Particle.news
Download on the App Store

China's LineShine Tops TOP500 List

Showing Chinese progress in building domestic CPUs, the Top500 HPL result does not show superiority in GPU-led AI training.

Overview

  • The Top500 update published on June 24, 2026 placed LineShine at number one with an HPL measurement of about 2.198–2.2 exaflops, overtaking the U.S. system El Capitán.
  • LineShine is installed at the National Supercomputing Centre in Shenzhen and is built with domestically designed LX2 CPUs, a proprietary LingQi interconnect and the Kylin operating system.
  • The ranking reflects performance on the HPL benchmark, a linear algebra test that rewards CPU-oriented, sequential workloads and does not directly measure GPU-style parallel training used for most modern AI models.
  • The United States still holds the majority of top slots on the list, and many cloud providers that run GPU-heavy AI systems do not submit their machines to Top500, so public rankings understate overall AI compute capacity.
  • Analysts say the result signals China’s push for semiconductor self-reliance in response to U.S. export controls and could accelerate domestic chip and system investment, but it does not by itself change who leads in AI model training.