At the International Supercomputing Conference in Germany, the latest TOP500 ranking named China’s Lingsheng the world’s fastest supercomputer, achieving a sustained double-precision performance of 2.198 exaflops (EFlops). It is the first supercomputer to surpass 2 EFlops of sustained performance and marks China’s return to the top of the global supercomputing rankings for the first time since Sunway TaihuLight led the list in 2017.
People’s Daily portrays Lingsheng’s achievement as the culmination of nearly a decade of accelerated domestic innovation spurred by U.S. export controls on advanced computing technologies. The system is reportedly built entirely on indigenous technologies, including a domestically developed high-performance CPU, high-bandwidth on-chip memory, high-speed interconnects, advanced storage, liquid cooling, and a full-stack software ecosystem. According to the article, this end-to-end domestic architecture enables China to develop advanced computing capabilities independently of foreign suppliers.
A key feature highlighted by People’s Daily is Lingsheng’s departure from the dominant CPU-GPU heterogeneous architecture employed by leading U.S. exascale systems. Instead, it adopts an all-CPU architecture with integrated AI acceleration units embedded directly into the processor, eliminating the data-transfer overhead between CPUs and GPUs. The article presents this design as a new technological paradigm that demonstrates China’s “changing lanes to overtake” approach to bypass Western dominant competitors.
Source: People’s Daily, June 29, 2026
http://finance.people.com.cn/n1/2026/0629/c1004-40749501.html