startup company, it is a top priority to find application scenarios based on its own technical capabilities and characteristics. “Landing” is this The theme of the year.
In the past year, the Matthew effect of China's artificial intelligence industry has become stronger. When small companies are faced with a general financing dilemma, head star companies are refreshing the financing records of this industry; BAT has already completed the AI industry layout. Entering the era of full application, the startups on the popular track face more intense competition; the next 2019, when the investment boom receded, small AI startups facing funding shortages and application scenarios were only found. Vertical field applications can win more living space.
Industry chain distribution
According to the China Institute of Information and Communication Research, as of September 2018, China (excluding Hong Kong, Macao and Taiwan) had 1,122 artificial intelligence enterprises, which were distributed in all aspects of the AI industry chain.
The AI industry chain can be divided into three categories according to the base layer, technology layer and application layer. The base layer provides computing power for the entire industry. The hardware includes chips, sensors and middleware. The software is a platform for computing power for the entire AI industry chain.
In the field of basic platforms, mainly provided by giants such as Tencent, Baidu and Ali, it is worth noting that the mining machine manufacturers such as Bit China are transforming into artificial intelligence chips.
AI industry chain distribution (Tencent deep network mapping)
The technical layer provides general AI technology capabilities for the overall industry chain. The sensing layer includes two machine-aware tasks: computer vision and speech speech recognition, which are relatively mature in the current technology. The cognitive layer is positioned as “machine brain”, including knowledge maps. /Semantic analysis and intelligent Q&A/virtual assistant two core areas; the top platform layer provides technical application services such as deep learning and pattern recognition in the form of a common technology application platform, and docks the application layer.
At present, the domestic startups mainly targeting the To C market are mainly concentrated in the technical layer, such as the University of Science and Technology, which provides voice services, and the business technology of the smartphone companies, such as Shangtang Technology and Yushi Technology.
The application layer is a specific consumer-grade terminal application and industry scenario application. The consumer-grade terminal application includes three directions: intelligent robot, intelligent drone and intelligent hardware; the industry scene application connects various AI application scenarios of external industries, such as automatic driving. , smart finance, smart healthcare, smart education, smart retail, smart security, smart marketing, smart city, and more.
In addition to some consumer applications, this is a huge To B market. The development direction of “Industry Internet” proposed by Tencent is closely related to artificial intelligence.
As the most important player in the domestic AI industry, the Internet giant BAT has completed a full range of AI technologies, platforms, application scenarios and foreign investment. BAT's focused business areas have also reflected its layout in the AI industry. In general, Baidu focuses on platforms and autonomous driving; Ali focuses on the application of artificial intelligence in the field of data services and the underlying technology; Tencent focuses on platform and technology open, and externally balanced layout.
BAT artificial intelligence layout (Tencent deep network drawing)
At the World Internet Conference at the end of last year, Li Yanhong said that Baidu hopes to build a basic platform to open up some of the basic capabilities of artificial intelligence, such as the framework of deep learning, speech recognition technology, and image recognition technology to the outside world. At the application level, Li Yanhong is more optimistic about two scenarios: smart home and intelligent transportation.
Ma Huateng believes that the interconnection between the entire future external network between the human brain and the human brain will be meaningful. There will be many changes in the new human-machine interface in the next 10 years. If the basis of 5G plus VR, the endurance and the computing power are solved, there will be some practical VR calculation software and AR calculation software will appear, which is also an epoch-making breakthrough.
AIUnicorn landing competition
At present, the head AI Unicorn startup is concentrated in the track of computer vision, speech recognition, AI chip, smart finance, smart medical technology and other technologies with mature technology or application scenarios.
Most of these companies were founded around 2012. Highly educated entrepreneurs with technical backgrounds felt the wave of AI technology development, set up a team, and started a business journey from the field they are good at.
Starting the business without knowing the AI, the company has won a large first-mover advantage for these companies. In the hottest 2016 of the venture capital market, these companies that have completed technology research and development, team running-in and initial market testing have won The extremely high valuation financing, with the blessing of funds and the accumulation of word of mouth, the status of the head startup company has been consolidated and become an important pole of the Chinese AI industry.
Tencent Shenwang organizes according to public information
In the most mature computer vision field, Shangtang, Yuncong, Defiance, and Yitu are all in the most easy-to-reach and broad-market areas of finance and security.
