The "2021-2022 Chinese Intelligent Computing Power Development Assessment Report" was released

On October 26, the "2021-2022 Chinese Intelligent Computing Power Development Assessment Report" (hereinafter referred to as the report) was released at the AICC2021 Artificial Intelligence Computing Conference, aiming to comprehensively consider and evaluate the development level of Artificial Intelligence in China through four aspects: macroeconomics, technological maturity, AI labor supply, industry and geography. The 2021 artificial intelligence city ranking has attracted much attention, with Beijing, Shenzhen and Hangzhou continuing to lead the top three, Nanjing entering the top five for the first time, Jinan entering the top ten for the first time, and Shanghai, Suzhou, Guangzhou, Chengdu and Hefei entering the top ten list.

The report was jointly released by IDC and Inspur Information for the fourth consecutive year since 2018. The report comprehensively considers the four aspects of macroeconomics, technological maturity, AI labor supply, industry and geography, evaluates the current situation of the development of Chinese intelligent industry, and provides references and action guidelines for promoting the development of industrial AI. The construction of computing power infrastructure reflects the comprehensive investment of a region in artificial intelligence, and for the first time this year, it was taken into account in the regional assessment.

The report pointed out that AI chips show a diversified development trend, AI chip computing power continues to increase to meet the growth trend of model scale; China's AI server market is growing rapidly, Chinese manufacturers lead the world, in 2020, the global AI server manufacturers wave, DELL, HPE market share ranked in the top three, the future AI server will be in the direction of diversified openness, green energy saving. The integration of AI and cloud is an inevitable trend, it is expected that by 2025, the proportion of public clouds of Chinese intelligent servers will exceed 50%, private clouds, government cloud, industry clouds, etc. are also booming, and hybrid IT is the first choice for enterprises. The development of algorithm models is becoming more and more complex, the huge model will be the foundation of large-scale innovation, and the emergence of huge models such as "source 1.0" has made the construction of large models and improving AI processing performance a development trend; application scenarios have transitioned from fragmentation to deep integration, from single-point application scenarios to diversified application scenarios. Compared with 2020, artificial intelligence has played a particularly significant role in finance, manufacturing, energy, utilities and transportation; the computing infrastructure represented by the intelligent computing center makes computing power services easy to use by providing public computing power, data and algorithm services, and solves the problem of supply of computing power services. (Zhan Zhao)

Top 10 Insights into Industrial AI

Insight 1: Countries accelerate the layout of artificial intelligence Technology companies lead technological upgrading China and the United States lead the world

More than 60 countries and regions around the world have issued artificial intelligence policies and issued national AI strategies. The United States is driven by technology companies and promotes the development of AI with continuous computing power and algorithm innovation. China has driven the layout of computing power facilities with new infrastructure, and accelerated the development of artificial intelligence with the integration of AI technology and traditional industry applications. IDC predicts that global enterprise investment in AI software, hardware, and services will exceed $85 billion in 2021 and increase to $204.5 billion in 2025, with a five-year compound growth rate (CAGR) of 24.5%.

Global technology giants will continue to increase investment in computing power in the future, and the investment scale of infrastructure in infrastructure represented by Amazon, Facebook, Google, Ali, Tencent, and Baidu will exceed half of the world's total. IDC predicts that by 2025, the world's top eight CSPs will consume more than 50 percent of their server and storage infrastructure.

Insight 2: Chinese the rapid development of intelligent industries, strengthen the construction of computing power infrastructure to support the landing of applications

Chinese intelligent infrastructure market size has maintained rapid growth, Chinese server manufacturers have become the backbone of the global server market, and Inspur ranks first in the global AI server market share. Affected by the epidemic, China's server market still maintains a high growth momentum, IDC's latest release of the "Global Artificial Intelligence Market Semi-annual Tracking Report" shows that the market size of Chinese intelligent server head manufacturers can increase by more than 50% year-on-year.

China has clearly proposed to lay out the national hub node of the computing power network throughout the country, and at the same time actively promote the construction of the "new infrastructure" of the intelligent computing center, and strive to build a modern infrastructure system and ecology that is universal, inclusive, safe and reliable. The report argues that businesses/organizations or countries with higher computing infrastructure capacity will be more likely to profit more from the dividends brought by AI.

Insight 3: The competition for artificial intelligence between cities is intensifying, and computing infrastructure is an important competitiveness

In the 2021 Chinese Smart Cities Ranking, the top five cities are Beijing, Hangzhou, Shenzhen, Nanjing, and Shanghai. Compared with 2020, Nanjing entered the top five for the first time, and Jinan entered the top ten for the first time. The competition for artificial intelligence between cities is intensifying, and computing power infrastructure is an important competitiveness. For example, Nanjing has landed the largest and most powerful intelligent computing center in the Yangtze River Delta region, providing a high-speed channel for the industrialization of urban artificial intelligence.

Insight 4: The penetration of the artificial intelligence industry accelerates and generates significant benefits

The penetration of artificial intelligence industry applications ranks first in the Internet, and the speed of artificial intelligence applications in the financial industry has accelerated and surpassed that of the government industry, ranking second. The manufacturing, transportation and energy industries are also more deeply used in artificial intelligence, ranking fifth, seventh and ninth respectively. Compared with last year, the penetration of artificial intelligence in various industries is increasing.

