Yang Zongkai, a core strategist behind the Ministry of Education’s digital transformation, recently revealed that China’s "15th Five-Year Plan" for education technology will pivot around three technological pillars: standardized construction of the underlying computing power platform, independent and controllable development of vertical education LLMs, and upgraded governance capabilities for full-chain data. The National Education Big Data Center will serve as a key computing power node, achieving a fully localized technology stack from the chip layer to the application layer—a move widely seen as a critical countermeasure in the current international AI computing power competition.\n\nOn the AI model strategy front, the authorities have for the first time proposed a dedicated "Education Large Language Model" R&D program, emphasizing the need to break through the limitations of existing Transformer architectures in personalized learning scenarios.
Notably, the plan specifically requires the establishment of a testing and certification system for education-specific NPUs (neural processing units), reflecting policymakers’ increased focus on underlying AI chip technologies.\n\nIn terms of intelligent deployment, the document highlights the concept of "scenario-specific computing power," requiring dynamic allocation of computational resources based on different scenarios such as teaching, assessment, and management. This elastic deployment model based on edge computing is likely to spawn new procurement standards for education-dedicated AI servers. On the international expansion front, China is exporting its proprietary AI standards through smart education platforms. The Kunpeng + Ascend computing power combination it adopts is already proving competitive against NVIDIA’s solutions in education cloud projects across several Southeast Asian countries.