As the next era after the mobile Internet, from 2015 to 2016, artificial intelligence companies have experienced explosive growth. According to the latest data released by the government, as of May 8, 2018, the number of artificial intelligence enterprises nationwide reached 4,040, of which There are 1,237 companies that have invested in venture capital.
In the "2017 China Unicorn Enterprise Development Report" released in March this year, Face++ (Xingshi Technology), Shangtang Technology, Yitu Technology, Going to Ask, Cambrian and many other artificial intelligence companies are on the list. Image recognition, intelligent voice, AI chip has become the main entrepreneurial direction.
At the same time, in 2018, artificial intelligence suffered from the pain of commercialization. Some research reports show that more than 90% of AI companies are still in the loss stage, and most of them have annual business income of less than 200 million.
Capital has also become more cautious. Xie Peng, vice president of technology, said: 'Since last year, all capital will ask, how to cash out?'
In 2018, what changes have occurred in the venture capital environment of artificial intelligence and even cutting-edge science and technology? Where are the entrepreneurial opportunities in the new situation? What specific problems are faced in the commercialization of the segment?
Capital flow
Head companies get large amounts of financing, investment tends to be rational
As a breakthrough in structural reform, artificial intelligence venture capital is still receiving much attention in 2018, and many head enterprises have successively completed large-scale financing. But compared with before, capital tends to be rational this year.
In July 2017, the State Council issued the “New Generation Artificial Intelligence Development Plan”. The document requires that the overall technology and application of artificial intelligence will be synchronized with the world advanced level in 2020, and the artificial intelligence industry will become a new important economic growth point. The scale of artificial intelligence core industry More than 150 billion yuan, driving the relevant industry scale of more than 1 trillion yuan.
Huge market cake attracts the attention of capital. IDG Global Executive Vice President, IDG Capital Founding Partner Xiong Xiaoge said at the public event that IDG ranked first in the next three investment areas announced by IDG at the beginning of the year. .
More and more funds are flowing to head startups.
In October last year, Computer Vision's contempt for Face++ announced that it had completed the C round of $460 million in financing. This year, the AI field financing boom became more and more prosperous, and many companies finalized large financing.
At the beginning of 2018, the artificial intelligence company Rokid, the auto-driving technology solution provider Xiaoma Zhixing received about 100 million US dollars in financing; On May 3, the artificial intelligence high-tech enterprise Ubisoft announced the completion of the 820 million US dollars C round of financing, refreshing AI Field single-round financing record; On May 31, computer vision company Shangtang Technology announced that it has received 620 million US dollars in C+ round financing.
At the beginning of June, Computer Vision Corporation cloud completed the latest round of RMB 1 billion financing from technology; On June 21, the smart chip company Cambrian completed a round of hundreds of millions of dollars in financing, and the valuation soared to 2.5 billion US dollars.
Capital pursuits have created a large number of unicorns in the AI field. According to public reports, there are currently 13 unicorns in China's AI field with a valuation of more than 1 billion US dollars. The highest valuation is 5 billion US dollars. The soup is worth 4.5 billion US dollars. There are also 9 prospective unicorn companies.
Compared with the influx of funds into the head artificial intelligence enterprises, investment institutions have gradually become more rational in their investment in the AI field. Zhang Tianlei, CEO of Beijing mainline technology, has publicly stated that the artificial intelligence investment boom has gradually receded.
This view can also be reflected in the investment data. The number of investment events in IT Orange's 2017 is 353, and the number has dropped. The number of new AI companies has declined, less than 50.
Falling pain
The AI era, which can be founded only by doctors, has passed.
The industry landing of "AI Commercialization's Second Revolution" - 2017 China AI Investment Market Research Report shows that from the perspective of investment structure, in 2017, China's AI investment market has shifted to the development period, but the gestation period and birth period enterprises The signs of investment growth are not obvious. As of the first half of 2017, more than 50 AI companies have closed down due to a break in the capital chain.
One of the 'death causes' is the pain of the company.
'Now the artificial intelligence startup company may be different from the previous two years. Before, there was a luxury team, a lot of money. From last year, all the capital will ask, how to realize it? 'Artificial Intelligence Unicorn Company, Defiance Technology Xie Peng, the vice president, made a speech in the public event not long ago, pointing to the pain of the commercialization that the artificial intelligence industry is experiencing.
