计算机行业周观点第50期：应用驱动Infra强需求-20260228

Investment Rating - The industry is rated as "Overweight," indicating an expected increase in value exceeding 10% compared to the market benchmark index over the next 6-12 months [6]. Core Insights - The AI Agent concept is transitioning from theory to practical application, with OpenClaw driving demand for autonomous execution. Token consumption is shifting from "dialogue consumption" to "task consumption," leading to rapid growth in usage [1]. - The release of Seedance 2.0 by ByteDance has highlighted significant demand for computational power, as users report long wait times for video generation due to high traffic and processing demands [2]. - The rapid development of AI applications is driving a surge in token consumption, which will necessitate systematic upgrades and expansions across the entire computational infrastructure, including chips, servers, and data centers [3]. Summary by Sections AI Agent Development - AI Agents are now capable of executing real tasks, with OpenClaw demonstrating a significant increase in token consumption through various operational modes, including multi-round self-correction and context expansion [1]. - The token consumption for active sessions can exceed 230,000 tokens, indicating a substantial increase in operational demands [1]. Seedance 2.0 Launch - Seedance 2.0 utilizes a sparse architecture to enhance training and inference efficiency, achieving industry-leading performance in generating high-quality audio-visual content [2]. - Despite its capabilities, the model faces significant computational bottlenecks, leading to user complaints about extended wait times for video generation [2]. Infrastructure Demand - The growth in AI applications is expected to drive a comprehensive upgrade of the computational infrastructure, highlighting the need for enhanced supply of computational resources [3].