Inspur Launches SD200 Ultra Supernode for 2.8-Trillion-Parameter LLM

At the 2026 AI Computing Conference (AICC 2026) on September 21, Inspur unveiled the Yuanmao SD200 Ultra supernode AI server. Based on domestic AI chips, the server can run the 2.8-trillion-parameter Kimi K3 large language model with token generation latency dropping below 5.85 milliseconds for the first time. Inspur also introduced the Yuanmao HC2000 multi-computing-power rack system.

Inspur is a key provider of AI computing infrastructure in China, and the new products expand its server lineup for large-model workloads.