Information source:
https://www.scmp.com/tech/tech-war/article/3324931/china-narrows-ai-development-gap-us-remains-behind-advanced-chips-clsa?module=top_story&pgtype=section
China's catching up speed in the field of artificial intelligence is exceeding external expectations. CITIC Lyon Securities' latest research shows that the gap between China and the United States in the development of AI technology has narrowed sharply from a significant lag more than a year ago to only about three months, and this dramatic change is reshaping the global technological competition landscape. However, in the field of more basic advanced chip manufacturing, China is still facing years of technological generation gap, highlighting the complexity and multi-level characteristics of this technology competition.
Zhang Tao, head of China Science and Technology Research at Lyon Securities, pointed out in his latest analysis that the rapid rise of China's AI industry is due to the synergy of multiple factors: the influx of a large number of top talents, the acceleration of technology iteration, and the explosive growth of AI application scenarios. This evaluation result is in sharp contrast with the industry's previous general expectations, and also provides an important reference for understanding the current competitive trend of China-US technology.
Despite the strict export restrictions on high-performance chips from the United States to China, Chinese AI companies are still making breakthroughs in multiple dimensions such as software algorithms, application innovation and system optimization, demonstrating strong adaptability under the pressure of technological blockade. This development model not only changes the competitive landscape of the global AI industry, but also has a profound impact on international science and technology governance and industrial policies.
Local advantages of AI application ecosystem
The rapid development of China's AI industry is largely due to its unique application ecological advantages. Unlike the United States that mainly focuses on general artificial intelligence and basic model research and development, Chinese companies pay more attention to the implementation and application of AI technology in specific scenarios, forming a comprehensive coverage from consumer Internet to industrial manufacturing.
Internet giants such as Baidu, Alibaba, and Tencent are making significant achievements in the field of large language models. Products such as Baidu's Wenxin Yiyan, Alibaba's Tongyi Qianwen, Tencent's Hunyuan Big Model have performed close to or even surpassed international advanced models such as GPT-4 in some tasks. Evaluation data from the Tsinghua University Institute of Artificial Intelligence shows that the gap between these Chinese models and top international products in core indicators such as language comprehension, text generation and knowledge questions and answers is rapidly narrowing.
More importantly, Chinese AI companies have shown obvious advantages in vertical applications. In the field of autonomous driving, the technical level of Baidu Apollo, Xiaopeng Motors, NIO and other companies has reached the international advanced level. Especially in the technical problem of autonomous driving on urban roads, Chinese companies are even ahead of international giants such as Tesla in some aspects with their rich road scenario data and intensive testing environment.
The same is true in the field of financial technology. AI risk control systems, smart investment advisors and digital payment technologies developed by companies such as Ant Group and Tencent Financial Technology are leading the way in the world. Data from the Digital Currency Research Institute of the People's Bank of China shows that digital RMB systems based on AI technology have surpassed traditional financial infrastructure in terms of transaction processing speed, security and user experience.
Accelerating appearance of talent aggregation effect
The rapid development of China's AI industry cannot be separated from the support of talent advantages. In recent years, a large number of top overseas AI talents have returned to China. At the same time, domestic universities and research institutions are also accelerating the cultivation of local AI experts, forming a strong talent aggregation effect.
The latest statistics from the Ministry of Education show that Chinese universities train more than 300,000 graduates of AI-related majors every year, of which the proportion of doctoral and master's students continues to increase. Top universities such as Tsinghua University, Peking University, and the Chinese Academy of Sciences have reached the world's first-class standards in core fields such as machine learning, deep learning, and computer vision.
What is more worthy of attention is that the engineering capabilities and industrialization experience of China's AI talents are accumulating rapidly. Unlike the United States that focuses on theoretical research, Chinese AI practitioners pay more attention to the practical application of technology and commercialization. This practically oriented talent training model provides strong support for the rapid development of China's AI industry.
The talent battle between large technology companies has further promoted the development of the industry. The salary and benefits offered by emerging technology companies such as ByteDance, Meituan, and Didi to attract AI talents are comparable to those of Silicon Valley technology giants, forming a strong talent siphon effect. This not only accelerates technological innovation, but also promotes the widespread application of AI technology in various industries.
Structural Challenges in Chip Manufacturing
Despite significant progress in AI software and application levels, China still faces major challenges in the field of underlying hardware, especially advanced chip manufacturing. The chip export restrictions imposed by the United States on China cover the complete industrial chain from design software to manufacturing equipment, and have had a substantial impact on China's semiconductor industry.
The current state-of-the-art processes of leading chip manufacturers in China, such as SMIC and Huahong Group, are still at the 14-nanometer to 7-nanometer level, which is a clear gap with international advanced manufacturers such as TSMC and Samsung in the 3-nanometer and 5-nanometer processes. This technological generation difference directly affects the performance of China's AI chips and also restricts the development speed of related industries.
The export ban on EUV extreme ultraviolet lithography machines by the Dutch ASML company is a key factor restricting the development of China's chip manufacturing industry. These hundreds of millions of dollars of precision equipment are indispensable for the production of advanced process chips below 7 nanometers, and Chinese companies currently do not have access to these critical equipment.
However, China's semiconductor industry is also seeking alternative development paths. Although Huawei HiSilicon's Kirin 9000S chip adopts relatively mature process technology, it achieves performance close to advanced processes through system-level optimization and architectural innovation. This "curve-saving the country" technical path provides new ideas for the development of China's chip industry.
Differential competition in the technology ecology
The development of the AI industry in China and the United States shows obvious differentiated characteristics. American companies are more focused on basic research and the construction of general technology platforms, while Chinese companies are better at technology applications and business model innovation in specific scenarios.
OpenAI's ChatGPT, Google's Bard, Microsoft's Copilot and other products represent the United States' technical advantages in the field of general artificial intelligence, which perform well in multilingual processing, cross-domain reasoning and creative generation. In contrast, China's AI products focus more on specific vertical fields, such as e-commerce recommendations, content reviews, smart customer service and other practical scenarios.
This differentiated development model actually provides different competitive advantages for the AI industries of the two countries. The leading position of American companies in basic technologies and general platforms will be difficult to shake in the short term, but the advantages of Chinese companies in application innovation and scenario-based solutions are equally important.
A research report by IDC of international data company IDC shows that China's AI application market has surpassed that of the United States, especially in terms of AI penetration in traditional industries such as retail, finance, manufacturing and transportation. This application-level advantage is transforming into a driving force for technological innovation, promoting China's AI industry to a higher level.
In the long run, the competition between China and the United States will be more reflected in the construction of ecosystems and standard formulation. China's "East Digital and West Calculation" project and the Artificial Intelligence Industry Development Alliance aim to build a complete AI industry ecosystem, and the United States is also strengthening its leadership in the field of AI through policy tools such as the "Chip and Science Act". The final result of this competition will largely determine the future landscape of the global AI industry.
