China's Future Investment Watch series - China's Man-Creating Expeditions: The Upcoming Golden Decade of AI
Antitrust, reform in education, real estate and medical care policies, Common Prosperity... all of these show that China is undergoing earth-shaking change. These changes have brought large uncertainty, making many investors afraid to invest in Chinese projects and companies. However, over the next decade, China sure will become the world's largest economy. How to better understand the opportunities and risks of the Chinese market and deal with certainty and uncertainty is a crucial problem. EqualOcean launched a series of research, China's Future Investment Watch, hoping to provide clues for global investors.
There is a view on the Chinese Internet that it took billions of years for the earth to create life, and the first computer was born only 70 years ago – many people think the human is more efficient than natural evolution in creating intelligent items. This reveals the soaring ambition of mankind. However, human beings are not satisfied with this. They want to create intelligent life like themselves by means of technology, just like God. Under this background, artificial intelligence (AI) has been pushed to the forefront of the world.
We have to admit that AI is one of the keys to further liberating productivity in the future. China has clearly recognized this. China started developing AI-related industries later than western developed countries (especially the United States), but it has made some achievements in all sub-sectors and is occupying the market mainstream. We believe that under China's elaborately planned, the golden decade of China's AI industry is coming.
Three layers in China
The AI industry map is mainly divided into three sub-sectors (layers): the input layer, the hidden (technology) layer, and the output layer, representing the upstream, middle and downstream of the industry. At the Governance Forum of the 2021 World AI Conference, the China Institute of Science and Technology information released the Global Innovation Index 2020 report. The report shows that China's AI innovation index ranks second among the participating countries, second only to the United States, and forms a gap among other countries. However, the development priorities of the two countries are different. American AI focuses on the input layer, while Chinese AI focuses on the output layer.
1. Input layer
Computing power, algorithm platform, and data resources are the three elements of the input layer of AI. The input layer is a more academic sector, and the number of patents is one of the important evaluation criteria. As the birthplace of AI, the United States has a world-leading position in the input layer. Among the list of 2020 global top 10 AI patent companies, more than half of the companies originated in the United States, while only two originated in China. However, the advantage of this leading position is declining year by year, an important reason is supported by governments. From 2017 to early 2021, the State Council, the National Development and Reform Commission, the Ministry of Industry and Information Technology of China, as well as local governments have issued a bunch of policies of AI. In 2021, the market scale of China's input layer of AI exceeded CNY 50 billion, accounting for about 30% of the total profit of the AI industry chain. We predict that in the next decade, the scale of China's AI input layer market will reach CNY 257.3 billion in 2030, with a 5-year CAGR of 37%, indicating a market with huge potential.
2. Hidden layer
AI hidden layer (technology layer) is a general AI technology developed based on the input layer. It is represented by cognitive and perceptual computing technology, which simulates human intelligence from two aspects of perception and cognition, and endows machine learning and inference ability. The hidden layer mainly includes professional technology (computer vision, speech recognition, NLP) and general technology (including machine learning, knowledge graphs). Compared with the United States, the development of China's AI industry relies too much on open-source code and existing mathematical models, resulting in a lack of professionalism and pertinence, which often cannot meet the actual requirements of specific tasks. Taking image recognition technology as an example, AI developed with open-source code can accurately recognize human faces, but it is difficult to meet clinical requirements in medical areas.
3. Output layer
China's AI industry is demand-driven. The number of output AI companies in China accounts for the highest, accounting for 84.05% of all the companies in the AI industry. In the output (application) layer, AI has a wide range of application scenarios, including more applications in finance, security, education, medical treatment, retail, and even robotics fields. China's AI output layer benefits from huge training data and international open-source algorithms. Chinese AI companies can more quickly apply AI technology to companies of different types and deeply explore the application scenarios of AI. During the period of rapid development of AI in China (from 2013 to 2017), the number of investments in the output layer of AI increased from 134 to 1,062, with an annual compound growth rate of 67.8%, far exceeding the number of investments in the hidden layer and the input layer.
In fact, the entrepreneurial age of AI is coming to an end, especially for the areas in the first two layers. According to the statistics of EqualOcean Intelligence, Chinese AI start-ups began to develop rapidly from 2012 and peaked from 2016 to 2017. Then the enthusiasm has been cooled down. In 2020, only 57 new AI companies were established, while the number used to be 184 in 2019. AI start-ups have basically completed the accumulation of original technology, and the focus is being transferred to developing application scenarios. AI companies have gradually developed from the early technology-driven stage to the business-driven stage. In particular, AI technology based on vision, voice and text is relatively mature. In the future, the development of AI companies will rely more on the insight of the solutions to the actual needs in life...