全栈计算基础设施
Search documents
全球AI算力革命,生态之争加速演绎
Huachuang Securities· 2026-01-16 04:15
Investment Rating - The report maintains a "Recommended" investment rating for the computer industry [3] Core Insights - The global AI computing revolution is accelerating, with a significant increase in demand for intelligent computing power, projected to exceed 16 ZFlops by 2030, with intelligent computing accounting for over 90% of this demand [6][14] - NVIDIA leads the market with a nearly 90% share in AI server GPUs, while companies like Broadcom and AMD are also making significant strides in the ASIC chip market [6][19] - The competition in the AI computing ecosystem is intensifying, with a shift from general-purpose to specialized chips, driving a trend towards customized solutions [8][12] Summary by Sections Global AI Computing Revolution - The demand for intelligent computing is rapidly growing, with the global computing power expected to reach 16 ZFlops by 2030, where intelligent computing will dominate [14] - NVIDIA's GPU market share is approximately 90%, with significant growth in AI chip sales projected for the coming years [19] NVIDIA's Data Center Business - NVIDIA has built a comprehensive computing infrastructure, investing over 582 billion in R&D, leading to innovations across chips, systems, and software [49] - The introduction of the Blackwell architecture has significantly enhanced performance, supporting models with up to 100 trillion parameters [53] Broadcom's Rise - Broadcom focuses on ASIC chips, holding a 55%-60% market share in the ASIC market, establishing long-term partnerships with major cloud service providers [43] - The company's AI business revenue reached 20 billion, growing by 65% year-on-year [6] Intensifying Competition in the AI Ecosystem - The AI market is shifting towards specialized chips, with major cloud providers like Google and Amazon developing their own chips to reduce dependency on external suppliers [8][12] - AMD is enhancing its ecosystem, with plans to release new chip series that promise significant performance improvements [19] Investment Recommendations - The report suggests focusing on A-share companies such as Cambricon, Haiguang Information, and Inspur, as well as U.S. companies like NVIDIA, Broadcom, and AMD, as potential investment opportunities in the evolving AI computing landscape [6][8]