The new TOP500 ranking of the world’s most powerful supercomputers, revealed today at the Supercomputing Conference in St. Louis, shows the HPE-Cray / AMD supercomputer El Capitan, housed at Lawrence Livermore National Laboratory, has retained the top spot while the Eviden JUPITER system at the EuroHPC / Jülich Supercomputing Centre in Germany has been declared the fourth certified exascale system and the first from Europe.
The JUPITER Booster system is listed at no. 4 on the TOP500 list by submitting a new measurement of 1.000 Exflop/s on the High Performance LINPACK benchmark. It is the first certified exascale system outside of the U.S. (China is known to have several exasc ale-class systems but no longer participates in the TOP500 list).
El Capitan, Frontier (Oak Ridge National Laboratory), and Aurora (Argonne National Laboratory) are still the first three systems on the TOP500.
The El Capitan system is an HPE Cray EX255a system remeasured with 1.809 exaflop/s on the HPL benchmark. LLNL also achieved 17.41 Petaflop/s on the HPCG (high performance conjugate gradients) benchmark, which makes the system the new No. 1 on this ranking as well.
El Capitan has 11,340,000 cores and is based on AMD 4th generation EPYC processors with 24 cores at 1.8 GHz and AMD Instinct MI300A accelerators. It uses the Cray Slingshot 11 network for data transfer and achieves an energy efficiency of 60.9 Gigaflops/watt.
The Frontier system is the No. 2 system on the TOP500 and is measured at 1.353 Exaflop/s. Frontier is based on the HPE Cray EX235a architecture and is equipped with AMD 3rd generation EPYC 64C 2GHz processors. The system has 9,066,176 total cores and also relies on Cray’s Slingshot 11 network for data transfer.
The Aurora system keeps the No. 3 spot on the TOP500 with 1.012 Exaflop/s on the HPL benchmark.
Aurora is built by Intel based on the HPE Cray EX – Intel Exascale Compute Blade, which uses Intel Xeon CPU Max Series processors and Intel Data Center GPU Max Series accelerators which communicate through Cray’s Slingshot-11 network interconnect.
JUPITER – JU Pioneer for Innovative and Transformative Exascale Research is located at the Forschungszentrum Jülich campus in Germany and is operated by the Jülich Supercomputing Centre. It is based on the Eviden’s BullSequana XH3000 direct liquid cooled architecture.
Top 10 Highlights
Microsoft’s Eagle cloud system stays at no. 5 with 561 Petaflop/s, showcasing cloud-based HPC scalability. Italy’s HPC6 follows at no. 6 (478 PFlop/s), while Japan’s Fugaku at no. 7 continues to lead Asia and remains second on the HPCG ranking (16 PFlop/s). Switzerland’s Alps system remains at no. 8 (435 PFlop/s), followed by Finland’s LUMI (no. 9, 380 PFlop/s) and Italy’s Leonardo (no. 10, 241 PFlop/s), both part of the EuroHPC initiative.
Green500
Measuring energy efficiency the Green500 ranking highlights the rise of NVIDIA Grace Hopper-based systems, which dominate the top three positions.
- KAIROS at CALMIP/University of Toulouse (France) — 73.28 GFlops/Watt, achieving 3.05 PFlop/s.
- ROMEO-2025 at ROMEO HPC Center (France) — 70.9 GFlops/Watt, delivering 9.86 PFlop/s.
- Levante GPU Extension at DKRZ (Germany) — 69.43 GFlops/Watt, reaching 6.75 PFlop/s.
All three systems share the BullSequana XH3000 design with Grace Hopper Superchips and NVIDIA NDR200 InfiniBand interconnects, highlighting the balance between energy efficiency and computing performance.
HPL-MxP: Mixed-Precision
The HPL-MxP benchmark emphasizes mixed-precision calculations crucial for AI and scientific simulation. The 2025 list confirms U.S. DOE leadership with El Capitan, Aurora, and Frontier occupying the top three positions, followed by JUPITER in fourth. Japan’s CHIE-4 system at Softbank ranks fifth with 3.3 Exaflop/s, demonstrating global diversification of HPC innovation.
