China’s LineShine supercomputer has taken first place in the June 2026 TOP500 ranking, marking China’s return to the list after a three-year absence from submissions.
The system is housed at the National Supercomputing Centre in Shenzhen and uses domestically designed chips, according to details published with the ranking. The submission marks China’s return to the list after it stopped submitting systems during a period of tighter US controls on advanced chips and computing technologies.
LineShine displaced El Capitan, the US system at Lawrence Livermore National Laboratory, which had previously held the top position. El Capitan is used by the US government to develop and maintain its nuclear weapons stockpile.
TOP500 said LineShine reached 2.198 exaflops on the HPL benchmark, making it the new No. 1 system on the list. El Capitan ranked second with 1.809 exaflops.
LineShine’s debut raised the number of publicly listed systems sustaining more than one exaflop on HPL from four to five. The five listed exascale systems are LineShine, El Capitan, Frontier, Aurora, and Germany’s JUPITER Booster. The ranking still reflects only systems submitted for measurement.
The ranking gives China the leading position on the TOP500 list, but experts cited by Reuters said the result does not establish LineShine as the world’s fastest system for artificial intelligence work.
The TOP500 list ranks supercomputers using benchmark tests associated with traditional high-performance computing. Its main HPL benchmark measures how fast a system solves dense systems of linear equations, and TOP500 notes that no single benchmark reflects a system’s overall performance.
On HPL-MxP, a mixed-precision benchmark relevant to AI and accelerator-heavy workloads, LineShine ranked fourth at 7.92 exaflops. LineShine also ranked first on the HPCG benchmark, with 22.00 petaflops.
Reuters cited technology and policy experts who said LineShine’s TOP500 position should be separated from its performance on AI-related workloads.
TOP500 said LineShine’s HPL-MxP result was consistent with a CPU-only design without dedicated low-precision accelerators. Many AI-focused systems rely heavily on GPUs or other accelerators.
LineShine is built on China’s custom LingKun platform, 304-core LX2 processors, LingQi interconnect, and Kylin OS. TOP500 said the system uses 13.79 million cores across LX2 processors running at 1.55 GHz.
The system draws about 42.2 megawatts of power and delivers an efficiency of 52.07 gigaflops per watt.
Cloud providers have built large systems for AI training and deployment. Microsoft, Amazon, and Google operate AI-focused computing systems, but most cloud companies do not submit their systems to the TOP500 process.
Since many private AI systems are not submitted, the TOP500 list does not provide a complete ranking of the largest AI computing systems in operation.
A 2025 Epoch AI study found that companies accounted for about 80% of AI supercomputer performance in its dataset, up from about 40% in 2019. The same study estimated that the US accounted for about 75% of AI supercomputer performance in the dataset, while China accounted for 15%.
Epoch AI also found that AI supercomputer performance doubled every nine months between 2019 and 2025, while hardware acquisition costs and power requirements doubled annually.
Epoch AI cited xAI’s Colossus as an example of private AI infrastructure, saying the system used 200,000 AI chips. The study also identified Colossus as the most performant AI supercomputer in its dataset as of March 2025.
Jimmy Goodrich, a senior fellow at the University of California’s Institute for Global Conflict and Cooperation, told Reuters that LineShine would not rank among the top five if major hyperscalers submitted their systems.
China first reached the top of the TOP500 list in 2010 and later traded the lead with the US and Japan. The country stopped submitting systems after a series of US export controls targeting chips and computing technologies under the first Trump administration and later under President Joe Biden.
Addison Snell, CEO of Intersect360 Research, told Reuters that the notable point was not LineShine’s performance, but China’s decision to submit the system for ranking. He said the move suggests Beijing wanted external recognition for its domestic chip design work.
The system’s disclosed specifications do not show the use of advanced AI chips. Reuters reported that this is likely because the tools required to manufacture those chips remain subject to US export controls.
The US Bureau of Industry and Security said its December 2024 rules targeted China’s ability to produce advanced semiconductors and advanced AI capabilities, including through controls on semiconductor manufacturing equipment and certain software and technology.
Goodrich said China appeared to be using the ranking to challenge perceptions about the effectiveness of export controls. He said the system’s technical details pointed to limits in its AI capabilities.
The ranking was released the same week US President Donald Trump signed executive orders on quantum computing. The orders included measures aimed at strengthening US work on quantum computing and related technologies.
The National Supercomputing Centre in Shenzhen did not immediately respond to Reuters’ request for comment.
Want to experience the full spectrum of enterprise technology innovation? Join TechEx in Amsterdam, California, and London. Covering AI, Big Data, Cyber Security, IoT, Digital Transformation, Intelligent Automation, Edge Computing, and Data Centres, TechEx brings together global leaders to share real-world use cases and in-depth insights. Click here for more information.
Tech Wire Asia is powered by TechForge Media. Explore other upcoming enterprise technology events and webinars here.
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。