China reclaimed the top spot in global supercomputing after a nine-year gap with the unveiling of LineShine, a powerful system located at the Shenzhen Cloud Computing Center. This development was confirmed on Tuesday by researchers who evaluate supercomputer performance using standardized tests. LineShine outperformed the leading U.S. system, El Capitan, housed at Lawrence Livermore National Laboratory in California, by more than 20 percent, ending the United States’ long-standing dominance of the twice-yearly Top500 ranking.

Distinctively, LineShine relies exclusively on standard microprocessors, known as CPUs, rather than the specialized graphics processing units (GPUs) commonly used in high-end supercomputers for intensive computations. According to Jack Dongarra, professor of computer science and electrical engineering at the University of Tennessee and one of the organizers of the Top500 list, this architecture represents a novel approach that blends artificial intelligence (AI) capabilities with traditional scientific computing tasks. Dongarra noted that most top-performing systems worldwide incorporate GPUs, making China’s design both impressive and unorthodox.

The emergence of LineShine intensifies the ongoing technological competition between China and the United States, with implications for science, national security, and geopolitics. While U.S. technology companies like OpenAI, Anthropic, and Google have advanced leading AI models supported by powerful Nvidia-produced GPUs, China has pursued alternative technological pathways. For example, the Chinese startup DeepSeek recently introduced a cutting-edge AI model operating with only a small fraction of specialized AI chips. This use of CPUs in place of GPUs could allow Chinese developers to sidestep existing U.S. export controls and tariffs targeting high-performance AI chip exports to China.

Jimmy Goodrich, senior fellow at the University of California Institute on Global Conflict and Cooperation, highlighted this development as a regulatory loophole, calling for stricter controls on CPU exports to China. Nonetheless, he also pointed out that despite China’s achievement with LineShine, it still lags behind the ultra-fast AI supercomputers developed by leading U.S. firms.

Supercomputers, traditionally designed for highly precise calculations in scientific fields such as climate modeling, cryptography, and nuclear weapons research, have operated with 64-bit data formats. However, commercial AI applications often employ lower-precision calculations to accelerate performance. U.S. national laboratories are increasingly incorporating these AI techniques alongside traditional methods to enhance scientific research.

Foreign supercomputers have occasionally topped the global rankings, with Japan’s systems holding No. 1 from 2020 to 2022. Industry observers like Addison Snell, an analyst at Intersect360 Research, caution against perceptions that only the United States is capable of producing the world’s fastest supercomputers, noting that other countries have developed competitive capabilities.

LineShine’s nearly 14 million computing cores are housed in 90 hardware cabinets, employing specialized circuitry embedded within CPU chips that accelerate matrix and vector calculations traditionally managed by GPUs. These chips are built on an instruction set licensed from Arm Holdings, a British technology company owned by Japanese firm SoftBank. Arm’s technologies are widely used in smartphones and data centers globally, including in China, where the company operates in compliance with export regulations.

Details about the chip manufacturers and production technology remain undisclosed. According to Dongarra, LineShine’s developers—experienced Chinese supercomputing experts—claimed the project was not government-funded, allowing them to submit their performance results for international recognition. They have also pursued acknowledgment through multiple submissions to the Gordon Bell Prize, an award honoring advancements in scientific computing.

Among its applications, LineShine has been employed in complex Earth system simulations and detailed modeling of the human brain. As competition in high-performance computing intensifies, China’s breakthrough with LineShine signals new strategic dynamics in the global race for technological leadership.