At SC25, NVIDIA announced advancements in BlueField DPUs, quantum computing, networking, AI physics, and more. NVIDIA founder Jensen Huang introduced the compact DGX Spark AI supercomputer and discussed the Grace Blackwell platform. NVIDIA also unveiled Apollo, a family of open models for AI Physics, and Warp, a Python framework for GPU acceleration in physics and AI.

NVIDIA introduced BlueField-4 DPUs at GTC Washington, powering the operating system of AI factories. The BlueField-4 combines a 64-core NVIDIA Grace CPU with NVIDIA ConnectX-9 networking for unprecedented performance and efficiency. Leading storage innovators like DDN, VAST Data, and WEKA are adopting BlueField-4 for AI and scientific workloads, transforming data centers into intelligent engines for AI and supercomputing.

NVIDIA unveiled Quantum-X800 InfiniBand switches and Quantum-X Photonics networking switches at GTC 2024, designed to power trillion-parameter-scale generative AI models. The Quantum-X Photonics platform addresses power efficiency, signal integrity, and reliability challenges for large-scale AI workloads. Companies like Siemens, Neural Concept, and Luminary Cloud are adopting NVIDIA Warp, a Python framework for GPU-accelerated simulation and AI workflows.

NVQLink, a universal interconnect linking accelerated computing to quantum processors, is being adopted by top scientific computing centers worldwide. NVQLink enables scalable error correction, hybrid applications, and real-time quantum-GPU workflows. NVIDIA and RIKEN are collaborating on two new GPU-accelerated supercomputers to advance Japan’s leadership in AI and quantum computing, scheduled to be operational in spring 2026.

Read more at Nvidia: Accelerated Computing, Networking Drive Supercomputing Ahead in Age of AI