Investment Rating - The report provides a "Buy" rating for the industry, indicating an expectation of stock performance exceeding the market by more than 10% over the next 12 months [37]. Core Insights - The report emphasizes the significant advancements in AI technology, particularly through the introduction of the NVIDIA Vera Rubin POD platform, which is designed to enhance performance for Agentic AI workloads [4][12]. - It highlights the collaboration within the AI industry chain, suggesting that model innovation and capital expenditure (CAPEX) are foundational for growth, with a continuous upward trend in AI-driven storage cycles [31]. - The report suggests focusing on key beneficiaries within the industry chain, as the demand for AI inference drives both production expansion and upgrades [31]. Summary by Sections Section 1: NVIDIA Vera Rubin POD Platform - NVIDIA introduced the Vera Rubin POD platform, which includes five rack-level systems tailored for Agentic AI workloads, enhancing throughput and reducing latency [4][12]. - The platform consists of various racks, including the MGX NVL rack for GPU computing and the MGX ETL rack for CPU and storage, showcasing a heterogeneous collaborative system architecture [4][12]. Section 2: Groq3 LPX Rack for Decoding Acceleration - The Groq3 LPX rack is designed to accelerate decoding processes, integrating 256 LPU processors with high bandwidth capabilities, significantly improving performance metrics [4][17]. Section 3: Vera CPU Rack for RL/Agent Sandbox Support - The Vera CPU rack supports over 22,500 concurrent reinforcement learning (RL) or agent sandbox environments, facilitating testing and validation of outputs from the Vera Rubin NVL72 and LPX systems [4][28]. Section 4: Investment Recommendations - The report recommends attention to core beneficiaries in the AI industry chain, driven by model innovation and collaborative development, with a focus on the upward trend in AI inference and storage cycles [31].
AI的进击时刻25:英伟达发布 Vera Rubin POD 平台