The emergence of DeepSeek serves as a vivid case for reevaluating and understanding China’s distinctive approach to disruptive innovation. Photo: IC PHOTO

---

During the 2025 Spring Festival, the generative AI model DeepSeek, developed by the Chinese AI startup of the same name, made a striking debut, sending shockwaves throughout the global tech community. Its extraordinary performance, delivered at a fraction of the cost of its competitors, has continued to ignite discussions worldwide. The emergence of this homegrown large language model serves as a vivid case for reevaluating and understanding China’s distinctive approach to disruptive innovation. In interviews with CSST, experts offered insightful explorations of the “key” to China’s AI-driven innovation, examining it from the perspectives of technological progress, industrial transformation, and shifting innovation models.

First, policy-friendly environment and institutional innovation have worked together to provide strong backing for disruptive innovation in the field of AI. The New Generation Artificial Intelligence Development Plan, launched in 2017, established a clear direction for growth, with subsequent policies further advancing industrial progress and providing solid institutional guarantees for corporate innovation.

Liu Huajun, a professor of economics at Shandong University of Finance and Economics, explained that local governments have facilitated the synergistic evolution of institutions and technology by reducing the institutional transaction costs of innovation activities through experiments such as establishing markets for AI compute and developing mechanisms to verify data ownership. This dynamic partnership between national policy guidance and local institutional innovation has cultivated a thriving environment for AI breakthroughs, ensuring the continued advancement of China’s AI landscape.

Second, an open-source strategy has become one of the defining characteristics of China’s AI innovation. Jiang Qiping, a research fellow from the Institute of Quantitative and Technological Economics (IQTE) at the Chinese Academy of Social Sciences, noted that the open-source approach adopted by DeepSeek has significantly accelerated the dissemination and application of innovative breakthroughs, engaging global developers to collaborate and thus advancing the development of the AI ecosystem.

This success stems from China’s institutional advantage of encouraging shared development, Jiang explained. The integration of open-source sharing with commercial value creation has injected new vitality into China’s AI sector. Despite resource constraints, Chinese enterprises have leveraged algorithmic optimization and engineering design to achieve low-cost training with high-performance output. This reflects China’s pragmatic approach to AI development, focused on delivering practical solutions that drive efficiency and cost savings across industries.

Liu’s concept of “demand-driven innovation” further elucidates this phenomenon. By identifying and addressing specific “pain points” in vertical fields, companies adopt a “scenario-defining technology” strategy, propelling innovation in a reverse-engineered fashion. This approach enables products and services to be better aligned with market demands, boosting the practical utility and market acceptance of new innovations.

Third, China’s emphasis on fundamental research and talent cultivation has laid a strong foundation for the rapid advancement of AI. Peng Xushu, a research fellow from the IQTE, observed that China’s higher education reforms have nurtured a large pool of STEM talent, providing essential intellectual support for the thriving AI sector.

Liu also stressed the crucial role of basic research in the innovation process, introducing the concept “dynamic aggregation effect” of innovation elements. This effect facilitates the synergistic interaction of knowledge, human, and financial capital, creating a multiplier effect that pushes enterprise innovation to new heights.

Edited by CHEN MIRONG