NVIDIA Vera Rubin Platform Enters Production with Seven-Chip AI Infrastructure for Agentic Workloads at GTC 2026

NVIDIA announced at the GTC 2026 conference in San Jose on March 16 the Vera Rubin platform, with seven new chips now in full production. The platform provides configurable AI infrastructure optimized for pretraining, post-training, test-time scaling, and agentic inference phases. The Vera Rubin platform integrates the NVIDIA Vera CPU, NVIDIA Rubin GPU, NVIDIA NVLink 6 Switch, NVIDIA ConnectX-9 SuperNIC, NVIDIA BlueField-4 DPU, NVIDIA Spectrum-6 Ethernet switch, and the newly integrated NVIDIA Groq 3 LPU. These components function as a unified AI supercomputer.

The platform comprises five rack types. Vera Rubin NVL72 GPU racks integrate 72 Rubin GPUs and 36 Vera CPUs connected by NVLink 6, along with ConnectX-9 SuperNICs and BlueField-4 DPUs. They train large mixture-of-experts models using one-fourth the GPUs compared to the NVIDIA Blackwell platform and achieve up to 10x higher inference throughput per watt at one-tenth the cost per token. Scaling occurs with NVIDIA Quantum-X800 InfiniBand and Spectrum-X Ethernet.

Vera CPU racks deliver dense, liquid-cooled infrastructure on NVIDIA MGX with 256 Vera CPUs for reinforcement learning and agentic AI workloads. They provide twice the efficiency and 50% faster performance than traditional CPUs, integra...