Sou Hu Cai Jing·2025-11-03 04:32Core Insights - Google is preparing to launch its next-generation AI model, Gemini 3, which is expected to be released in 2025, as confirmed by CEO Sundar Pichai during the latest earnings call [1][3] - Gemini 3 aims to surpass the current Gemini 2.5 Pro and narrow the performance gap with OpenAI's GPT-5, focusing on "agent-like" capabilities for complex, multimodal task handling [3] - The company acknowledges that while the overall pace of improvement is accelerating, significant advancements may take more time to achieve [3] Financial Performance - Alphabet's quarterly revenue is projected to exceed $100 billion for the first time in Q3 2025, marking a significant milestone for the company [3] - Monthly active users of Gemini have surpassed 650 million, with query volume doubling compared to the previous quarter [3] AI Integration and Growth - Google's search business has seen significant growth due to AI features, particularly among younger users, with AI Mode supporting 40 languages and achieving 75 million daily active users [4] - The revenue from products based on generative AI models has increased by over 200% year-on-year, with a nearly 34% increase in new cloud customers and a backlog of $155 billion, up 46% from the previous quarter [4] AI Infrastructure and Demand - There is a growing demand for AI infrastructure, with AI startup Anthropic planning to connect up to 1 million Google TPUs, indicating a strong need for computational power [5][6] - Google is set to launch a new generation of Ironwood TPUs and cloud service products based on NVIDIA GB300 chips to meet this demand [6] Expansion of Autonomous Driving - Waymo, Alphabet's autonomous driving division, is accelerating its expansion plans, aiming to enter London and Tokyo by 2026 and expanding its U.S. operational network [6] - Waymo has received permits for fully autonomous airport shuttle services in San Francisco and San Jose, with ongoing testing in New York City [6]