Nvidia Launches Vera Rubin Architecture at CES 2026 with Major Performance Gains
Nvidia CEO Jensen Huang unveiled the revolutionary Rubin computing architecture at CES 2026, featuring a six-chip system that delivers 3.5x faster AI training and 5x faster inference performance compared to Blackwell. The architecture, now in full production, combines Vera CPU and dual Rubin GPUs with advanced networking components, targeting major cloud providers including AWS, Anthropic, and OpenAI for deployment in the second half of 2025.

*this image is generated using AI for illustrative purposes only.
Nvidia CEO Jensen Huang has officially launched the company's revolutionary Rubin computing architecture at the Consumer Electronics Show (CES) 2026 in Las Vegas, marking a significant milestone in AI chip development. Named after astronomer Vera Florence Cooper Rubin, this next-generation architecture addresses the exponential growth in computational demands of artificial intelligence systems.
Six-Chip Architecture Design
The Rubin architecture represents a comprehensive six-chip system that combines cutting-edge components for maximum performance efficiency. "The Rubin platform uses extreme codesign across the six chips to slash training time and inference token costs," according to Nvidia.
| Component: | Specification |
|---|---|
| CPU: | NVIDIA Vera CPU (88 cores, designed for agentic reasoning) |
| GPU: | NVIDIA Rubin GPU (2 units per system) |
| Networking: | NVIDIA NVLink 6 Switch |
| SuperNIC: | NVIDIA ConnectX-9 SuperNIC |
| DPU: | NVIDIA BlueField-4 DPU |
| Ethernet Switch: | NVIDIA Spectrum-6 Ethernet Switch |
Performance Breakthrough Over Blackwell
Nvidia's internal testing demonstrates substantial performance improvements compared to the current Blackwell generation, establishing new benchmarks for AI processing capabilities.
| Performance Metric: | Rubin vs Blackwell Improvement |
|---|---|
| AI Training Performance: | 3.5x faster |
| AI Inference Performance: | 5x faster |
| Peak Performance: | 50 petaflops |
| Inference Compute Efficiency: | 8x more per watt |
| Operational Cost: | Lower cost per result using fewer components |
The enhanced performance addresses evolving AI system requirements, particularly for specialized networks processing massive datasets through multistage processes. Huang emphasized during his presentation that "Vera Rubin is designed to address this fundamental challenge that we have: The amount of computation necessary for AI is skyrocketing."
Production Status and Market Deployment
The Rubin architecture has moved beyond testing phases into full-scale production. "Today, I can tell you that Vera Rubin is in full production," Huang announced to the CES audience, with expansion expected throughout the second half of the year.
| Deployment Information: | Details |
|---|---|
| Production Status: | Full production active |
| Timeline: | Second half 2025 expansion |
| Early Customers: | Amazon Web Services, Anthropic, OpenAI |
| Supercomputer Integration: | HPE's Blue Lion, Doudna at Lawrence Berkeley National Lab |
| Product Availability: | DGX SuperPod systems and modular components |
Advanced Storage and Infrastructure Solutions
The architecture incorporates significant improvements in storage and interconnection capabilities through enhanced Bluefield and NVLink systems. Nvidia's senior director of AI infrastructure solutions, Dion Harris, highlighted critical storage innovations: "As you start to enable new types of workflows, like agentic AI or long-term tasks, that puts a lot of stress and requirements on your KV cache."
The new Vera CPU specifically targets agentic reasoning applications, while the dual Rubin GPUs provide unprecedented parallel processing capabilities for complex AI workloads.
Strategic Market Impact
Nvidia's accelerated development cycle has established the company as the world's most valuable corporation, with the Rubin architecture positioned to further strengthen its AI market dominance. The comprehensive six-chip approach represents a departure from traditional single-component upgrades, offering integrated solutions for next-generation AI applications across robotics, healthcare, and heavy industry sectors.
Major cloud providers and research institutions are already integrating Rubin technology into their infrastructure plans, indicating strong market confidence in the architecture's capabilities and Nvidia's continued technological leadership in the rapidly evolving artificial intelligence landscape.



























