Nvidia's Vision For AI Factories / Not Just GPU's at GTC

This Week in HPC explores NVIDIA's transformative vision unveiled at GTC 2026, where the company positioned itself not just as a GPU leader but as a full-stack AI factory builder. The episode highlights the massive scale of the event—spilling across San Jose’s convention center, civic venues, hotels, and parks—and emphasizes Jensen Huang’s two-hour keynote that went far beyond graphics to showcase a new ecosystem of specialized chips: the Grace CPU, Grok LPU, BlueField 4 DPUs, SuperNICs, and silicon photonic interconnects under the Kyber initiative. The focus is on building integrated, factory-scale AI systems like the NVL72 and the upcoming NVL1152, which will combine hundreds of GPUs, CPUs, and accelerators with optical interconnects to overcome the physical limits of copper. A major theme is the rise of AI agents via OpenClaw and NVIDIA’s Nemo Claw, which are accelerating development at unprecedented speed—what the hosts call the 'third derivative of change.' The episode also raises critical long-term questions about the lifecycle, maintenance, and sustainability of these massive AI infrastructures, while noting the vibrant partner ecosystem, from DDN’s mobile AI factory to AWS’s AI-themed haunted house. The discussion underscores a pivotal shift: AI is no longer just about compute, but about entire industrial-scale systems designed for agent-driven workloads and global deployment. Key takeaways include: 1) NVIDIA is evolving into an AI factory company, not just a hardware vendor; 2) The future of AI infrastructure lies in heterogeneous, photonic-connected systems; 3) OpenClaw’s explosive adoption signals a new era of autonomous AI agents; 4) Building AI factories requires planning for long-term TCO and lifecycle management; 5) The convergence of AI, quantum interconnects (via NVQ-Link), and open ecosystems is redefining HPC. The overall sentiment is highly positive, reflecting excitement and awe at the scale and ambition of NVIDIA’s vision, while acknowledging the complexity and long-term challenges ahead.