On September 21, during the 2026 Artificial Intelligence Computing Conference (AICC2026), Inspur Information made a significant announcement by launching the Yuanmao SD200 Ultra super-node AI server. Leveraging domestically developed AI chips, this server is uniquely capable of independently supporting the operation of the Kimi K3 large-scale model, which boasts an impressive 2.8 trillion parameters. Notably, for the first time, the latency in Token generation has been reduced to below 5.85 milliseconds, marking a substantial advancement in AI server performance. Concurrently, Inspur Information also presented the Yuanmao HC2000 multi-computing power unit, further solidifying its position at the forefront of AI technology innovation.
