By Xia Ri
In late 2022, ChatGPT made its appearance, signaling the start of a global acceleration in AI development. By the end of 2024, the number of large-scale AI models worldwide had surpassed 1,300, with China and the United States collectively holding 80% of the market share.
The U.S., capitalizing on its technological strength and innovative capacity, remains at the forefront of large AI model development. In contrast, China has swiftly emerged as a formidable competitor, driven by its vast market and proactive government support, with more than 300 domestically developed large AI models—second only to the U.S. What, then, is the future trajectory of AI development in China? Drawing on extensive long-term research, a senior analyst at ANBOUND offered a multi-dimensional analysis of the emerging trends in China’s AI landscape.
1. A series of “small AI” models similar to DeepSeek will continue to emerge. According to ANBOUND’s definitions, “large AI” refers to general-purpose large models, while “small AI” refers to smaller models designed for specific application scenarios. Currently, the development of “large AI” in China faces significant challenges, including enormous costs related to computing power, energy, and finances, as well as security concerns such as data leaks. In contrast, China holds local advantages in developing “small AI” models: certain industries and sectors have a strong market foundation and application scenarios, there are considerable strengths in engineering implementation capabilities, and the policy environment is highly supportive in this. Given the difficulties facing “large AI” development and the advantages of “small AI”, a wave of “small AI” models, like DeepSeek, will continue to emerge, becoming new competitors.
2. Truly original AI large models in China have yet to make a breakthrough. Large model development in some other countries started earlier and is clearly ahead in terms of parameter model development and application. In contrast, China’s development of large models began later and faces two main challenges: On the one hand, foreign technological monopolies will force Chinese AI to reveal its limitations. Currently, so-called Chinese large models, which in fact, are “small AI” models, are primarily based on extensively packaged and assembled models. If foreign advanced open-source models shut down their interfaces, China’s “small AI” models would inevitably be exposed for what they really are. Even without such a move, China would still face challenges related to computing power restrictions. With the U.S. continuing to ramp up export controls, chip supplies in China will eventually face problems. On the other hand, the unique information environment in China hinders the growth and upgrading of “large AI.” Due to the firewall system in the country, model training and learning are forced to be limited to information within a designated range. As a result, it will be difficult for China’s “large AI” to make breakthroughs in the short term.
3. The use of AI is set to become a trend, particularly in the training industry and some low-risk sectors. With breakthroughs in modern technologies such as deep learning and natural language processing, AI is gradually penetrating various industries and fields, including education, finance, and employment. Regardless of one’s level of education, those who cannot use AI will be seen as falling behind. However, in the future, AI usage will likely be concentrated in the training industry and certain low-risk sectors. In the field of training, AI offers distinct advantages: vast knowledge reserves that can be instantly retrieved and presented; the ability to accurately assess learners’ situations and create personalized learning plans; and the ability to build highly realistic scenario models for practical environments. Additionally, while AI is still in its early stages, it has made significant progress in areas such as formal logic. However, due to its technological limitations, including issues with complex logical reasoning, data leaks, and inconsistent quality, AI’s use will primarily be concentrated in low-risk industries.
4. The high-risk sector for AI applications is healthcare, as individuals with no medical training diagnosing themselves could lead to serious problems. Currently, China’s healthcare industry is accelerating the development and application of AI. However, due to the seriousness, complexity, and low tolerance for error in medical scenarios, the AI healthcare sector is fraught with risks. The most immediate challenge is the limited and often poor-quality medical data used to train AI doctors. Even if data issues gradually improve in the future, risks in AI healthcare will remain, such as data breaches, privacy violations, and legal and ethical concerns. Of particular note is the accuracy of AI-generated prescriptions. Since most patients lack professional medical knowledge, they are unable to describe their symptoms as accurately or comprehensively as a doctor would. If the patient’s description is insufficient or inaccurate, the AI’s response could be significantly compromised. If patients follow non-optimal or even incorrect answers, it could hinder their treatment or, in extreme cases, pose a threat to their lives.
5. Apps like Baidu face a survival crisis and may eventually have to rely on AI. As AI becomes more widespread, search apps like Baidu and Google have begun experimenting with AI-powered search. In the traditional search experience, users are accustomed to searching with keywords and matching information through hyperlinks, then reading, filtering, and organizing the answers they need. In the AI search experience, an increasing number of users are choosing to communicate with AI using natural language, allowing AI to help filter information and organize answers. This shift weakens the competitive advantage that traditional search engines have built over years in terms of matching efficiency. Currently, traditional search apps, after introducing AI search, are seeing a gradual decline in their original business. This suggests that the future of traditional search engines may evolve into AI assistants, ultimately becoming dependent on AI.
6. The rise of AI computers is expected to lead to the decline of traditional “dumb computers”. In February 2024, ANBOUND mentioned the concept of AI computers, which would replace traditional keyboards with natural language, create competition between Microsoft’s Windows and Apple’s OS, and pose a significant challenge to conventional computing. Hardware designs are likely to resemble today’s iPads, AI computers will break down language barriers, software tools will become more conversational, and AI will improve work efficiency. This was exemplified by the launch of Microsoft’s Copilot+PC on May 24, 2024. However, due to increasing technological restrictions and limited innovation, progress in China’s AI computers has been slow. Currently, the focus remains on imitation and using low prices to gain market presence. As a result, once AI computers become mainstream, the market for traditional “dumb computers” (i.e., non-smart laptops) is likely to shrink, with this shift expected to occur between 2026 and 2028.
