AMD is still dominating the supercomputing scene, with its El Capitan and Frontier systems holding firm at the No. 1 and No. 2 spots on the latest Top500 list, announced during ISC High Performance 2025. That makes seven consecutive Top500 rankings where an AMD-powered system has claimed the crown—and now, AMD gear runs 172 of the world’s fastest systems, accounting for over a third of the list.
El Capitan, located at Lawrence Livermore National Laboratory, once again ranked as the fastest supercomputer on the planet, clocking in at 1.742 exaflops on the High Performance Linpack (HPL) benchmark. It’s built by HPE and powered by AMD Instinct MI300A APUs, which merge CPU and GPU computing into one chip to handle everything from simulations and data analytics to AI inference, essentially combining HPC and AI into a single platform for ease of use.
Meanwhile, Frontier which is based at Oak Ridge National Laboratory, retained its No. 2 status with a score of 1.353 exaflops. As the world’s first exascale system, it’s powered by AMD EPYC CPUs and MI250X GPUs, driving research in climate, energy, and AI modeling.
In the AI-specific HPL-MxP benchmark, El Capitan debuted at No. 1 with a staggering 16.7 exaflops of mixed-precision performance, while Frontier and LUMI (also AMD-powered) followed close behind in third and fourth place. El Capitan also topped the High-Performance Conjugate Gradient (HPCG) ranking with 17.4 petaflops, thanks to the high memory bandwidth delivered by AMD’s APU architecture.
Even with all that performance, energy efficiency remains a key differentiator where AMD now powers 12 of the top 20 systems on the Green500 list, including both El Capitan and Frontier, which still scored impressively given their massive compute power at ranks 26th and 32nd in efficiency, respectively.
Beyond the headline systems, AMD’s footprint is growing fast as the latest Top500 list includes 17 new AMD-powered entries, among them the Hunter system at HLRS in Stuttgart and the Viper supercomputer at the University of Hull, both using MI300A APUs. CINECA’s new Pitagora system in Italy also joins the ranks, running on 5th Gen EPYC CPUs.
AMD executive Forrest Norrod summed up the momentum by emphasizing that AMD isn’t just chasing records, but it is also setting the pace for where HPC and AI are heading. With the convergence of AI and scientific computing now accelerating, AMD is positioned at the center of it all, offering a full stack of high-performance and energy-efficient CPUs, GPUs, and APUs that are already driving breakthroughs from exascale computing to generative AI.