On June 30, the Beijing Municipal Commission for Economic and Information Technology released the White Paper on the Development of Beijing Artificial Intelligence Industry for the first time, and found out the number of artificial intelligence enterprises in the country: 'Artificial intelligence enterprises showed explosive growth from 2015 to 2016, as of 2018 On May 8, a total of 4,040 artificial intelligence enterprises nationwide, Zhongguancun became the highland of artificial intelligence innovation in China. 'The Deputy Director of the Software Division of the Commission, Yu Jing said.
However, according to the relevant report of the billion-dollar think tank, more than 90% of the AI companies that have developed with the AI investment and entrepreneurial tide after 2016 are still at the loss stage.
The industry generally believes that 2018 is the first year of artificial intelligence commercialization, how to become less than 10% of the profitable companies in the industry, is the focus of capital and business.
The approach taken by the artificial intelligence company Yuntian Lifei, which focuses on the field of visual intelligence, is to find industry experts who understand cross-border and set up a comprehensive cross-border team.
'Doing vertical applications spans many different fields, but this is not to say that experts in several fields can easily get things done by sitting together. Vertical application scenarios involve many fields, so you need to have the quality of cross-border innovation. From R&D products to market expansion to team building, we will do cross-border exploration in all aspects. ' 'Thousands of people plan' National special experts, Yun Tian Lifei CEO and co-founder Chen Ning said.
Chen Ning has already considered the issue of productization of technology-based enterprises. He believes that at this stage, AI companies need to provide services for enterprises in other industries, and can directly solve their problems. This requires vertical fine In the sub-areas, AI companies must go deep into the industry to capture the pain points of the other side. Based on this, the Yuntian Lifei team is constantly introducing macro security, experts in the security field, and assisting Yuntian Lifei to explore business models. , make mature technology land as soon as possible.
Chen Ning’s point of view coincides with the search for China’s maker tutor, Li Kaifu, chairman and CEO of Innovation Works.
In Kai-Fu Lee's view, in the early days of AI, a doctor with a scarce knowledge resource can build a company, and into the AI-energy business era, AI companies also need a business leader who understands technology, understands products, and understands the land. The AI era of doctoral entrepreneurship alone has passed. '
Autopilot
Incoming people but have already opened the position
Artificial intelligence has different landing speeds in each subdivision scene, and AI combined with automatic driving is in front.
Capital and entrepreneurs have already predicted that autonomous driving will open the next personal computer interaction era after PC, Internet and smart phones. Li Kaifu once asserted: 'After Windows and Google, autopilot is likely to become the next stage of operating system. . '
In 2012, as Hilton in the United States broke through some of the shortcomings of deep learning in local convergence, in the ImageNet competition that year, convolutional neural networks won with overwhelming advantages, and technology giants such as Google and Facebook began to apply this breakthrough. In the development of automatic driving, the tide of self-driving entrepreneurship began here.
According to the statistics of China Electric Vehicles 100, from 2015 to November 2017, there were 193 investment and financing events in the field of automatic driving, amounting to 143.8 billion US dollars. According to German management consulting firm Roland Berger, by 2030, The global revenue of driverless taxis will be as high as 1.5 trillion euros (about 1.7 trillion US dollars). The trillion-dollar market cake also makes this track into the public.
China's automatic driving is late, but it is developing rapidly. Old car manufacturers, new car power, autopilot solution providers, fleet operators... The number has been more than 40.
In the "2017 China Unicorn Enterprise Development Report" released by the Ministry of Science and Technology in March this year, Xiaopeng Automobile, Weimar Automobile, Weilai Automobile and other new vehicles that are involved in the autonomous driving business are on the list. Recently, the company unveiled the company's driverless strategy and shared travel strategy – creating a driverless Robo-taxi customized for shared travel. It is reported that the first batch production plan is put into commercial operation in 2025, providing users with daily services. 100 million kilometers of travel services.
In terms of autopilot solutions, Xiaoma Zhixing, Jingchi Technology, Heduo Technology, etc. 'Baidu Department' (the founding team is mostly from the Baidu unmanned vehicle team) has become the most eye-catching star.
