NVIDIA Unveils Rubin HPC GPU for Science and Engineering at SC26
NVIDIA used the SC26 supercomputing conference in Chicago to announce Rubin HPC, a new graphics processing unit in its Vera Rubin platform aimed at scientific simulation, engineering and analytics workloads. In a company blog post published Tuesday, NVIDIA said the chip is designed for workflows in which AI “agents” generate candidate materials, molecules or designs and then rely on simulations and data processing to test those ideas.
NVIDIA said Rubin HPC will have “more than 50,000” CUDA cores on a single GPU. The company also said the product includes “a sparse data acceleration capability” meant to help keep those cores supplied with data, and that it uses seventh-generation NVIDIA NVLink, the company’s technology for linking GPUs together at high speed. NVIDIA did not publish a full public technical brief alongside the announcement, so those company claims were not accompanied by detailed public specifications such as exact core count, memory capacity, bandwidth or power draw.
The company said Rubin HPC is expected to be available in 2028. Ian Buck, NVIDIA vice president of hyperscale and high-performance computing, said in the blog post: “For agents tackling science and engineering, faster simulation means faster discovery. Rubin HPC accelerates the simulations agents use to test ideas and guide their next steps, helping scientists and engineers explore more possibilities and bring breakthroughs closer.” SC26 is taking place Nov. 15-20 in Chicago.
Rubin HPC sits within NVIDIA’s broader Vera Rubin data center platform, which the company introduced earlier in 2026 as its architecture after Blackwell for AI and high-performance computing. NVIDIA is framing the new GPU as part of that larger push into both AI and traditional HPC tasks such as simulation and analysis. One notable detail is that earlier public Vera Rubin materials referred to sixth-generation NVLink, while this Rubin HPC announcement uses seventh-generation NVLink terminology. NVIDIA did not release a full public spec sheet with the announcement.
The disclosure so far is high level and comes entirely from NVIDIA’s own announcement. No public detailed product brief was located with exact information on Rubin HPC’s memory, precision-specific performance, form factor, pricing or thermal design power. NVIDIA also did not disclose customer commitments, shipment allocations or other commercialization details in the announcement.
Stocks: NVDA