The landing of artificial intelligence technology has brought more value to the industry. The survey found that enterprises that have adopted artificial intelligence for more than three years have achieved significant benefits, with the average revenue of the surveyed companies increasing by 9.8%, shortening the process time by 20.4%, and increasing production efficiency by 21.6%.

Insight 5: The coordinated development of computing power and applications promotes the diversification of scenarios

Subject to development time, capital investment, computing power requirements, model maturity, technical limitations and complex needs, autonomous driving, new drug research and development, smart oilfield and other scenarios are in the early stage of development, and there is a broad space for development in the future.

Compared with the same period last year, application scenarios such as anti-fraud, intelligent risk control, intelligent investment advisory, intelligent claim settlement and intelligent recommendation have gradually entered a more mature application stage, and the degree of enterprise computing power investment is relatively high.

Application scenarios such as smart supply chain, intelligent quality inspection, intelligent equipment operation and maintenance, intelligent line inspection and power station equipment monitoring, smart insurance, and smart transportation have developed rapidly.

Insight six: The demand for artificial intelligence applications is becoming increasingly rich, giving rise to the diversified development of chips

GPUs remain the first choice for data center acceleration, with more than 90% of the market share, and other chips such as ASICs, FPGAs, and NPUs are increasingly being adopted in various fields. Other non-GPU chips such as ASICs, FPGAs, and NPUs are also increasingly being adopted in various industries and fields, with an overall market share of nearly 10%, and it is expected to account for more than 20% by 2025.

Driven by new technologies, 5G and other connection technologies reduce the transmission and processing speed of data, artificial intelligence processing on the edge side will become a key growth area of enterprises, side side of the artificial intelligence chip deployment scale and dispersion, requiring chips with the ability to adapt to various complex environments, artificial intelligence at the edge and end side will have a wide range of application scenarios, from automatic driving to industrial manufacturing, to consumer smart homes and wearable devices.

Insight seven: Artificial intelligence server has broad prospects, diversified openness, green energy saving is the development direction

The acceleration of artificial intelligence applications has promoted the high growth of AI servers. According to IDC China Accelerated Computing Market Report, the artificial intelligence acceleration server market size is expected to reach $5.69 billion in 2021, an increase of 61.6% compared with 2020, and by 2025, the Chinese intelligent acceleration server market will reach $10.86 billion.

Open computing is an innovation in the industrial collaboration model in the field of global data centers, which will greatly promote the construction and development of green and efficient data centers. Server manufacturers comprehensively lay out advanced energy-saving and consumption-reduction technologies such as cold plate liquid cooling and immersion liquid cooling, providing enterprises with stable, safe, efficient and green infrastructure and practicing carbon neutrality.

Insight 8: Artificial intelligence algorithm models show a large-scale trend, huge models will be the basis for large-scale innovation, and computing power will become the fundamental guarantee.

Since 2011, the parameters of global artificial intelligence models have increased dramatically, and have exceeded the level of 100 billion. The GPT-3 deep learning model released in 2020 has a parameter of up to 175 billion, which was the world's largest AI giant model at that time; the "source 1.0" parameter released in 2021 rose to 245.7 billion, with a parameter growth of up to 40%, which is currently the world's largest Chinese AI giant model. Building large models and improving artificial intelligence processing performance have become very popular model development trends nowadays.

The training of huge models requires the support of strong computing power, and the continuous improvement of computing power and the rapid development of cloud platforms are an important benefit to the development of "huge" models. In China, the construction of computing infrastructure such as intelligent computing centers led by local governments solves the problem of supply of computing power services by providing public computing power, data and algorithm services to make computing power services easy to use.

Insight 9: Accelerated convergence of AI and cloud drives business innovation

Chinese intelligent server public cloud will usher in rapid development. In 2020, China's artificial intelligence server market deployed in the public cloud has accounted for more than 30%, and the overall market size has exceeded 800 million US dollars, while the proportion of the global artificial intelligence server public cloud has reached 50%, and by 2025, the proportion of public cloud for Chinese intelligent servers will exceed 50%. AI public cloud service solutions focus on computer vision, intelligent speech, natural language processing and machine learning, leading AI cloud service vendors, integrating these technologies to continuously introduce new application scenarios and solutions.

Industry leaders are actively deploying private clouds to support their emerging business applications, including artificial intelligence. The trend toward hybrid IT architectures with public, private, and traditional data centers has had a significant impact on enterprise technology and business innovation.

Insight 10: Infrastructure construction and government guidance "two-pronged approach" to lead the development of the digital economy

The government has increased the construction of new artificial intelligence infrastructure, and helped traditional industries establish a systematic understanding of intelligent development through algorithm infrastructure represented by the intelligent computing center and the algorithm infrastructure represented by the huge model. The Smart Computing Center is regarded by more and more local governments as a key information infrastructure to support and lead the development of the digital economy, smart industry, smart city and smart society, and provide platform support for the development of computing power, data, ecology and industry.

Computing power has become the core productivity of the digital age and a new driving force for the development of the digital economy. The investment in AI computing power will also accelerate the integration of artificial intelligence, an important digital technology, with the real economy, empower the transformation and upgrading of traditional industries, spawn new industries and new business models, and make due contributions to Making China's digital economy stronger, better, and bigger.

Source: Guangming Network