AMD at the HPC User Forum 2026: AMD sets new bar for HPC with AMD Instinct MI430X GPU FP64 performance

The HPC User Forum (HPCUF), taking place May 5–6 in Austin, Texas, will once again bring together leaders from national labs, academia, and industry to discuss the future of high-performance computing. AMD is proud to participate as a sponsor and contributor to this important community event, reinforcing its role at the forefront of HPC and AI convergence.

AMD Instinct™ MI430X GPU: A New FP64 Performance Class

Today, AMD is previewing the upcoming AMD Instinct™ MI430X GPU, which is designed to redefine the limits of what a GPU can offer to future HPC systems. Projected to deliver more than 200 TFLOPs of native FP64 performance¹, the MI430X represents a new class of accelerator for simulation, modeling and AI-driven science. MI430X is projected to provide more than six times the FP64 performance of the next-gen NVIDIA Rubin architecture, positioning it to become the highest performance FP64 GPU ever built. ^2,3

For AI Gigafactories and high-performance computing centers, where numerical fidelity and throughput are mission-critical, MI430X is designed to be a step change in capability. The training data for next generation artificial intelligence models will be built on high fidelity simulations. Beyond traditional simulation workloads, this capability is increasingly central to the future of artificial intelligence itself. Many next-generation AI models will rely on training data generated from high-fidelity scientific simulations, from climate and materials science to nuclear engineering and fluid dynamics. As AI systems evolve toward surrogate models, automated laboratories, and closed-loop discovery pipelines, the quality of the data used to train those systems becomes the limiting factor. Models trained on low-precision or numerically unstable data inherit those limitations; models trained on accurate physical simulations can capture the true structure of the underlying science. As scientific teams push toward AI‑driven discovery, one question keeps coming up: can tomorrow’s infrastructure still be trusted to produce correct answers at scale?

Ultimately, accelerators such as the AMD Instinct™ MI430X, which is designed to provide leadership FP64 and low precision AI capabilities, in a single package, is planned to be the foundation of the emerging AI-for-science ecosystem.

Engaging the HPC Community in Austin

This year’s HPCUF agenda reflects the rapidly evolving landscape of HPC—where AI, simulation, and hybrid computing models are increasingly intertwined. Attendees can expect a range of technical sessions and panels covering middleware innovation, hybrid workflows, and domain-specific HPC applications.

A highlight of the event will be an industry panel in which AMD will participate in discussing “The Future of Precision in HPC: FP64, Reduced Precision, and Emulation”, where industry, national lab, and academic experts will discuss this evolving HPC landscape, the requirements of future systems, and the role software-based emulation techniques will play.

Powering the Next Wave of Supercomputers

AMD’s leadership in HPC is not theoretical—it is already being realized in major global deployments powered by AMD Instinct GPUs and the next-generation systems below to be powered by the AMD Instinct MI430X GPU.

Discovery (Oak Ridge National Laboratory, USA) 

The upcoming Discovery system is planned to be deployed in 2028 in cooperation with the U.S. Department of Energy (DOE) at ORNL under the Genesis Mission, which is a national program designed to expand America’s leadership in artificial intelligence (AI) and high-performance computing (HPC). It is expected to be the DOE’s next flagship system and will help drive breakthroughs in energy, biology, advanced materials, national security, and manufacturing innovation. Discovery will leverage AMD Instinct MI430X GPUs alongside next-generation AMD EPYC™ CPUs to enable large-scale AI training and inference, agentic AI, and scientific simulation. It is expected to represent one of the first “AI factory” supercomputers in the United States. See “AMD Powers U.S. Sovereign AI Factory Supercomputers, Accelerating an Open American AI Stack.”

Alice Recoque (Europe)

In Europe, the Alice Recoque system, which is expected to be Europe’s new supercomputer, will be powered by next-generation AMD Instinct MI430X GPUs and AMD EPYC™ CPUs. It is being deployed in cooperation with Grand équipement national de calcul intensif (GENCI), operated by Commissariat à l'énergie atomique et aux énergies alternatives (CEA). 

Alice Recoque is designed to deliver exascale-class performance for both AI and traditional HPC workloads, emphasizing energy efficiency and scientific throughput. It will tackle Europe’s most pressing societal, scientific, and industrial challenges by combining large-scale simulations, data analysis, and AI. Alice Recoque is expected to deliver more than one exaflop of HPL performance. Learn more about this system in the AMD release “AMD and Eviden to Power Europe’s New Exascale Supercomputer, the First Based in France”.

These deployments highlight the AMD growing footprint across sovereign AI and HPC infrastructure, enabling nations and institutions to build open, high-performance computing ecosystems.

Looking Ahead

As the HPC community gathers in Austin, the AMD message is clear:

• Precision matters — FP64 is foundational for scientific discovery
• Convergence is here — infrastructure must support both HPC and AI-for-Science
• Performance leadership — with AMD Instinct MI430X GPU setting a new bar

From thought leadership at the HPC User Forum to powering the world’s next-generation supercomputers, AMD is helping shape the future of computing—one breakthrough at a time.

To reach an AMD HPC expert, email HPC@AMD.com