Sugon, a Chinese leader in supercomputing technology, has unveiled its Chinese AI supercluster called scaleX. This cutting-edge platform is a 10,000-card supercluster that delivers more than five exaflops of computing power. It is now one of China’s most powerful infrastructures, designed to meet the high demand for AI and complex computational tasks.
An exaflop (eFLOP) refers to a system capable of performing a quintillion floating-point operations per second, and the scaleX supercluster takes advantage of this immense processing capacity to handle intricate AI models that require substantial computing power.
ScaleX Supercluster: Tailored for Complex AI Applications
Sugon’s scaleX is optimized for AI-driven research, designed to support trillion-parameter models. This platform is a direct response to the increasing demand for AI infrastructure capable of handling more advanced and resource-intensive tasks. Li Bin, Senior Vice President at Sugon, emphasized that scaleX was developed with Sugon’s expertise in large-scale computer systems. The scaleX AI supercluster has already outperformed some components of NVIDIA’s Rubin Ultra NVL576 rack, which is expected to launch in 2027.
Competing with NVIDIA and Huawei in AI Infrastructure
The launch of scaleX represents a major step for China in the AI infrastructure market. As AI technology continues to transform industries, Sugon’s Chinese AI supercluster will likely challenge existing platforms from NVIDIA and Huawei. scaleX is a testament to China’s growing dominance in the field, driving forward innovation in computing and AI.
This new infrastructure will help China secure its position in the global AI race, offering immense potential for future development in fields ranging from digital science to data analytics.
