The expansion of storage space, increase of data availability, and continuous improvements in machine learning algorithms such as neural networks and deep learning have increased the applicability of artificial intelligence, or AI, to everyday life.
This article summarizes the state of the global AI industry, compares the involvement of U.S. and Chinese VC firms, and details the current progress of and challenges facing the AI market.
Among Chinese VC firms, ZhenFund has been a pioneer in AI investment, championing AI’s untold impact on the nexus of society and technology.
Over the past few decades, people have debated whether AI has the potential to spark another technological revolution and usher in a new era in human history.
In 2006, Geoffery Hinton, a professor at the University of Toronto, invented a new method of training artificial networks and published three papers demonstrating the potential of deep learning methods. Companies soon invented CPUs compatible with deep learning algorithms, raising computing power limits. With the increased ability of algorithms and computing power, researchers were able to harness the ability of existing data to train artificial intelligence.
Although deep learning has led to major advancements in voice and image recognition, the method is far from perfect – the training method is only applicable in specific fields and relies heavily on existing data. AI is also still far from being compatible with human brains, especially given its shortcomings in logical reasoning and intuition processes.
Despite the relatively nascent nature of the AI industry, its future is extremely bright. The emergence of new AI applications demonstrates the potential impact of AI to transform various industries.
Can Consumer Focused AI Applications Work?
Currently, AI start-ups can be broadly separated into three categories: robotics, natural language processing and computer vision/graphics.
According to research statistics from the Tencent Research Institute and IT Juzi (The State of Art and Trend of Venture Investments in China and the U.S. in 2017), the most prominent AI start-ups in China are those focused on intelligent robots and drones. Natural language processing start-ups, focusing on things like semantic analysis, speech recognition, and chatbots are in the second tier, while computer vision start-ups focusing on facial recognition, surveillance, autonomous driving and image recognition are in the third tier. Start-ups focusing on affective computing (a field which combines psychology, semantics, vision, and environment sensing) are also growing in number, a positive development as this area is currently underinvested compared to the rest of the industry.
Providing enterprise solutions for the business market is also a logical entry point for AI start-ups. AI can drastically improve efficiency and user experiences, as well as enhance a large variety of other industrial applications.
While most AI companies are focused on enterprise solutions, an individual consumer-focused segment of the AI industry has emerged in recent years. Individual consumer-focused companies are exploring ways to monetize user traffic and data. A first wave of AI-generated advertisements for media, cosmetics, and design products were well-received by consumers. The demands of the individual consumer market will play a significant role in the future development of AI. According to statistics, the U.S. already has more than 50 individual consumer-facing AI companies with financing totaling more than 800 million RMB.
What Are Viable Entry Points Across Industries?
In addition to the core competency of technological strength, critical factors for success include product viability and industrial applicability. The next generation of AI products must have real-world applications.
According to statistics, the healthcare industry has embraced AI, especially in services such as medical imaging and case analysis. The automobile industry has also welcomed AI technology, especially in the research and development of self-driving cars. Of the 80 AI companies focusing on automobiles, 30 focus specifically on self-driving AI. Other industries with major AI research include education, financial services, manufacturing, defense, and home furnishings.
What Are Fundamental Differences Between China and the U.S.?
A recent foreign news article praised the development of AI in China, suggesting that China could overtake the U.S. as the global leader of the AI industry. However, this might be a pre-mature conclusion. Below, we take a closer look at underlying market differences:
- Number of companies: As of June 31, 2017, there were 2,542 AI companies worldwide, of which 41%, or 1,078, were located in the U.S. China ranks second with 592, or 23%, of global AI companies. However, the difference between the U.S. and China – 486 companies – is huge.
- Capital investment: Based on existing data, U.S. – based AI companies have raised a total of 97.8 billion RMB (50.1% of global AI investment) compared to China-based companies that have raised 63.5 billion RMB (33.2% of global AI investment)
- Human capital: China has 39,200 AI specialists compared to the U.S.’ 78,700. Training and retaining talent has become a critical impediment to the development of China’s AI industry.
- Number of investment firms: The U.S. has three times the number of AI investment firms as China does. There are 620 AI investment firms in China, whereas the U.S. has 1,800. Furthermore, the number of Chinese firms who have invested in AI projects more than once is 203, whereas in the U.S. this number is 596.
The U.S. remains the leader of the global AI industry development. China’s AI industry still lags far behind the U.S. China does, however, enjoy a unique structural advantage: its AI entrepreneurial environment.
But Is China’s Entrepreneurial Environment More Attractive?
While the U.S. AI industry is much more developed than China’s, Chinese AI start-ups are receiving far more investments than their American counterparts.
Investment Rate: The investment rate for Chinese AI start-ups (69%) is much higher than the U.S. rate (51%). It can be inferred that the Chinese AI market lacks high-quality projects; while there is an abundance of financing available for AI companies, this is tempered by a shortage of human capital talent and technology.
Investment Time Frame: The investment time frame represents period between company establishment and first-round fundraising. On average, Chinese start-ups receive early stage funding (within 9.7 months) much faster than U.S. start-ups (within 14.8 months).
How to Escape The Industry’s Dilemma
There are three critical components of AI development: algorithm, computing power, and data. While China has the requisite database scale and algorithm applications, China’s AI industry suffers from a lack of sufficient computing power. Because computing power is closely related to the development of chip, the weakness of China’s chip industry has affected the development of its AI industry.
According to data, the U.S. has 33 chip firms with fundraising totaling 30 billion RMB. In contrast, China has 13 chip firms with fundraising totaling 1.3 billion RMB, a fraction of the U.S. total.
