NVIDIA Unveils Vera Rubin Platform and Nemotron 3 Suite at Berlin GTC, Targeting 10x Inference Cost Reduction

NVIDIA announced a sweeping set of infrastructure and AI model advances at its GTC Berlin keynote on October 21, positioning itself to dominate the emerging agentic AI and physical AI markets. Jensen Huang unveiled the Vera Rubin platform, a next-generation compute system that delivers 10x lower cost per token for inference workloads compared to Blackwell, alongside a raft of open-source Nemotron language models and partnerships spanning robotics, autonomous vehicles, and creative AI. The announcements signal a strategic shift toward inference-heavy workloads as enterprises begin deploying autonomous AI agents at scale.