7. As AI becomes more widely used, information analysis will gain mainstream influence. In the age of information overload, while AI can execute various tasks through programs and algorithms, quickly gather data, and provide answers, it still cannot think or ask questions in the same way humans do. AI serves as humanity’s “answer factory”, but the ability to pose questions remains a “core asset” of human beings. However, due to limitations in knowledge structure, many people find it difficult to ask deep, precise, and meaningful questions. In contrast, information analysis will become increasingly influential. It not only makes full use of AI as an auxiliary tool to quickly gather vast amounts of data from various fields and domains, but also applies basic information analysis models to solve practical problems and predict future trends. As AI technology advances and knowledge becomes less critical, the importance of information analysis will only grow.
8. AI may experience one more wave of development, but it will eventually enter a prolonged stagnation period. Looking back at the development of AI in China, after the emergence of a vast number of models during the early chaotic growth phase in 2023, the industry underwent a major shakeout in 2024. However, with support from government policies and initiatives like DeepSeek, China’s AI sector is expected to experience another wave of growth. For instance, the “AI+” initiative was included for the first time in the 2024 Government Work Report, signaling a surge in AI-related policies nationwide. That said, there are concerns that after this wave, China’s AI sector will enter a lengthy period of stagnation. On one hand, the high costs of computing power, electricity, and labor are prohibitive, especially for small and medium-sized companies. Additionally, many Chinese AI companies are only just managing to secure enough funding to sustain operations. On the other hand, the industry is entering a phase of intense competition. During the market’s explosive growth window, the players with the resources and capabilities have already established their positions, and future competition will likely revolve around price. Even if there are new entrants, the number of newcomers is expected to be limited.
9. AI-driven shopping optimization is intensifying competition among e-commerce platforms. In the past, when shopping online, many consumers faced information gaps that led to indecision, time-wasting, and difficulty in making the best choices. Today and in the future, AI can help us make smarter shopping decisions. For instance, some e-commerce platforms use AI recommendation systems to suggest products based on one’s shopping history, browsing patterns, and feedback. Additionally, AI can assist in verifying the authenticity of products, checking the credibility of sellers, and reviewing feedback. As Chinese platforms like JD.com and Pinduoduo actively integrate AI, the efficiency of the shopping experience will significantly improve, which will increase competition. On one hand, there is a technology-driven upgrade in user experience, with AI analyzing user behavior and needs to provide accurate product recommendations. On the other hand, operational efficiency and cost optimization are achieved, such as through AI automation in areas like product listing, customer service, and data analysis. Furthermore, AI on the client side can pose challenges for e-commerce platforms, as customers can use AI to make decisions and optimize purchasing behaviors, reducing impulse buying. Ultimately, the future success of e-commerce platforms will be heavily influenced by their AI capabilities.
10. Similar to its impact on consumption, AI will gradually transform society. As AI continues to develop and be applied, much like its integration into the consumer sector, it will make people smarter and enable them to make more informed decisions. At the same time, some fundamental tasks, particularly those that are repetitive, routine, and highly standardized, as well as jobs involving data and information processing, will be replaced. This transformation is not a sudden disruption, but rather a gradual process, much like water permeating soil. On this foundation, AI will evolve from being a tool to becoming a core infrastructure, prompting society to rethink the development of various industries and sectors. It will progressively influence people’s production and lifestyles, ultimately transforming society and reshaping the trajectory of human civilization’s evolution.
AI, as one of the most transformative technologies of the 21st century, will have a widespread and profound impact on society. This article has conducted a trend analysis from various perspectives, including technology, risks, industries, and information, presenting ten key judgments to provide a better reference for the future development of AI in China.
Final analysis conclusion:
Since the launch of ChatGPT at the end of 2022, China’s AI large models have experienced rapid growth. Moving forward, the development of AI in China is expected to follow ten major trends: First, a growing number of “small AI” models, similar to DeepSeek, will emerge. However, technological monopolies and other challenges may make it difficult for “big AI” to break through in the short term. AI usage will be concentrated in training and low-risk industries, driven by its popularity. Meanwhile, healthcare will become a high-risk sector as AI continues to be integrated. Search apps, like Baidu, are likely to evolve into AI assistants. The rise of AI-powered computers will lead to the decline of traditional “dumb computers”. Information analysis, empowered by AI, will become a mainstream tool for decision-making. After an additional wave of development, AI may enter a prolonged stagnation period. E-commerce platforms will face intensified competition as AI optimizes shopping decisions. Finally, much like its impact on consumer behavior, AI will gradually transform society. These trends reflect both the immense potential and the challenges that will shape the future of AI in China.
Xia Ri is an Industry Researcher at ANBOUND, an independent think tank.