On June 26th, the World Mobile Congress Shanghai started. XU Ziyang, the CEO of ZTE, mentioned in his speech that the world has entered the stage of AI-driven industrial revolution. However, besides illusions, security and ethical issues, generative AI development is also facing challenges such as computing power, energy consumption, dataset, standardization, and commercial application. The application of large models in various industries also needs to solve a series of rigid challenges such as professionalism, accuracy, robustness, and traceability. At the same time, professional data governance, incremental training, and toolkits are also essential when building domain models based on large models. For example, in the telecom self-intelligence network scenario, the combination of large and small models, RAG (Retrieval Augmented Generation), multi-agent collaboration, digital twins, and multi-modal CoT (Chain-of-Thought) and other technological innovations have already achieved certain results. In the journey of commercializing AI applications, training and promotion are the key drivers of real economic takeoff. With abundant application scenarios and private domain data resources, China can make significant contributions to the global AI industry.