LineShine, built with Chinese CPUs, to surpass the world’s fastest supercomputer

China has unveiled a new supercomputer built from entirely home-grown CPUs, with the aim of beating the world’s fastest machine while sidestepping U.S. export controls. Known as Lingsheng or LineShine, it is designed to reach 2 exaflops – or two quintillion calculations per second – edging past the 1.8-exaflop El Capitan, the current world record holder at the Lawrence Livermore National Laboratory in California. Unlike other exascale supercomputers, which rely on graphics processing units (GPUs), LineShine will run entirely on central processing units (CPUs). It uses 47,000 CPUs across 92 cabinets, according to Chief Designer Lu Yutong from the National Supercomputing Center in Shenzhen. Huang Xiaohui, the Center’s Deputy Director, told a conference in Shenzhen on April 24 that LineShine had achieved full-stack independence, from underlying hardware to core software, as a fully domestic supercomputer.

“By the end of 2025, we completed full system deployment and activation, with sustained performance exceeding 2 exaflops. Its performance has already surpassed that of the United States’ El Capitan, returning China to the world’s No 1 position,” she said. Huang said LineShine used the world’s most powerful CPUs and adopted an integrated architecture that supported both traditional high-performance computing and artificial intelligence (AI) workloads. Most of today’s fastest supercomputers use GPUs – chips that can do the same calculation thousands of times at once, making them ideal for AI and large simulations. CPUs, by contrast, handle a wider range of tasks but process them in smaller batches, requiring far more chips to reach the same overall speed. That is why nearly all of the world’s top systems, including El Capitan, are built around GPU-based designs built with chips from companies such as AMD and Nvidia. There are four exascale supercomputers on the widely watched Top500 supercomputer ranking, all of them powered by GPUs.

China’s earlier supercomputers depended on a mix of domestic and foreign chips. Systems such as Sunway TaihuLight ran on Chinese-designed processors, while others incorporated U.S.-made components, including chips from companies such as Intel and Nvidia. That landscape began to shift after the U.S. government imposed export controls on advanced chips starting in 2019, restricting China’s access to high-end processors used in supercomputing and AI.

LineShine is expected to run on a new CPU known as LX2, with each chip containing more than 300 computing cores. In its initial phase, the project used servers powered by Huawei Kunpeng processors, the South China Morning Post reports.