The supercomputer leaderboard changed in 2026. China is back at No. 1, Europe has crossed the exascale line, and five machines now sustain at least one quintillion floating-point operations per second.

This countdown uses the June 2026 TOP500 and ranks systems by HPL Rmax — measured sustained performance, not theoretical peak. For more on the silicon underneath modern compute, see our Top 7 AI Accelerators and 7 Most Energy-Efficient AI Models.

Measured HPL Rmax from the June 2026 TOP500. One exaflop equals one quintillion floating-point operations per second.

#7 — Eagle

561.2 petaflops. Microsoft Azure’s Eagle uses more than 2 million cores in an NDv5 architecture built around Intel Xeon processors, NVIDIA H100 GPUs and NVIDIA InfiniBand. Unlike the national-lab machines above it, Eagle demonstrates that hyperscale cloud infrastructure can compete directly with purpose-built supercomputers.

Microsoft Azure and NVIDIA discuss the H100 GPU infrastructure at the center of Azure’s modern accelerated-computing stack.

#6 — HPC7

571.5 petaflops. Eni’s new HPC7 in Italy is the world’s fastest industrial supercomputer on the June list. It combines AMD EPYC CPU cores with AMD Instinct MI300A APUs inside an HPE Cray EX255a system. Eni uses the machine for energy, geoscience, carbon storage, batteries, fusion research and AI.

AMD’s official MI300 overview explains the integrated CPU-GPU architecture used by HPC7.

#5 — JUPITER Booster

1.000 exaflop. Europe’s JUPITER Booster is built around roughly 24,000 NVIDIA GH200 Grace Hopper Superchips and an Eviden BullSequana XH3000 platform. The system at Forschungszentrum Jülich is designed for both large scientific simulations and AI training. We previously covered the JUPITER compute architecture as Europe expanded its domestic HPC stack.

Forschungszentrum Jülich explains JUPITER and the infrastructure behind Europe’s first exascale system.

#4 — Aurora

1.012 exaflops. Argonne National Laboratory’s Aurora pairs Intel Xeon CPU Max processors with Intel Data Center GPU Max accelerators. Its 9.26 million cores are used for simulation, AI and data-intensive science ranging from fusion to cosmology and materials.

Argonne shows how researchers are using Aurora’s exascale compute and AI capability for scientific discovery.

#3 — Frontier

1.353 exaflops. Frontier at Oak Ridge National Laboratory was the first machine to break the exascale barrier on TOP500. Its HPE Cray EX235a architecture combines AMD EPYC processors with AMD Instinct MI250X accelerators and remains one of the most important scientific computers ever built.

Oak Ridge’s official Frontier video explains the machine that first pushed sustained supercomputing past one exaflop.

#2 — El Capitan

1.809 exaflops. Lawrence Livermore National Laboratory’s El Capitan uses AMD Instinct MI300A APUs and HPE’s Cray EX255a platform. Built for the U.S. National Nuclear Security Administration, it held the world No. 1 position until June 2026 and remains the fastest U.S. system.

Lawrence Livermore shows El Capitan’s hardware and its role in national-security and scientific simulation.

#1 — LineShine

2.1984 exaflops. LineShine at the National Supercomputing Centre in Shenzhen is the new world leader. The custom Chinese LingKun system uses 13.79 million CPU cores, LX2 processors, the proprietary LingQi interconnect and Kylin OS — notably without the GPU accelerators used by most of the other leaders. Its measured HPL result is about 21.5% above El Capitan.

This English-language 2026 overview walks through the current TOP500 leaders and LineShine’s unusual CPU-only route to No. 1.

Why this ranking matters

The interesting story is not only raw speed. The top seven use radically different strategies: cloud GPUs, industrial HPC, Grace Hopper superchips, Intel GPU Max, AMD MI250X and MI300A, and finally LineShine’s enormous CPU-only design. The TOP500’s LineShine report also shows the system leading HPCG, suggesting its design is competitive beyond the headline Linpack test.

The next race will be about more than exaflops: useful AI throughput, memory bandwidth, networking, cooling and power efficiency will determine which architectures scale economically. That same shift is why our recent coverage of chiplets and exascale computing matters as much as the ranking itself.

BitcoinVersus.Tech Editor’s Note:
We volunteer daily to ensure the credibility of the information on this platform is Verifiably True. If you would like to support to help further secure the integrity of our research initiatives, please donate using the QR code provided

For best QR-code scanning results, please scan this donation code from a desktop or larger screen. Mobile screens may prevent camera recognition.