On July 18th, news emerged that the 2026 World Artificial Intelligence Conference (WAIC) took place in Shanghai from July 17th to 20th. During the conference, Biren Technology showcased its cutting-edge NPO optical interconnect and distributed decoupled architecture super node solution. This innovative solution enables the expansion of a single super node to accommodate up to 1024 cards.

Biren Technology has been at the forefront of developing high-performance computing chips using a Chiplet architecture. This pioneering technology continues to be a cornerstone in its latest-generation BR2xx series GPUs. The BR2xx series stands out for its support of FP8 and FP4 low-precision, high-performance calculations, offering enhanced memory bandwidth and seamless integration of super node interconnect capabilities.

The company's proprietary BLink2.0 super node interconnect protocol facilitates the sharing of a unified memory space among up to 1024 GPUs. Furthermore, leveraging the BR2xx and BLink2.0 technologies, Biren Technology has established a three-tier super node product matrix. This matrix includes a 16-card standard server super node (utilizing electrical interconnect), a 128-card high-density full-rack super node (also employing electrical interconnect), and a 1024-card distributed decoupled architecture super node (featuring NPO optical interconnect).