AMD powers world’s fastest supercomputers & expands AI reach
El Capitan, powered by AMD technology, has maintained its top position on the most recent Top500 list, which ranks the world's fastest supercomputers. The company now powers 40% of the top ten supercomputing systems globally, including both El Capitan and Frontier, recognised as the world's first operational exascale system.
Top500 presence
AMD is present in 177 systems on the current Top500 list, making up almost 35% of all supercomputers featured. These installations span a range of research and commercial applications. The company's involvement extends to energy efficiency as well, with 26 of the top 50 systems on the Green500 list-focused on sustainable supercomputing-operating with AMD hardware.
Cloud computing expansion
High performance computing (HPC) offerings from major cloud providers are increasingly built on AMD's fifth generation EPYC processors. Amazon Web Services (AWS) recently made available the EC2 Hpc8a instance, targeting workloads that require significant memory such as computational fluid dynamics, weather forecasting, and pharmaceutical research. Microsoft Azure's HBv5-series virtual machines and Google Cloud's H4D virtual machines, both based on the latest AMD CPUs, have been released for general use. Google reports almost four times the performance of previous models across typical HPC workloads, ranging from manufacturing to healthcare.
AI infrastructure
As part of ongoing government-backed technology initiatives, AMD is providing the architecture for new supercomputers, dubbed Lux and Discovery, to support the US national AI strategy. Lux will enable AI research in fields including materials science, biology, and clean energy, and will be deployed at Oak Ridge National Labouratory in 2026. Discovery, also for Oak Ridge, aims to further increase performance and energy efficiency compared to previous systems by integrating the latest generation GPUs, CPUs, and open source software. Together, these systems are intended to support the infrastructure described as the nation's 'AI factory'-combining high performance computing with next-generation artificial intelligence.
Software developments
AMD has introduced the AMD Enterprise AI Suite, an open-source platform intended to help enterprises manage and deploy AI workloads on a large scale. The software package is optimised for AMD's GPUs and integrates with Kubernetes to support the entire AI process pipeline, including development and monitoring.
Recent scientific projects have drawn on AMD-based infrastructure. At Lawrence Livermore National Labouratory, researchers undertook what is described as the largest and fastest protein structure prediction workflow run to date, using the El Capitan supercomputer. AMD and the RIKEN Institute have also started collaboration on new scientific research and technology exchanges.
"High performance computing and AI are transforming how the world innovates - and AMD is driving that change. From powering the fastest supercomputers on the planet to enabling energy-efficient data centres and frontier AI systems, AMD technologies are accelerating discovery, insight, and progress," said Lawrence Livermore National Labouratory, Lawrence Livermore National Labouratory.