Established in the second half of 2016, Xiao Ma Zhixing completed the US$112 million Series A financing in early 2018. Not long ago, Xiaoma Zhixing was granted the Beijing Municipal Government's road test license for self-driving vehicles, becoming the first to obtain Beijing's automatic driving road. A licensed start-up company. A person familiar with the matter said that the current valuation of Xiao Ma Zhixing has exceeded $1 billion and has become a unicorn in the field of autonomous driving.
In September 2017, Jingchi Technology received a $52 million Pre-A round of financing led by Qiming Ventures, and is currently launching a new round of financing; in October, Hedo Technology was led by IDG Capital, NavInfo Tens of millions of dollars of angel round financing; In May 2018, Roadstar.ai received $128 million in Series A financing, setting the highest financing amount for the newly created start-up autopilot company at that time...
From the capital level alone, the autopilot startup has already opened its position, and the ensuing change is the autopilot entrepreneurial situation.
'The window of unmanned entrepreneurship is about to close soon.' In the L4 level auto-driving truck company Tucson future CEO Chen Mo, if at the end of 2018, a team of more than 50 people has not been formed, regardless of funds, talents In terms of technical accumulation, it is basically difficult to compete with the first echelon.
In the current competitive stage, Xiaoma Zhixing CEO Peng Jun believes that 'the most important thing is to grab people and steal money.'
The characteristics of autonomous driving determine that this is a heavy asset business, and the threshold for small players to follow up imitation is higher. This also means that the competitive logic of the current autopilot industry is: Who can find more money faster? , gather more top talents, who can lead this track.
AI chip
'China Core' spurt development
The 'Zhongxing Incident' in April this year made the people realize the importance of mastering the core technology of the chip. A group of artificial intelligence startups and even Internet giants have announced their intention to enter the AI chip field.
On April 19th, Alibaba revealed that Alibaba Dharma is developing a neural network chip, Ali-NPU, which will be used for image and video analysis, machine learning and other AI reasoning calculations; May 16th, Yunzhisheng Released the first AI series chip UniOne for the Internet of Things and the first generation chip 'Yuyan'; On May 24, I went out and asked about the release of the first mass-produced AI voice chip module in China, 'Qingxin' Mobvoi A1; On June 26, Rokid released its AI voice-specific SoC chip KAMINO18...
In June, AI startup company Shenjian Technology also announced that its AI chip listening Tao series will be available in the second half of this year; Spirent CEO Gao Shixing confirmed at the same time that the company is building an AI voice chip, which is expected to be released in the second half of this year. (Made the chip through a series of process steps like a pipeline).
Counting the Cambrian, Horizon Technology, Bitcoin and other startups, as well as Huawei, Hikvision, Dahua, Dajiang and other famous hardware manufacturers, the artificial intelligence chip market is already a red sea.
Even without the ZTE incident, the development of AI-specific chips has been put on the agenda of these companies. Yun Zhisheng founder and CEO Huang Wei has publicly stated that: No chip, it will die. This view is also recognized by Rokid CEO Zhu Mingming.
This is because the rapid development of artificial intelligence is inseparable from three core elements: data, algorithms and computing power. The chip is equivalent to AI gasoline, providing artificial intelligence with the computing power to survive. If you purchase existing chips on the market Will greatly increase the cost of these startups.
On the one hand, the existing chips on the market are sold at high prices. For example, the NVIDIA Drive PX2 used in self-driving cars is priced at $15,000. On the other hand, most of the existing energy products are general-purpose chips, functional. The pertinence is not so strong. Considering the versatility and flexibility of modification, many reserved modules need to be added in the design, so the power consumption will be high. In other words, the electricity bill is extremely expensive.
'If Rokid doesn't research the chip, the price of 'Rokid Me' is now 799 yuan, and the price is raised to at least 1,500 yuan. ' Zhu Mingming said.
However, Horizon CEO Yu Kai also issued a warning: 'We must also be aware that chip research and development is a complex project, and we need a scientific and rigorous attitude. Chinese chips should be self-reliant, but they should also be calm, and avoid the big leap-forward national core.'