(US Chip Companies)
The vast majority of Chinese chip companies are focused on ASIC and FPGA, while only one is focused on brain-like chips. In comparison, the U.S. has four GPU companies, with the remainder split evenly between focusing on FPGA and ASIC. It is noteworthy that Rigetti Computing has been researching and developing quantum chips, for which it has received six rounds of financing totaling 472 million RMB.
(China Chip Companies)
All told, China’s chip industry lags far behind that of the U.S., especially in financing and technological core competency. The emphasis of Chinese chip companies on GPUs over brain-like chips has negative potential impact on the growth of China’s chip industry.
What Are The Key Drivers For AI?
Venture capital firms are partly responsible for encouraging interest in AI in both China and the U.S. Some pioneering VC firms have already begun calculating ROIs and thinking about their exit strategies, while other latecomers are just getting in the game.
The first AI investment can be traced back to 1999, when an American VC firm invested in an AI platform called Enkia. Over the past 18 years, investors have clamored to invest in promising AI start-ups as AI-related investment has reached a total of 202.6 billion RMB.
The old PC/mobile-based practice of focusing on founders during Round A, products in Round B, and data in Round C is outdated and not directly applicable to AI investment. Instead of focusing on financials, investors should approach the core business model from a strategic standpoint and understand the underlying advantages of a company. This better demonstrates whether the business has the potential to innovate and improve efficiency, and whether it can scale reasonably.
(Investments by Chinese Funds)
(Investments by U.S. Funds)
Y Combinator has invested 34 rounds in 25 AI start-ups, including Sift Science, Chute, Qventus and SimpleLegal. Sift Science has received over 364 million RMB in financing from Founder Collective, SV Angel, Y Combinator and 14 other VC firms. Y Combinator’s other AI-related investments range from 76,000 RMB to 114 million RMB, and from Seed/Angel to C rounds. Among Chinese VC firms, ZhenFund has been the most prolific AI investor, with more than 37 investments, followed by 28 investments from Sinovation Ventures and 22 from IDG Capital.
Amongst AI fields, computer graphics and imaging has the largest number of investors (291), followed by robotics and machine learning. AI investments reached an all-time high in 2016, with an annual total of 23.19 billion RMB. However, recent investments have demonstrated more caution and a bias towards middle and late stage projects. The Chinese AI industry is oversaturated with investors while also suffering from a dearth of high-quality investments.
The Chinese AI industry suffers from two major problems. First, the demand from investors far outstrips the supply of investable projects and start-ups. Second, the expectation from the market is far higher than what is realistic based on quality and user experience of existing AI applications.
How to Overcome the Various AI Industry ‘Chasms’
Securing funding does not guarantee success for start-ups. According to E.M. Rogers’ diffusion of innovations theory, receiving funding increases a company’s likelihood of bankruptcy. This is a recurring pattern in the technology industry, and AI start-ups are no exception.
As start-ups grow, they will encounter certain barriers and setbacks, or a ‘innovation chasm’. Technology, project, and funding are the three most important elements for any technology start-up. If any one element is insufficient, then the whole model runs the risk of falling into the chasm and never recovering.
There are three major challenges that a start-up overcome to continue growing:
- Technology chasm: Continued technological advancement is critical to a start-up’s sustained success. Only start-ups with proprietary technological advantages and research capabilities can sustain their development and growth. Initial technological capability may be sufficient for a short while, but without additional funding and research, start-ups could be beaten by latecomers.
- Product chasm: The ability to transform a proprietary technology into a viable product is key to a start-up’s success. The entry point and method determines a start-up’s fate. Strong cash flow characteristics enable a start-up to grow.
- Market chasm: If the product doesn’t meet a specific consumer demand or need, then the competitive market environment will pose severe challenges to the company.
All three “chasms” must be crossed in order for a company to succeed. Without continued innovation, the growth of the AI industry is unsustainable. The peak period for investment in AI has already ended. As an AI investment bubble starts to grow, investors need to carefully weigh risks before making decisions.
What Type of New Start-up Projects Will Likely Fail?
According to recent statistics, AI-related investments have experienced tremendous growth in the past two years. However, a lot of AI start-ups struggle to obtain and retain customers; as a result, their business model proves to be unsustainable and they go into bankruptcy.
The mismatch between market over-supply and lower demand reached a tipping point between 2013 and 2015, especially in context of the number of drone, robot waiter, virtual assistant, and intelligent hardware companies; approximately 50 AI start-ups have gone bankrupt in recent years.
Virtual assistant technology is still not mature enough to enter the consumer market. Yingying (应应-雨恒矩阵) and Zhineng Wanshiwu (智能万事屋), two promising virtual assistant technology companies, have gone bankrupt. Drone start-ups are also encountering cash flow problems; recently, Zerotech, Ehang, and even Parrot, a company ranking third in global drone sales, have made layoff announcements. Pearl Automation, a company founded by former senior employees at Apple, has stopped operations following poor product sales despite more than $50 million USD of financing across two rounds.
Start-up failures serve as valuable lessons for everyone else. There are five major reasons why companies fail:
- The underlying technology is not mature enough to guarantee a good product
- Stable growth and market share are hard to ensure
- It is difficult to harness that power to compete with existing players
- The cost is too high and unaffordable for the mass market
- A company lacks the capital for necessary continued product research and development
In a market where product homogeneity is prevalent, the cutthroat competition between AI start-ups will continue. In the foreseeable future, most of the start-ups will be out of the game, and only a few will survive.