NVIDIA’s Next-Gen Vera Rubin AI Platform Now in Full Operation
NVIDIA has set the stage at CES 2026 with a significant announcement. The tech giant’s next-generation Vera Rubin AI platform has now entered full production, surpassing the initially planned H2 2026 timeline. This Rubin platform signifies NVIDIA’s first-ever “extreme-codesigned” system, amalgamating six new chips to function as a cohesive AI supercomputer.
Christened after the trailblazing astronomer Vera Rubin, the platform fuses the Vera CPU and Rubin GPU with four additional networking and storage chips. These include the NVLink 6 Switch, ConnectX-9 SuperNIC, BlueField-4 DPU, and Spectrum-6 Ethernet Switch. NVIDIA asserts that the Vera Rubin NVL72 rack-scale system offers up to a 10x reduction in inference token costs and delivers “more bandwidth than the entire internet.”
“Rubin arrives at exactly the right moment, as AI computing demand for both training and inference is skyrocketing,” stated NVIDIA CEO Jensen Huang during his CES keynote. He further revealed that leading cloud providers such as AWS, Google Cloud, Microsoft, and Oracle are set to be among the first to deploy Vera Rubin-based instances in the latter half of 2026.
Moreover, AI labs including OpenAI, Anthropic, Meta, and xAI are projected to utilize the platform for training larger, more competent AI models and implementing advanced reasoning systems.
Source: NVIDIA News