Suiyuan Technology exhibited high-performance super nodes at WAIC, with six major advantages to improve deployment efficiency

Zhixi Author | ZeR0 Editor | Mo Ying Zhixi reported on July 23 that during the World Artificial Intelligence Conference WAIC 2026, domestic cloud AI chip company Suiyuan Technology participated in the exhibition and held a forum with the theme of "Building a Token Economic Core Ecosystem for Collaborative Innovation", focusing on displaying its system-level innovation results in AI computing power infrastructure construction and industrial applications, and presenting the development picture of domestic AI computing power from technological breakthroughs to large-scale applications. In the eight years since its establishment, Suiyuan Technology has independently developed and iterated five cloud AI chips with four generations of architecture, building a complete product system covering AI chips, AI accelerator cards and modules, intelligent computing systems and clusters, and AI computing and programming software platforms. As the scale of large model training and inference continues to move towards the Wanka era, super nodes are becoming the key system architecture to support high-performance AI computing power infrastructure. At the exhibition, Suiyuan Technology focused on exhibiting high-performance super nodes based on self-developed acceleration modules, including: "Yunsui ESL64-O" based on advanced orthogonal architecture solutions. "Yunsui ESL64-C" based on the mature Cable-tray solution can meet the needs of large-scale cluster networking above 10,000 cards. Yunsui OGX standardized AI system products provide flexible and cost-effective computing power deployment solutions for training and inference scenarios. Self-developed optical interconnection prototype system solution based on NPO technology to explore a new generation of intelligent computing network interconnection technology. Relying on self-developed acceleration modules, Suiyuan Technology continues to promote innovation in high-performance super-node systems to meet the needs of large model training and high-parallel reasoning, and provide support for the construction of ultra-large-scale intelligent computing clusters. Among them, the orthogonal architecture solution "Yunsui ESL64-O" super node jointly created with ZTE has six major advantages: low latency, high reliability, strong signal, excellent heat dissipation, easy expansion, and convenient operation and maintenance. It further improves deployment efficiency through high-density integration, zero cables, and high scalability design. ▲"Yunsui ESL64-O" super node This super node was jointly released by ZTE and Suiyuan. It uses domestic CPU, AI chip, switching chip, domestic DPU, domestic network card chip, etc. to meet the requirements of information innovation and embody the domestic industrial chain association.