The newly announced NVIDIA Vera Rubin platform is set to deliver advanced supercomputing capabilities for scientific research. NVIDIA announced the platform at the ISC High Performance 2026 conference, combining native double-precision (FP64) performance with the company’s artificial intelligence technologies.
The platform integrates NVIDIA Rubin GPUs and NVIDIA Vera CPUs using NVLink-C2C interconnects. Furthermore, it incorporates ConnectX-9 SuperNICs and BlueField-4 DPUs within a direct liquid-cooled architecture. This setup delivers 7 exaflops of AI performance for scientific applications and 5 petaflops of native FP64 computing power.
Technical Specifications of the Platform
Furthermore, the platform integrates with the NVIDIA CUDA-X libraries and the full suite of the company’s AI software. This integration allows researchers to run traditional physics simulations alongside modern machine learning models on a single unified system.
Specifically, the architecture allows up to 144 GPUs per rack, transforming individual racks into high-density supercomputers. Consequently, researchers can run larger simulation models with improved accuracy. The platform is designed to accelerate workloads such as climate modeling, computational fluid dynamics, and energy exploration.
Next-Generation Supercomputing Deployments
Several leading research centers plan to deploy the new technology. For instance, the Leibniz Supercomputing Centre will use the platform to power its “Blue Lion” supercomputer, built by HPE, which is scheduled to begin operations in 2027. Meanwhile, the National Energy Research Scientific Computing Center will deploy the “Doudna” supercomputer, a Dell Technologies system utilizing the platform.
Additionally, the Los Alamos National Laboratory has selected the platform for its upcoming Mission, Vision, and Veritas systems. These systems, built by HPE, will support national security, open science, and quantum computing research. Notably, the Veritas system is designed specifically for software agents to advance scientific discoveries.
The Impact of NVIDIA Vera Rubin on Scientific Research
Jensen Huang, founder and CEO of NVIDIA, stated that scientific discovery has become a race between the complexity of global challenges and the computing systems built to solve them.
“NVIDIA Vera Rubin is a new tool for science, a rack-scale supercomputer that brings simulation, AI, and data processing together to help researchers and industries design and discover faster than ever.”
Jensen Huang, Founder and CEO of NVIDIA
Global System Integrators and Availability
Global system manufacturers, including Bull, Dell Technologies, GIGABYTE, HPE, and Supermicro, are bringing the NVIDIA Vera Rubin NVL4 platform to market. These liquid-cooled racks will allow organizations to deploy accelerated computing at scale. According to the company, the systems are expected to become available in the fourth quarter of 2026.
