Nvidia Kicks Off the Next Generation of AI with Rubin – Six New Chips
4 months ago
- #AI
- #NVIDIA
- #Rubin
- NVIDIA Rubin platform delivers up to 10x reduction in inference token cost and 4x reduction in GPUs needed for training MoE models compared to NVIDIA Blackwell.
- NVIDIA Spectrum-X Ethernet Photonics switch systems offer 5x improved power efficiency and uptime.
- New NVIDIA Inference Context Memory Storage Platform with BlueField-4 accelerates agentic AI reasoning.
- Microsoft’s Fairwater AI superfactories will feature NVIDIA Vera Rubin NVL72 rack-scale systems.
- CoreWeave among first to offer NVIDIA Rubin, operated via CoreWeave Mission Control.
- Expanded collaboration with Red Hat to optimize Rubin platform with Red Hat Enterprise Linux, OpenShift, and AI.
- Rubin platform includes six new chips: NVIDIA Vera CPU, Rubin GPU, NVLink 6 Switch, ConnectX-9 SuperNIC, BlueField-4 DPU, and Spectrum-6 Ethernet Switch.
- Rubin introduces five innovations: NVLink 6, Transformer Engine, Confidential Computing, RAS Engine, and Vera CPU.
- Broad ecosystem support from AWS, Google, Meta, Microsoft, OpenAI, and others.
- Rubin-based products will be available from partners in the second half of 2026.