NVIDIA Vera Rubin Platform Now in Production — 10x Agentic Throughput vs Blackwell, Shipping Fall 2026
NVIDIA Newsroom·high signal
NVIDIA announced at Computex that Vera Rubin is in full production on TSMC 3nm, combining an 88-core Vera CPU with next-gen Rubin GPU. The platform delivers 10x agentic AI throughput, 10x reduction in inference token cost, and 4x reduction in GPUs needed to train MoE models versus Blackwell. AWS, Google Cloud, Microsoft, OCI, CoreWeave, Lambda, and Nebius will be among the first to deploy Vera Rubin instances in late 2026.