Skip to main content

Rafay outlines AI factory stack with Aviz ONES and Spectrum-X

2 companies named across 9 categories, one of 856 articles referencing Aviz Networks. Previous coverage: Aviz Networks details Diameter GX session visibility (Sep 2026).

Companies mentioned

Best suited for

Seniority
Director
Job function
IT Infrastructure
Persona
Platform / Infrastructure Architect
Buyer role
Architect / Technical Evaluator
Buyer journey
Need to Buy
Adoption curve
Early Majority
Technology maturity
Operational Expansion
Industry
Information Technology / Software & Services / IT Services / Cloud Services

Our classification, not the publisher's statement. Best suited for, not only for.

The blog describes how Rafay, Aviz ONES, and NVIDIA Spectrum-X are combined to support multi-tenant AI factories and neocloud environments. It presents the stack as a way to add governance, observability, and GPU utilization controls that matter for enterprise platform teams.

Research Overview

The post frames AI factories as infrastructure built around high-density GPU fabrics for training and inference. It argues that GPU capacity alone does not address enterprise requirements for shared service delivery across tenants.

To close that gap, the article centers on a fabric-to-cloud approach that connects network fabric performance with workload scheduling and governance. It presents the partnership as a unified set of components for private, hybrid, and sovereign environments.

Product Update

Rafay Platform is described as the scheduler, governance layer, and scale platform for multi-tenant service delivery in AI factories and neoclouds. The blog says it orchestrates cloud-based AI fabrics together with NVIDIA Spectrum-X Ethernet and Aviz ONES fabric orchestration.

The article lists several capabilities, including lifecycle orchestration, secure multi-tenancy, policy and cost governance, GPU-aware allocation, workload orchestration, and ecosystem integration. It says the platform supports deployment, scaling, and upgrades across private, hybrid, and sovereign computing environments.

Technical Breakdown

NVIDIA Spectrum-X is presented as the lossless Ethernet fabric layer with SuperNICs. Aviz ONES is positioned as the component handling GPU-aware fabric orchestration, tenant segmentation, and lifecycle automation.

Rafay is shown as the layer that turns networked GPUs into governed compute across bare metal, Kubernetes, or virtual machines. The blog also says the combined stack provides telemetry for link health, ECMP balance, and GPU utilization from a single view.

Operational Impact

The post says the stack is intended to make GPU use more predictable, more shareable across tenants, and easier to attribute for billing and chargebacks. It also says the model can support self-service access to GPU clusters and AI workbenches through governed catalogs.

Use cases listed in the blog include AI/ML training and inference, enterprise AI factory deployments, and GPU platform-as-a-service. The article says these use cases rely on scheduling, lifecycle automation, and cost controls rather than manual handling by platform teams.

Leadership Perspective

The blog’s broader message is that GPU infrastructure can be managed as a service rather than a fixed cost. It says that approach can support faster delivery and better utilization across the AI infrastructure stack.

The post closes by describing the reference architecture as a secure, multi-tenant AI platform for enterprise environments. This Blog Summary is a fact-based summary of the vendor blog.

Blog post, originally published by Ram Mohan Hariprasad at aviznetworks.com.