China's domestically developed LineShine supercomputer has reached a major milestone in next-generation computing, becoming the first system to sustain more than 2 EFLOPS on the High Performance Linpack benchmark. The achievement secured the top spot on the latest TOP500 ranking at the International Supercomputing Conference (ISC) High Performance 2026 in Germany.

Developed after years of research by Chinese scientists, LineShine adopts an innovative all-CPU architecture rather than the conventional CPU-GPU design. By integrating AI matrix acceleration units directly into domestically designed CPUs, the system improves computing efficiency, reduces data-transfer bottlenecks and simplifies the migration of scientific applications.

Every key component, from processors and memory packaging to networking and system software, was developed in China, strengthening the country's independent computing capabilities. LineShine is already supporting research in atmospheric and ocean science, engineering, materials science, drug discovery, brain science and artificial intelligence, with continued efforts to expand its software ecosystem for broader scientific and industrial use.