China has taken the top spot in the global supercomputing race with a machine that challenges the assumption that the fastest systems must be built around advanced GPUs.
The LineShine supercomputer, installed at the National Supercomputing Center in Shenzhen, has been ranked first in the TOP500 list, overtaking the US system El Capitan in California. The ranking measures high-performance computers every six months using standardized benchmarks for speed, real-world performance and energy efficiency.
LineShine’s reported performance is 2,198 exaflops, meaning it can carry out more than 2 quintillion operations per second. Its power draw is about 42.2 megawatts, according to the benchmark figures.
The development could affect how governments and research institutions think about computing infrastructure, especially in countries facing limits on access to advanced chips. Most new supercomputers lean heavily on GPUs, the same kind of processors used to train and run large AI systems. LineShine instead relies on roughly 45,000 China-developed LX2 central processing units.
The system also runs on domestically developed infrastructure, including the LingKun platform, the LingQi high-speed network and Kylin OS, a Linux-based operating system used in Chinese scientific and government computing.
The approach challenges the idea that export restrictions on GPUs can fully determine the pace of supercomputing competition. US controls on advanced chips and related technologies have pushed China to invest in alternative architectures, while Beijing has answered with its own measures.
Early results suggest China has found one path around those limits. The more difficult test is whether LineShine’s domestic design can support the scientific research, simulation work and AI workloads that make supercomputers matter beyond the rankings.
