From AI factory blueprint to running an AI cloud.
NVIDIA sets the standard for how AI factories are designed and built. The Rafay Platform turns that standard into governed, self-service offerings that your customers and teams use and pay for, from bare metal to tokens.
The Rafay Platform is NVIDIA AI Cloud Ready
The Rafay Platform meets NVIDIA’s software standard for operating production-grade AI cloud infrastructure. With NVIDIA Infra Controller, it gives neocloud and sovereign AI cloud operators a validated AI cloud platform from day one. Rafay is among the first independent software vendors to earn the designation.
READ THE ANNOUNCEMENT →NVIDIA provides the AI factory components. The Rafay Platform runs them as a cloud business.
NVIDIA DSX and the NCP reference architecture give every operator the same blueprint for AI factory hardware and software. Turning that blueprint into revenue takes an operational layer on top.
The Rafay Platform is that layer. It brings NVIDIA infrastructure online, isolates each tenant, and delivers bare metal, VMs, Kubernetes, Slurm, Run:ai and token services from one self-service portal, with metering and billing on every service.
From GPU owner to NVIDIA Cloud Partner
NVIDIA cloud partner programs expect governed, multi-tenant services on NVIDIA reference designs. The Rafay Platform provides that software layer, so operators can meet each requirement without building it themselves.
READ THE NCP GUIDE →Recognized across the NVIDIA ecosystem
Few platforms work this closely with NVIDIA. The Rafay Platform is validated and certified across the stack, from GPU Kubernetes and virtualization to AI cloud software and enterprise AI factories.
How Rafay turns each NVIDIA component into a service
NVIDIA DSX brings reference designs and software together to design, simulate, build and operate AI factories. DSX OS is its open-source, modular software for running multi-tenant AI factory infrastructure. The Rafay Platform turns each DSX OS capability into a service operators can offer, with governance, quotas and audit logging on every one.
Why rack-scale systems need a new operating layer
NVIDIA NVL72 systems change what an AI cloud operator has to manage. The same rack-scale design carries forward to GB300 and Vera Rubin NVL72.
How it comes together
The Rafay Platform runs on NVIDIA DSX OS and Infra Controller, and turns NVL72 racks into services operators sell from one control plane.
What Rafay brings to NVIDIA rack-scale
What operators need to turn NVL72 racks into revenue, from one control plane.
Production deployments at scale
Telcos hold power, sites and trusted networks across a country. With the Rafay Platform on NVIDIA telco reference architectures, they turn that footprint into sovereign AI clouds and token-metered inference at the edge.
Not every NVIDIA estate runs on Kubernetes alone. The Rafay Platform also delivers Slurm and GPU virtual machines as self-service, multi-tenant services on NVIDIA infrastructure.
