海外AI逐步闭环，继续看好AI基建带来高增长

Core Insights - Nvidia has announced a strategic partnership with OpenAI, committing to invest up to $100 billion in infrastructure, including the deployment of at least 10GW of Nvidia systems, which will consist of millions of GPUs [1][2][3] Investment Details - OpenAI will deploy a minimum of 10GW of Nvidia systems, which translates to approximately 10 million kilowatts, and Nvidia will continuously invest as the computing power comes online [1][2] - The total investment from Nvidia could reach $100 billion (approximately 711.47 billion RMB) [1][2] Infrastructure Development - The first GW-level data center based on Nvidia's Vera Rubin platform is set to be deployed in the second half of 2026 [1][2] - The estimated cost to build 1GW of computing power is around $50-60 billion, with Nvidia's chips and systems accounting for about $35 billion of that cost [3] Market Implications - The investment scale for the 10GW data center project is estimated to be between $500 billion and $600 billion, similar to Nvidia's previously announced Stargate project [3] - OpenAI's CEO, Sam Altman, emphasized that this infrastructure is crucial for achieving OpenAI's goals and delivering services to users [3] AI Industry Trends - The AI landscape is shifting from model competition to a focus on infrastructure expansion, with significant investments being made to overcome computing power bottlenecks [4] - OpenAI has signed a $300 billion five-year contract with Oracle and plans to spend $100 billion on backup servers from cloud service providers over the next five years, in addition to Nvidia's investment [4] Investment Recommendations - The current focus in AI infrastructure is shifting from cloud vendor spending to AIDC cloud computing centers, with companies like Oracle, Coreweave, and others being highlighted for their GPU card advantages [4] - In the domestic market, attention remains on AI infrastructure, with recommendations for companies like Runze Technology and Aofei Data [5]