1 hour ago
26
China just broke the supercomputing sound barrier. LineShine, developed by the National Supercomputing Centre in Shenzhen, debuted at No. 1 on the TOP500 list with a sustained performance of 2.198 exaflops on the High Performance Linpack benchmark. That makes it the first system in history to crack the 2-exaflop threshold using only CPUs.
The announcement came at the International Supercomputing Conference in Hamburg, Germany, on June 23, 2026. It dethrones the American-built El Capitan, which had held the crown since late 2024 with its 1.809 exaflops. For context, LineShine is roughly 21% faster than the machine it replaced.
What makes LineShine different
Most top-tier supercomputers lean heavily on GPUs to hit their performance numbers. LineShine took a different path entirely, built on an all-CPU architecture using Armv9 LX2 processors. The system packs 13.79 million cores across 304-core chips running at 1.55 GHz.
The machine doesn’t just top the main ranking. It also claimed the No. 1 spot on the High-Performance Conjugate Gradients (HPCG) list with 22.00 petaflops. On the HPL-MxP list, which measures mixed-precision workloads, it landed fourth with 7.92 exaflops.
Power consumption sits at 42.2 megawatts, giving it an efficiency rating of 52.07 gigaflops per watt.
This is the first time a Chinese supercomputer has topped the TOP500 since Sunway TaihuLight held the crown back in 2017.
The geopolitical computing race
The US has spent the better part of a decade restricting China’s access to advanced semiconductor technology, particularly high-end chips from companies like Nvidia and AMD. LineShine’s all-CPU design suggests China found a workaround. Rather than relying on the GPU architectures that dominate Western high-performance computing, the Shenzhen team scaled out with domestically aligned processor technology. Whether the Armv9 LX2 chips were designed entirely in-house or licensed through Arm’s architecture remains a detail worth watching, but the end result speaks for itself.
What this means for crypto and AI investors
Supercomputers of this caliber are primarily used for scientific simulation, climate modeling, and increasingly, training massive AI models. The rise of decentralized AI networks, projects that distribute machine learning workloads across blockchain-coordinated infrastructure, depends on the broader trajectory of computing power.
There’s also a security dimension that crypto investors shouldn’t ignore. Sustained exaflop-scale computing is a reminder that the cryptographic assumptions underpinning most blockchains aren’t eternal. Projects investing in post-quantum cryptography may find renewed attention as headlines like this circulate.
Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our Editorial Policy.
English (US) ·