ZHONGTAI SECURITIES·2025-11-06 12:47Core Insights - The report highlights Google's comprehensive AI stack advantage, which includes chips (TPU), networks (OCS), models (Gemini), and applications (cloud computing/search/advertising), indicating a significant potential for capital expenditure increase [3] - Google's capital expenditure for 2025 has been raised to between 91 billion and 93 billion, with expectations for continued growth in 2026, reflecting a strong demand for AI capabilities and hardware innovation [3] - The Gemini model has surpassed 650 million monthly active users, with token processing volume increasing over 20 times within a year, showcasing robust growth in Google's cloud revenue and order backlog driven by AI [3] Summary by Sections AI Data Center and OCS Technology - OCS (Optical Circuit Switching) technology is crucial for AI data centers, providing advantages such as reduced latency and energy consumption by directly transmitting data using optical signals, which is essential for large-scale AI computing needs [4] - Traditional data center architectures face challenges in power consumption and cost, while OCS offers a key solution for dynamic computing demands through server decoupling and efficient long-distance signal transmission [4] Google TPU and OCS Market Outlook - Google's TPU shipments are expected to increase, with the introduction of OCS technology enhancing network throughput by 30%, reducing power consumption by 40%, and significantly decreasing network downtime [5] - The upcoming TPU V7 Ironwood will feature 1.6T optical modules, with various suppliers poised to benefit from the OCS technology, which includes MEMS and liquid crystal solutions [5] Investment Opportunities - The report suggests focusing on investment opportunities within Google's AI computing supply chain and OCS technology innovations, highlighting key companies such as Zhongji Xuchuang, Changxin Bochuang, and Tengjing Technology as potential beneficiaries [6]