.jpg)
NVIDIA has officially launched the Rubin platform, its next-generation AI supercomputing architecture, comprising six new chips engineered through extreme codesign. Unveiled at CES 2026, Rubin is designed to drastically reduce the cost and complexity of building and deploying massive-scale AI systems, promising up to a 10x reduction in inference token cost and requiring 4x fewer GPUs to train mixture-of-experts models compared to its Blackwell predecessor, accelerating mainstream AI adoption.
NVIDIA launches the Rubin platform, its next-gen AI supercomputing architecture built on six new chips.
Promises up to 10x lower inference token cost and 4x fewer GPUs needed for training MoE models vs. Blackwell.
The six chips are: Vera CPU, Rubin GPU, NVLink 6 Switch, ConnectX-9 SuperNIC, BlueField-4 DPU, and Spectrum-6 Ethernet Switch.
Designed for agentic AI, advanced reasoning, and gigascale inference with massive context memory.
Backed by broad ecosystem support from AWS, Google, Microsoft, Meta, OpenAI, xAI, Dell, HPE, and Lenovo.
Rubin-based products from partners will be available in the second half of 2026.
The Rubin platform represents NVIDIA's annual cadence in delivering a new generation of AI supercomputers, engineered to meet skyrocketing demand for both training and inference. By employing extreme codesign across hardware and software—spanning compute, networking, and storage—Rubin aims to set a new standard for performance per watt and total cost of ownership for building the world's most advanced AI systems, from large-scale training clusters to inference factories.
The announcement was met with unprecedented endorsements from across the technology and AI landscape, underscoring Rubin's expected impact.
Sam Altman, CEO of OpenAI: “Intelligence scales with compute... The NVIDIA Rubin platform helps us keep scaling this progress so advanced intelligence benefits everyone.”
Satya Nadella, executive chairman and CEO of Microsoft: “With the addition of NVIDIA Vera Rubin GPUs, we will empower developers and organizations to create, reason and scale in entirely new ways.”
Elon Musk, founder and CEO of xAI: “NVIDIA Rubin will be a rocket engine for AI. If you want to train and deploy frontier models at scale, this is the infrastructure you use.”
Mark Zuckerberg, founder and CEO of Meta: “NVIDIA’s Rubin platform promises to deliver the step-change in performance and efficiency required to deploy the most advanced models to billions of people.”
The platform introduces five key innovations engineered for complex, multi-step AI workloads:
Sixth-Gen NVLink: Delivers 3.6TB/s per GPU and 260TB/s per rack for seamless communication in massive models.
NVIDIA Vera CPU: A power-efficient Arm-based CPU designed as the control plane for large-scale AI factories.
Rubin GPU: Features a 3rd-gen Transformer Engine delivering 50 petaflops of NVFP4 compute for inference.
3rd-Gen Confidential Computing: Provides data security across CPU, GPU, and NVLink domains at rack-scale.
2nd-Gen RAS Engine: Enhances real-time health monitoring, fault tolerance, and serviceability for maximum uptime.
Rubin is backed by a comprehensive ecosystem. Major cloud providers like AWS, Google Cloud, Microsoft Azure, and OCI, along with NVIDIA Cloud Partners like CoreWeave, plan to deploy instances in late 2026. System manufacturers including Cisco, Dell, HPE, Lenovo, and Supermicro will deliver Rubin-based servers. Furthermore, leading AI labs and software partners are aligning their roadmaps with the platform. NVIDIA also highlighted an expanded collaboration with Red Hat to deliver a complete, optimized AI software stack for the Rubin platform.
About NVIDIA
NVIDIA is the world leader in AI and accelerated computing.
TechIntelPro empowers B2B technology decision-makers with the latest industry trends, deep insights, leadership perspectives and actionable intelligence to drive both organizational and professional milestones—and thrive on the tech pulse.