According to a research report by Huatai Securities, the AI industry is projected to shift from large-scale model pre-training to the commercialization phase of AI Agents by 2026, leading to a significant surge in demand for inference computing capabilities. The report identifies three key trends:

Firstly, within the AI supply chain, China is accelerating the development of a closed-loop domestic computing power ecosystem, accompanied by advancements in super-node interconnection and storage enhancements. The turning point for inference demand has arrived, and the advent of new product cycles, such as AI-integrated foldable smartphones and AI-powered glasses, presents substantial opportunities for structural innovation.

Secondly, in the realm of power and passive components, the rising power consumption of AI systems is driving concurrent increases in both volume and pricing for MLCCs (Multilayer Ceramic Chip Capacitors), inductors, capacitors, and power semiconductors. This price surge aligns with ongoing efforts to promote domestic substitution in these components.

Thirdly, concerning independence and controllability, the localization process for upstream manufacturing, equipment, and components is gathering momentum, resulting in a systematic increase in the value of advanced packaging solutions.