CloudMatrix 384 super-node uses 6912 x 400G OSFP silicon photonic (SiPh) Linear Drive Pluggable Optics (LPO) optical modules and 3168 fibers, connecting 384 Ascend 910C computing chips through a complete mesh interconnection architecture. Large AI models involve both scale-up and scale-out networks, and AI computing supernode technology aims to integrate hundreds of GPUs/NPUs to form high-density computing units, such as the NVIDIA GB200 NVL72, Huawei CM384, and ODCC ETH-X. As the number of GPUs multiplies, bandwidth demands exceed TB/s, and rack power density climbs to over 40kW. Based on Xizhi Technology's world-first distributed optical switching technology, this supernode utilizes silicon photonics-based optical interconnect and optical switching chips, BiRen Technology's proprietary high-computing general-purpose GPU liquid cooling module, and a new carrier board. This article analyzes Panjiu AL128 supernode AI servers and their interconnect architecture, explaining what supernodes are, how GPUs connect, and how they advance AI computing. At the Apsara Conference 2025, Alibaba Cloud unveiled its new-generation Panjiu AI Infra 2. 2% failures stem from optics & how QSFPTEK cuts costs by 69. On May 14, 2025, the "2025 Chip and Optical Forum" hosted by HiSilicon and organized by. While the industry-standard OSFP (Octal Small Form-Factor Pluggable) module has successfully enabled 400Gbps, 800Gbps, and 1. 6Tbps optical pluggable modules, it is limited to 32 modules per Rack Unit (RU), typically requiring 2 RUs to achieve 102.