智能体基建厚积薄发，商业化应用曙光乍现

Investment Rating - The report does not explicitly state an investment rating for the industry Core Insights - The AI development stage is transitioning from reasoning models to intelligent agents, with improvements in model capabilities and the emergence of commercial applications [4][6][29] - The infrastructure for AI agents is rapidly evolving, with a focus on enhancing developer ecosystems and commercial viability [4][22][25] - Initial products in the AI agent space are generating significant revenue, indicating a promising outlook for commercialization [4][24][29] Summary by Sections 1. AI Development Stage - AI is evolving from reasoning models to intelligent agents, which can understand goals and utilize external memory and reasoning capabilities [4][6] - The current AI product landscape is characterized by a shift from training expansion to testing expansion, focusing on commercial capabilities [4][8] 2. Agent Model Layer - The foundational intelligence level of AI models is improving, with reasoning capabilities becoming central to their functionality [4][12] - The report outlines a four-core capability framework for agents: reasoning, memory, planning, and tool usage [12][15] 3. Agent Intermediate Layer - The development of middleware tools and protocols is accelerating, with significant growth in developer ecosystems [4][22][72] - Notable protocols like MCP and A2A are being established to facilitate communication and collaboration among AI agents [69][74] 4. Agent Application Layer - Initial AI agent products are achieving substantial annual recurring revenue, showcasing their potential for widespread application [4][24] - The report highlights the emergence of new pricing models based on performance metrics, indicating a shift towards more sophisticated commercial strategies [4][24] 5. Related Companies - Key players in the AI agent ecosystem include Nvidia, Broadcom, Google, Snowflake, Salesforce, SAP, Shopify, Amazon, Microsoft, and Google [4][24]