The burgeoning demand for computing power in AI large - scale models is propelling the localization of high - end interconnect technologies. Recently, the UB (Unified Bus, with its Chinese name being "Lingqu") interconnect technology tailored for super - nodes made its debut. This technology fills the void in domestic ultra - high - speed interconnect protocols and shatters the foreign monopoly held by NVLink technology. The UB protocol facilitates ultra - large - scale node interconnectivity. It cuts down on latency and boosts both bandwidth and reliability. This is achieved through its innovative peer - to - peer architecture and unified memory access semantics. By doing so, it advances the autonomy and controllability of AI computing power infrastructure, which is in line with the global trend of technological self - reliance and is of great significance in the context of China's pursuit of independent innovation in the AI field. In Western tech ecosystems, there's also a growing emphasis on reducing reliance on external technologies for strategic sectors, and the UB protocol's development aligns with such global movements towards greater technological sovereignty.