In the financial field, Shangtang Technology's customers cover many Internet financial institutions and commercial banks; Yuncun Technology, which is headquartered in Chongqing, is the first face recognition in China's banking industry, although its propaganda power and geographical advantages are not dominant. Large supplier; Defiance Technology is backed by the investor Alibaba, providing face recognition login support for Alipay clients, and also providing face recognition services for many Internet finance companies and commercial banks; The real-name authentication solution relies on face comparison and live detection technology to provide financial companies with omni-channel solutions from counter to self-service to mobile.
In the field of security and smart cities, the main landing scene of Shangtang Technology is to solve anti-terrorism, criminal arrest and public safety through intelligent video; cloud is mainly based on core technologies such as large library retrieval, dynamic deployment control, and trajectory tracking, and various business lines. The combination of algorithm items has been applied in many public security systems; Vision Technology is based on smart camera hardware, covering face recognition access control, sky-eye monitoring system, dynamic face recognition monitoring, and human-integrated application. According to the research and development of "Technology", the large-scale portrait platform has also been applied to the public security field. It also provides portrait comparison systems for the General Administration of Customs and China's border inspection.
In addition to finance and security, Shang Tang, Yun Cong, Defying and Yitu also have different levels of competition in the fields of medical care, transportation and retail. When the technology and the landing scene are not much different, the fierce killing is inevitable.
However, for the AI Unicorn Company, which is still in the capital winter, it is not only bad to expand the application scenarios in different fields, but also to accelerate its commercial landing through equity investment can become another choice.
On August 7, 2018, Shangtang Technology led the projection technology of 1.36 billion yuan in D-line financing. On June 21st, Shangtang Technology also led the medical Internet company Helian Health's $75 million B-round financing, the latter. The post-mortem valuation was nearly $1 billion — — the unicorn started making small unicorns. Shang Tang's current public investment cases include 51 VR, Helian Health, Suning Sports, and Shadow Technology.
On November 2nd, Vision Technology appeared in the chain convenience store “Good Neighbors”; in the new round of tens of millions of dollars of financing investors, it is widely believed that the current chain convenience store industry continues to invest in store AI equipment. And digital transformation is the basis for cooperation between the two. The phenomenon of Shang Tang and contempt for investment while financing is also seen as the epitome of the current commercialization of head-to-head AI enterprises.
Investment institution layout: funds poured into the head company
The artificial intelligence track covers all major Chinese investment institutions, including Zhenge Fund, Sequoia Capital China, IDG, Innovation Workshop and other industrial capitals such as Tencent, Ali, Baidu and Lenovo.
The active capital environment has obviously greatly promoted the cultivation of the AI industry, and the mutual promotion between capital and enterprises has also enhanced the overall strength of the Chinese AI industry.
(Tencent deep network finishing, ranking in no particular order)
2018In the year, the concentration of funds to the head company is a major trend in the field of AI venture capital.Even in the face of capital winter, the star projects on the popular track such as Shangtang Technology, Youbie, and Defiance Technology are still sought after. Taking Shangtang Technology as an example, this computer vision technology unicorn company only this year 4 From January to September, a total of $2.22 billion in financing was completed for the C, C+ and D rounds, and the valuation climbed to $6 billion.
The statistics of IT Orange show that there were 410 investment events in the AI field in 2018, with a total investment of 107.8 billion yuan. Compared with 2017, the investment event decreased by 1/3, but the total investment increased by 1/4. Earlier projects in the project were reduced in investment, and the proportion of projects in and after the C round increased. This aspect means that the difficulty of financing for start-up AI companies, on the other hand, also confirms the reality that funds are flowing to the head company.
From a technical perspective, computer vision is still the most popular artificial intelligence technology in the venture capital circle, which is related to its high technical maturity and wide application scenarios in security and finance industries. In addition to Shangtang Technology, the world's CV field unicorn company's 旷世科技, 依图科技 and Yun Cong technology have also completed hundreds of millions of dollars in financing in the past year.
But another relatively mature AI technology —— speech recognition, but rarely see large amounts of financing, the first echelon of computer vision track company valuation has reached billions of dollars, the first echelon company on the speech recognition track The valuation is generally only about 3 billion yuan.
