How AMD EPYC CPUs and AMD Instinct GPUs are powering the world's HPC facilities

TL;DR

• Two of the three most powerful supercomputers in the world are powered by AMD EPYC CPUs and AMD Instinct GPUs
• AMD EPYC CPUs and AMD Instinct GPUs power 40% of the 10 most powerful supercomputers in the world
• Next generation AMD EPYC CPUs and next generation AMD Instinct GPUs slated to power the Alice Recoque supercomputer
• Next generation AMD EPYC CPUs and next generation AMD Instinct GPUs slated to power the Discovery supercomputer
• The next-gen AMD Instinct MI430X GPU is designed to redefine what a GPU can offer HPC systems
• The next-gen AMD Instinct MI430X GPU represents a new class of accelerator for simulation, modeling and AI-driven science.

Over the past several years, AMD has made a name for itself in the world of high-performance computing (HPC) and its influence only continues to grow.

This is impressive on its own, but perhaps even more notable is AMD knocked itself off the top spot. Frontier, which was also built by HPE and featured an AMD Optimized 3rd Generation EPYC 64C 2GHz processor and AMD Instinct MI250X GPU accelerator, had held the top spot since coming online in 2022.

In fact, 40% of the top 10 fastest supercomputers in the world, as measured by TOP500, are powered by AMD EPYC CPUs and AMD Instinct GPUs and 32% of the top 50 supercomputers use AMD EPYC and/or AMD Instinct processors.

AMD EPYC CPUs and AMD Instinct GPUs also power three of the top five supercomputers ranked by the high-performance conjugate gradient (HPCG) benchmark – an alternative to tests like LINPACK.

With environmental responsibility also high on the agenda for businesses today, it’s notable that 40% of the top 10 slots in the Green500 list of most energy efficient supercomputers are also occupied by systems using AMD EPYC chips.

AMD is committed to HPC excellence

Of course, HPC isn’t only confined to the publicly listed Top500 supercomputers. Some organizations will have their own HPC infrastructure or will be accessing cloud-based HPC that, for one reason or another, doesn't feature on the list.

The AMD commitment to serving this level of HPC, as well as the larger supercomputers, has also been repeatedly demonstrated.

Most recently, in May 2026 at the HPC User Forum, the company previewed the AMD Instinct MI430X GPU.

In the words of AMD, the Instinct MI430X GPU “is designed to redefine the limits of what a GPU can offer to future HPC systems”.

Projected specs for the Instinct MI430X GPU are:

• More than 200 TFLOPs of native FP64 performance
• Up to 6x higher FP64 performance of next generation Nvidia Rubin architecture in select HPC workloads
• Designed to deliver among the highest FP64 performance of any GPU

Native FP64 – as offered by AMD Instinct GPUs – is extremely important for traditional HPC workloads, where accuracy matters. While some newer architectures use emulation-based approaches to accelerate FP64 operations, their performance and efficiency can vary depending on the application and numerical method.

The fact that AMD Instinct GPUs are uniquely well-suited to accelerate even the most demanding AI and HPC workloads is recognised in two upcoming supercomputers: Alice Recoque at GENCI in France and Discovery, at Oak Ridge National Laboratory (ORNL) in the USA.

Alice Recoque powered by AMD EPYC and AMD Instinct

Alice Recoque, which is expected to be deployed in 2026 as Europe’s newest supercomputer, is slated to feature AMD Instinct MI430 GPUs and next-gen AMD EPYC CPUs, codenamed “Venice”.

The supercomputer is expected to deliver more than one exaflop of HPL performance and will support both traditional HPC and AI workloads.

In the words of Philippe Lavocat, CEO of GENCI, the Alice Recoque supercomputer powered by AMD Instinct MI430 GPUs, will “[pave] the path toward post-Exascale services … [and] propel Europe ‘Beyond HPC,’ federating high-performance computing, artificial intelligence, and quantum technologies to empower science, accelerate innovation, and strengthen our technological sovereignty and global competitiveness”.

Discovery powered by AMD EPYC and AMD Instinct

The Discovery supercomputer, designed to feature AMD Instinct MI430X GPUs, is being created under ORNL’s Genesis Mission in partnership with the US Department of Energy. It’s expected to be deployed two years after Alice Recoque, in 2028, and is slated to similarly feature next-gen AMD EPYC “Venice” CPUs.

"The Discovery system will drive scientific innovation faster and farther than ever before," said ORNL director Stephen Streiffer. "ORNL's leadership in supercomputing has dramatically shortened researchers' time from problem to solution across a host of fields and industries. With Discovery, the integration of high-performance computing and AI promises breakthroughs at the accelerated speed and scale necessary for continued US leadership in an increasingly competitive global environment."

Overall AMD Instinct MI430X GPU is designed to be a “step change” in capability for AI gigafactories and HPC centers, where, in the words of AMD, “numerical fidelity and throughput are mission-critical”.

The future of HPC and advanced AI systems with AMD

AMD has established itself as a leader in HPC and supercomputing by several measures, including its listings in the Top500 and Green500.

AMD EPYC CPUs and AMD Instinct GPUs are used in some of the most powerful supercomputers in the world, with more AMD technology-powered advanced supercomputers on the way.

AMD Instinct GPUs are uniquely well-suited to accelerate even the most demanding AI and HPC workloads, thanks in part to their native FP64 support.

Next-gen AMD EPYC CPUs, code named “Venice”, and next-gen AMD Instinct MI430 GPUs – both of which are expected to be released in 2026 – will bring even more power to HPC and supercomputing facilities.

Cautionary Statement

This article may contain forward-looking statements concerning Advanced Micro Devices, Inc. (AMD), which are made pursuant to the Safe Harbor provisions of the Private Securities Litigation Reform Act of 1995. Forward-looking statements are commonly identified by words such as "would," "may," "expects," "believes," "plans," "intends," "projects" and other terms with similar meaning. Investors are cautioned that any forward-looking statements in this article are based on current beliefs, assumptions and expectations, speak only as of the date of this article and involve risks and uncertainties that could cause actual results to differ materially from current expectations. Such statements are subject to certain known and unknown risks and uncertainties, many of which are difficult to predict and generally beyond AMD's control, that could cause actual results and other future events to differ materially from those expressed in, or implied or projected by, the forward-looking information and statements. Investors are urged to review in detail the risks and uncertainties in AMD’s Securities and Exchange Commission filings, including but not limited to AMD’s most recent reports on Forms 10-K and 10-Q.

AMD does not assume, and hereby disclaims, any obligation to update forward-looking statements made in this article, except as may be required by law.