Skip to content

NVIDIA Announces Rubin HPC — A New GPU to Accelerate Agents for Science, Engineering and Analytics

NVIDIA AI

The Vera Rubin platform GPU features over 50,000 CUDA cores and seventh-generation NVLink interconnect technology.

Key points

  • Designed to accelerate simulation, sparse matrix operations, and data processing for science and engineering workflows
  • Includes sparse data acceleration to keep compute cores continuously supplied with data
  • Pairs with NVIDIA CUDA-X libraries for agent-invoked simulation and analytics algorithms
  • Expected to become available in 2028

Why it matters

Researchers and engineers running large-scale scientific simulations will have dedicated hardware tailored for agent-driven discovery workflows.

Sources