LineShine became the first supercomputer to exceed 2 exaflops in sustained performance, putting China back atop the TOP500 ranking for the first time since 2017.
Its processors embed AI acceleration directly into the chip, letting scientific simulations and AI workloads run on one platform without shuttling data between separate CPUs and GPUs.
Researchers have already used the system for East Asia rainfall forecasting, and developers say it also targets materials science, drug discovery and large language model inference.
Built largely with domestically developed technology after tighter US export controls, LineShine highlights China's push for scientific self-reliance and a broader shift toward AI-integrated supercomputing.
Is China's new supercomputer a true leap forward or just a power-hungry workaround for US sanctions?
With China leading in supercomputing, what does this signal for the future of global technology competition?
As supercomputers demand more power than small cities, can we innovate faster than we consume energy?
China’s LineShine Achieves 2.2 Exaflops, Reclaims Global Supercomputing Leadership with CPU-Only Architecture
Overview
In June 2026, China reclaimed the global supercomputing lead as its LineShine system debuted at number one on the TOP500 list, achieving a remarkable 2.198 exaflops on the HPL benchmark. Unlike many modern exascale systems such as the U.S. El Capitan, which rely on GPU accelerators, LineShine stands out with its innovative CPU-only architecture using domestically designed LingKun processors. This breakthrough was driven by US export controls, pushing China to develop independent technology. Although LineShine's energy efficiency is lower than some GPU-based systems, its success marks a major shift in global high-performance computing and technological competition.