NVIDIA's New Platform Vera Rubin Is Here

NVIDIA CEO Jensen Huang announced at CES 2026 that the company's next-generation Vera Rubin platform is now in full production and will ship to customers including Microsoft and Amazon in the second half of 2026. 

The announcement accelerates the timeline by six months compared to NVIDIA's previously stated schedule and just shows that demand for next-generation AI infrastructure is so urgent the company is ramping production ahead of schedule.

The Vera Rubin platform comprises six integrated chips, in deluging the Vera CPU, Rubin GPU, NVLink 6 switch, ConnectX-9 SuperNIC, BlueField-4 DPU, and Spectrum-6 Ethernet switch, designed to function as a unified AI supercomputer rather than as independent components. 

The architecture will lead to what NVIDIA claims a 10x reduction in inference token cost compared to the previous Blackwell generation and enables training of Mixture of Experts models using only 25% of the GPUs that Blackwell required.

Companies like Microsoft and Amazon making decisions to deploy Rubin at scale are betting that agentic AI, long-context reasoning, and massive-scale inference are the dominant workloads of 2026 and beyond.