Chinese supercomputer LineShine tops new TOP500 list

The Chinese supercomputer LineShine heads the 67th edition of the TOP500 list, the ranking of the world's most powerful supercomputers. The list was presented on 23 June 2026 at the ISC 2026 conference in Hamburg. It is the first time a Chinese machine has led the list since the era of Sunway TaihuLight and Tianhe-2, which alternately held the top position during the 2010s.

LineShine is installed at the National Supercomputing Centre in Shenzhen and was built by the Shenzhen Cloud Computing Center. On the High Performance Linpack benchmark, the standard measure used by the TOP500, the system achieves 2.198 exaflops. That is more than twenty percent higher than El Capitan, the American system that previously held the top spot. The theoretical peak performance of LineShine stands at 2.736 exaflops; the measured HPL score therefore represents approximately eighty percent of that theoretical maximum.

The system contains 13.79 million computing cores and is built around a custom-designed Chinese processor. Further details about that processor had not been fully disclosed at the time of publication.

What the TOP500 list measures

The TOP500 is published twice a year and ranks supercomputers based on their performance on the HPL benchmark. That test solves a large system of linear equations and provides a standardised indication of computing power. The result is expressed in flops, floating-point operations per second. Two exaflops is equivalent to two trillion million such operations per second.

The list is used as a reference by scientists, governments and the technology industry to track the global distribution of computing capacity. A position at the top therefore carries not only technical but also geopolitical significance, particularly as computing power is increasingly linked to strategic autonomy in areas such as climate modelling, materials science and AI training.

Position of the United States and Europe

El Capitan, installed at Lawrence Livermore National Laboratory in California, drops to second place. When introduced in 2024, that system became the first machine ever to exceed one exaflop on the HPL benchmark. Frontier, also American, previously held the top position for several years.

Europe has traditionally held a limited share of the absolute top of the TOP500. The most powerful European machine on recent lists was JUPITER, installed at Forschungszentrum Jülich in Germany. European supercomputers are partly funded through EuroHPC, the joint initiative of the EU and participating countries to strengthen European computing capacity.

Significance of Chinese processors

A notable aspect of LineShine is that the system runs on a custom-designed Chinese processor, at a time when China has restricted access to advanced Western chips, including Nvidia's most powerful GPUs, due to American export controls. If the specifications of that processor are confirmed, LineShine demonstrates that Chinese manufacturers are capable of building systems that reach the global top at the system level, even without the chips commonly used in Western HPC systems.

Details about the architecture, the production location of the processor and the role of any Chinese chip designers have not yet been made public by the Chinese authorities or the Shenzhen Cloud Computing Center.

Relevance for the European AI and tech scene

For European policymakers and investors, the rise of LineShine illustrates how closely computing power and technological sovereignty have become intertwined. The EU is investing in its own infrastructure through EuroHPC and the AI Continent Action Plan, partly to reduce dependence on non-European systems. The fact that LineShine tops the list using Chinese chips underscores that building a fully independent semiconductor chain, from design to fabrication, is a multi-year undertaking requiring parallel efforts on multiple levels. For founders and researchers in the Dutch and broader European AI scene, access to large-scale computing capacity remains a practical bottleneck; the question of how that capacity will be distributed and priced in the coming years therefore remains a pressing one.