Amazon(US:AMZN) 36氪·2025-12-11 10:00Core Insights - The article discusses the emergence of AI Agents and the current state of AI infrastructure, highlighting the gap between the rapid development of AI Agents and the readiness of the underlying infrastructure to support them [3][5][9]. Group 1: AI Agent Development - The AI Agent era is recognized as having arrived, with significant announcements from Amazon Web Services (AWS) regarding AI infrastructure and management [5]. - There is a notable increase in interest and investment in AI Agents, with many developers and companies focusing on this area during major events like re:Invent [5][6]. - However, there is a contrasting sentiment among developers regarding the current capabilities of AI infrastructure, which is perceived as inadequate to support the demands of AI Agents [9]. Group 2: Infrastructure Challenges - Developers express concerns about the current state of AI infrastructure, citing weaknesses in cost management and AI-first capabilities [9][11]. - The high costs associated with AI model inference are a significant barrier, with estimates indicating that 80-90% of AI Agent costs are tied to inference [11]. - There is a call for a software revolution to better accommodate AI Agents, including the need for simpler interaction interfaces and the elimination of data silos [13][14]. Group 3: Investment Trends - A new wave of investment in AI infrastructure is emerging, with companies focusing on optimizing AI infrastructure to reduce inference costs [15]. - Major players like NVIDIA are making significant investments in AI infrastructure startups, indicating a trend towards enhancing the foundational technologies that support AI Agents [15]. - Database companies are also recognizing the importance of adapting their products to better interact with AI Agents, emphasizing the need for scalable solutions to meet the growing demand [15].