Like the Internet industry, the business model of artificial intelligence companies can be basically divided into To C, To B and To G. In particular, the security market with a budget of more than one trillion yuan per year in the To G business has already become the Red Sea of many AI startups. However, due to the lack of To G application scenarios in the speech recognition track, the scale of revenue and valuation is not the same as that of companies on the visual recognition track. According to data released by the billion-dollar think tank, the computer vision head company's revenue in 2018 can reach about 2 billion yuan, while the revenue of the head company in the voice recognition field is mainly concentrated around 200 million yuan.
From the perspective of industry applications, intelligent robots, smart healthcare and smart finance are still favored by various investment institutions. In the intelligent robot track, UBS Technology has become the second-year financing runner-up after Shangtang Technology with a C-round financing of US$820 million. Players such as Daxie Technology, Geek+, and Turing Robot also obtained capital. Financing completed financing; in the field of smart medical care, Huihui Huiying, Pushu Technology, and Shenrui Medical have all made some gains in financing; the fourth paradigm of Smart Finance Circuit also recently completed 1 billion yuan of C round financing, this Artificial intelligence technology and service providers are among the new unicorns in the industry with a valuation of $1.2 billion.
The era of AI application landing has arrived
If 2016 is the first year of AI industrialization and 2017 is the year of AI industrialization, then 2018 is undoubtedly the year of AI application. With the giants such as BAT completing their layouts on their own tracks, in 2018, several giants handed over beautiful responses to the AI industry application.
At the Baidu Developer Conference in July, Li Yanhong announced that Baidu's L4-class autonomous bus Apollo has achieved mass production. In November, Li Yanhong announced at Baidu World Congress that Baidu will launch the AI Smart City “ACE Ace Plan” and announce There are also plans for Baidu Apollo and FAW Red Flag production L4-class self-driving passenger cars and Baidu and Volvo to produce electric self-driving cars.
In March's MWC, Alibaba showed the city's brain in front of the global media; in June, Gaode map teamed up with Aliyun to release the city brain & middot; smart transportation strategy, urban brain & middot; smart transportation public service version also debuted. Ali Gaode City Brain & middot; Smart Transportation will be the first to land in 50 cities including Beijing, Shanghai, Guangzhou and Hangzhou. The vision is to save an average of 10% of each user's travel time;
In June, Tencent officially released the first AI medical auxiliary treatment open platform, announced the opening of its first AI+ medical product & ldquo; Tencent & & 辅 & auxiliary engine, to help the hospital HIS system, Internet medical services to achieve intelligent upgrade, build Cover the wisdom of medical science before, during and after diagnosis. In November, Tencent AI Lab released AI's auxiliary translation product Transmart, a product based on Tencent's self-developed neural network machine translation engine, which provides intelligent recommendation of real-time translations and provides complementary translation APIs and customized services for partners. The era of application of AI has arrived.
While BATs are leading the application on their respective tracks, AI startups facing high valuation and low revenue pressure will also find application scenarios that match their technical characteristics and achieve revenue as the focus of their work, but the reality is In the various stages of the industrial chain, the players are crowded, and the competition for commercialization has become fierce.
In 2019, more AI startups need to find applications for technology and algorithms.
Liu Qingfeng, chairman of Keda Xunfei, said that the next three to five years will be the most critical pattern determination window for artificial intelligence. In these three to five years, who can make the artificial intelligence application truly scale and let the application fall, who can take the lead in the future smart industry. But from the past year, the time left for AI startups may not be that long.
Under the cold winter, “landing” is the theme that runs through 2018. When the giants complete the layout, they turn to their respective areas to start commercial staking; AI unicorns are also expanding the application scene. It will be transformed into a platform by means of equity investment; small AI startups that face funding shortages and limited landing scenarios will be eliminated or inevitable by market reshuffle.
It is predicted that the view that a large number of AI companies will fail in 2019 will gradually become mainstream, but entrepreneurs' ability to adapt to environmental changes should also be taken seriously. In the cold winter, the balance of investment tilted to the scene, advocating data and algorithms is no longer a pass on the financing road, entrepreneurs have become more down-to-earth, and began to emphasize orders, customers, commercial layout. In 2019, AI startups that can find application scenarios in the vertical arena will also win more space.