Nvidia expands AI infrastructure push into high-performance computing Nvidia unveiled Rubin HPC, a new GPU in its Vera Rubin platform with more than 50,000 CUDA cores per GPU, seventh-generation NVLink, sparse-data acceleration and optimizations for sparse matrix operations, stencil computations and sorting and reduction, at the Supercomputing '26 show in Chicago; the chip is expected to ship in 2028. Nvidia also announced a rack-scale Vera CPU configuration combining 192 Vera CPUs per rack with Quantum-X800 InfiniBand networking and SHARP in-network computing, plus Nemotron for Science, a collection of open AI models, scientific datasets and post-training tools built on Nemotron 3.5 Super VL for building specialized scientific AI agents. Nvidia extended its CUDA-X libraries with cuPhoton, which cuts astrophysics and telescope data analysis from over a month to eight minutes, cuPotential at up to 3x faster molecular simulations, Daviator at up to 22x faster materials science and quantum-mechanical calculations, and cuSpa at up to 900x faster than CPUs for biological and cellular analysis. Perhaps anxious https://www.networkworld.com/article/4248180/nvidias-leadership-crisis-puts-enterprise-ais-dependence-on-one-company-under-scrutiny.html to talk about technology instead of its CEO drama https://www.networkworld.com/article/4248533/whats-next-for-nvidia.html , Nvidia unveiled a raft of HPC-related news at the Supercomputing ’26 https://sc26.supercomputing.org/ show taking place in Chicago. Nvidia’s https://www.networkworld.com/article/3562856/nvidia-latest-news-and-insights.html announcements at SC26 continue the company’s efforts to expand beyond selling GPUs into supplying complete AI and high-performance computing infrastructure. Rubin HPC: A new GPU for scientific computing The news covers next-generation GPU and CPU platforms, networking and interconnect technology, accelerated scientific computing, and systems designed for large-scale AI workloads. Here is a breakdown of the major developments. Nvidia announced Rubin HPC https://blogs.nvidia.com/blog/rubin-hpc/ , a new GPU within its Vera Rubin platform designed specifically for high-performance computing applications. Unlike GPUs primarily optimized for AI training and inference, Rubin HPC is designed to boost scientific simulations, engineering applications, and large-scale data processing. Rubin HPC has more than 50,000 CUDA cores on a single GPU. The core count for regular Roman GPUs has not been disclosed. It includes seventh-generation NVLink for accelerated GPU-to-GPU communication, sparse-data acceleration, and optimizations for sparse matrix operations, stencil computations, and sorting and reduction operations. Rubin HPC is expected to ship in 2028. Vera CPU platform expands into HPC Nvidia also announced a rack-scale version of its Vera CPU platform specifically designed for HPC. When it first introduced its own CPU, Nvidia positioned it as an alternative to traditional server processors from Intel and AMD. Now Nvidia is positioning its CPU for a specific market. The new configuration combines 192 Vera CPUs per rack plus Quantum-X800 InfiniBand networking, and SHARP in-network computing technology. It uses existing NVL72 rack components, including chassis, power distribution, and cooling infrastructure. Nemotron for Science: AI agents for scientific research Nemotron is a family of AI models developed by Nvidia, designed to power AI applications, particularly autonomous AI agents. Nemotron for Science https://huggingface.co/nvidia/NVIDIA-Nemotron-3.5-Super-VL-120B-A12B-BF16 is a collection of open AI models, scientific datasets and post-training tools designed to help researchers develop specialized AI agents. The platform is built on Nemotron 3.5 Super VL https://developer.nvidia.com/blog/nvidia-nemotron-3-5-super-vl-powers-specialized-agents-with-multimodal-reasoning/ and includes scientific evaluation tools, an open chemistry dataset, post-training recipes for specialized scientific applications, and integration with CUDA-X accelerated computing libraries. The objective is to enable AI agents to do more than answer scientific questions. They can interpret research information, generate code, execute computational tools and assist with scientific workflows. CUDA-X: Accelerating scientific applications Nvidia announced additional capabilities for its CUDA-X https://www.nvidia.com/en-us/technologies/cuda-x/ software libraries, which are designed to accelerate specialized scientific applications. Some notable capabilities include: cuPhoton, astrophysics and telescope data that can reduce analysis time from over a month to eight minutes; cuPotential for molecular simulations that are up to 3x faster than the prior generation; Daviator for materials science and quantum-mechanical calculations up to 22x faster than the prior generation; and cuSpa for biological and cellular analysis up to 900x faster than CPUs. The broader objective is to make GPU acceleration available throughout scientific workflows, including drug discovery, materials research, astrophysics and biological research. Nvidia Photonics: Networking for massive GPU clusters Nvidia highlighted its photonics technology https://www.networkworld.com/article/4233992/cisco-nvidia-push-ultra-dense-networking-for-ethernet-ai-fabrics.html , particularly co-packaged optics CPO switches. The announcement included a deployment by cloud provider Lambda, which is using Nvidia Photonics switches to connect GB300 NVL72 racks containing more than 10,000 GPUs. CPO integrates optical connectivity closer to switching silicon, reducing reliance on conventional electrical connections for high-bandwidth data movement. The intended benefits promise lower networking power consumption; greater network reliability; improved scalability; and better connectivity for increasingly large GPU clusters. This is an important development because networking power consumption and bandwidth are becoming major constraints as AI clusters grow.