Nvidia (NASDAQ: NVDA) announced its Vera Rubin accelerated computing platform at ISC High Performance 2026 in Hamburg, Germany, targeting scientific research institutions and national laboratories with high-performance computing and AI capabilities.

According to the company, a single Vera Rubin rack system can deliver more than 7 exaflops of AI performance and 5 petaflops of native double-precision (FP64) computing with up to 144 GPUs, performance the company says is comparable to systems ranked on the TOP500 list of the world's most powerful supercomputers.

Several major research institutions have committed to deploying Vera Rubin-based systems. The Leibniz Supercomputing Centre in Germany plans to bring its Blue Lion system online in 2027, built on Vera Rubin and HPE Cray infrastructure, targeting approximately 30 times the computing capacity of its current system.

The National Energy Research Scientific Computing Center at Lawrence Berkeley National Laboratory will deploy a Dell Technologies system named Doudna, intended for workloads spanning molecular dynamics, high-energy physics, fusion energy and drug discovery.

Los Alamos National Laboratory has selected Vera Rubin for three systems — Mission, Vision and Veritas — to be built by HPE, covering national security workloads, open science research and agentic AI applications.

System manufacturers Bull, Dell Technologies, GIGABYTE, HPE and Supermicro have announced direct liquid-cooled Vera Rubin rack systems for the broader market. Nvidia said Vera Rubin NVL4-based systems are expected to be available from these manufacturers in the fourth quarter of 2026.

"Scientific discovery is now a race between the complexity of the world's greatest challenges and the computing systems built to solve them," said Jensen Huang, founder and CEO of Nvidia.

The Vera Rubin platform combines Nvidia Rubin GPUs and Vera CPUs connected through NVLink-C2C interconnects, ConnectX-9 SuperNICs and BlueField-4 data processing units in a liquid-cooled architecture.