NVIDIA announced Rubin HPC, a GPU in its Vera Rubin platform, for simulation and data processing in science, engineering and analytics. The company says a single GPU will have more than 50,000 CUDA cores, and expects Rubin HPC to be available in 2028. NVIDIA’s announcement also describes sparse-data acceleration and seventh-generation NVLink.
What Rubin HPC is designed to do
Rubin HPC is intended to accelerate simulations and data processing, including work in which AI agents explore candidate materials, molecules and designs. In that workflow, a simulation tests an idea and its results can guide what an agent tries next. NVIDIA also names scientific and engineering analytics among the target workloads.
“For agents tackling science and engineering, faster simulation means faster discovery,”— Ian Buck, NVIDIA’s vice president of hyperscale and high-performance computing, in NVIDIA’s announcement.
The announced features and workloads
NVIDIA says Rubin HPC has more than 50,000 CUDA cores on one GPU. That is a hardware specification; it does not give a simulation’s run time. The company also describes sparse-data acceleration as a way to keep the GPU’s compute resources supplied with data.
The announcement names sparse matrix operations, stencil computations, sorting and reductions as workload patterns Rubin HPC is intended to accelerate. NVIDIA’s CUDA-X libraries provide optimized algorithm implementations for agents to use with the GPU. For connections between GPUs, NVIDIA names seventh-generation NVLink.
What NVIDIA’s 2028 timeline means
NVIDIA announced Rubin HPC at SC26 in 2026 and says it is expected to be available in 2028. That is the company’s projected availability year. The announcement offers no application benchmark, so the announced core count and features do not quantify how quickly a particular scientific workload would run.