Nine images provide a detailed breakdown of Nvidia's latest Rubin platform, from Blackwell to Rubin, indicating that AI computing power has entered the system level era. The emergence of Rubin is essentially a paradigm shift in AI infrastructure: when model size, inference load, and data traffic burst simultaneously, the bottleneck shifts from computing power itself to bandwidth, interconnection, and system collaboration, and single-chip capabilities no longer determine the upper limit. Vera CPU is responsible for scheduling, Rubin GPU provides computing power, HBM4 solves bandwidth, NVLink connects interconnects, and ultimately integrates in cabinet level systems - Nvidia no longer sells chips, but a complete set of AI factories. This not only drives GPU performance competition, but also upgrades the entire chain of HBM, advanced packaging, networking, liquid cooling, power supply, and data centers. From the semiconductor industry report, let's learn together.