Market research firm Gartner released its "Gartner China AI 25" report. The report predicts that by 2030, China's domestic semiconductor industry will provide more than 50% of the semiconductor products purchased by Chinese OEMs; by 2030, more than 30% of the revenue of Chinese semiconductor companies will come from sales outside mainland China; and by then, more than 80% of companies will adopt physical AI in all aspects of design, manufacturing, products, and services, compared to less than 1% currently.
The "Gartner China AI 25" report selected 25 companies from 1,000 listed companies by market capitalization in China, belonging to the sectors of autonomous mobility, drug development, consumer intelligence, smart energy, and industrial intelligence. These include industry leaders such as Xiaomi, BYD, BOE, TCL, Haier, and CATL. The report found that these companies are reshaping their business boundaries through AI, and their practices exhibit four common characteristics.
First, at the strategic level, all 25 companies are driving the implementation of their AI strategies with leading visions. Gartner categorizes AI into two types: everyday AI, which uses AI in daily operations, such as employees using AI to write code and improve productivity; and disruptive AI, which uses AI to revolutionize product services and R&D processes. These 25 companies all focus on disruptive AI innovation. They no longer view AI as a simple efficiency tool, but as a core driver reshaping products, services, and R&D processes.
Secondly, in terms of operational models, these 25 companies exhibit three main characteristics: AI-first organizations, closed-loop data feedback, and platform-based scaling.
An AI-first organization is a strategic approach that recognizes the transformative potential of AI and emphasizes fully considering this potential in all aspects of corporate initiatives, rather than simply viewing it as a tool. The implementation of this strategy also demonstrates the company's tolerance and inclusiveness in embracing AI. For example, Haier launched its AI Year 2025, proposing a strategic direction of "embracing AI with all employees, comprehensively, and throughout all processes," allowing AI to "flow like blood through every aspect of operations."
A closed-loop data feedback system is an operational principle that helps systems continuously learn and optimize. Enterprises no longer view data as a static resource; in AI applications, new situations, new environments, and new data can lead to new strategies. For enterprises, data is not necessarily just high-quality data. With the participation of intelligent agents, a combination of "wrong" and "correct" data may be more important than simply having correct data. Giving an agent incorrect data and then requiring it to correct the data and learn the correct strategy can be more helpful to the business. Fang Qi, Research Director at Gartner, stated that enterprises should not only focus on high-quality data during data collection but should also seek representative data.
Platform-based scaling is a growth strategy that expands the reach and impact of AI solutions across the entire enterprise and even the ecosystem by building a platform architecture. The actual effectiveness of AI implementation depends not only on model capabilities but also on engineering capabilities. In other words, sometimes a model performs well under normal lighting conditions, but in the enterprise engineering process, problems such as insufficient lighting in the workshop may affect its performance. To solve these problems, enterprises need to continuously increase the accuracy of the system and enhance the generalization ability of the model. The goal of platform capabilities is to dilute and reuse the huge initial investment costs of AI, applying it to the processes of more enterprises or the entire ecosystem.
Tags：