浙商证券：AI端侧落地正当时 硬件产品蓄势待发

Core Insights - 2025 is expected to be a significant year for AIGC applications, driven by substantial investments in generative AI infrastructure, although revenue generation remains weak [1] - The development of AI edge hardware is anticipated to accelerate the integration of AIGC technology into various scenarios, enhancing technology adoption [2] - The AIGC is likely to reshape the terminal product market, particularly in mobile phones and PCs, as AI functionalities become a key differentiator [3] Group 1: AI Infrastructure and Market Dynamics - Significant capital expenditure is required for AI infrastructure, with companies needing to find ways to generate returns on these investments [1] - The emergence of various AI products in the market indicates a growing sales scale and continuous iteration after initial explorations [1] - The reduction in model training costs through innovations like the DeepSeek model will facilitate the entry of more non-large AI companies into the market [1] Group 2: AI Edge Hardware Development - AI edge hardware is expected to see rapid growth, serving as a user entry point and enabling personalized AI applications [2] - The introduction of products such as AIPC, AI smartphones, and AI wearables has laid a foundation for market maturity, with significant sales achievements noted in smart glasses [2] - The ongoing investment in AI infrastructure and the reduction in training costs will further catalyze product scaling [2] Group 3: Market Growth and Restructuring - The integration of AI functionalities in mobile phones and PCs is likely to alter market dynamics, affecting total sales, penetration rates, and component changes [3] - The AI capabilities in smart glasses and headphones are expected to expand market applications and increase sales volumes, benefiting both brand manufacturers and component suppliers [3] - AI companion products are in the early stages of commercialization, with significant demand potential driven by advancements in large models and reduced usage costs [4]