AI deployments thrive on pace, knowledge and scale. That’s why NVIDIA is increasing NVIDIA-Licensed Programs to incorporate enterprise storage certification — for streamlined AI manufacturing facility deployments within the enterprise with accelerated computing, networking, software program and storage.
As enterprises construct AI factories, entry to high-quality knowledge is crucial to make sure optimum efficiency and reliability for AI fashions. The brand new NVIDIA-Licensed Storage program introduced at present on the NVIDIA GTC international AI convention validates that enterprise storage methods meet stringent efficiency and scalability knowledge necessities for AI and high-performance computing workloads.
Main enterprise knowledge platform and storage suppliers are already onboard, making certain companies have trusted choices from day one. These embody DDN, Dell Applied sciences, Hewlett Packard Enterprise, Hitachi Vantara, IBM, NetApp, Nutanix, Pure Storage, VAST Knowledge and WEKA.
Constructing Blocks for a New Class of Enterprise Infrastructure
At GTC, NVIDIA additionally introduced the NVIDIA AI Knowledge Platform, a customizable reference design to construct a brand new class of enterprise infrastructure for demanding agentic AI workloads.
The NVIDIA-Licensed Storage designation is a prerequisite for companions creating agentic AI infrastructure options constructed on the NVIDIA AI Knowledge Platform. Every of those NVIDIA-Licensed Storage companions will ship personalized AI knowledge platforms, in collaboration with NVIDIA, that may harness enterprise knowledge to cause and reply to complicated queries.
NVIDIA-Licensed was created greater than 4 years in the past because the trade’s first certification program devoted to tuning and optimizing AI methods to make sure optimum efficiency, manageability and scalability. Every NVIDIA-Licensed system is rigorously examined and validated to ship enterprise-grade AI efficiency.
There at the moment are 50+ companions offering 500+ NVIDIA-Licensed methods, serving to enterprises cut back time, price and complexity by giving them a wide array of performance-optimized methods to energy their accelerated computing workloads.
NVIDIA Enterprise Reference Architectures (RAs) had been launched final fall to supply companions with AI infrastructure greatest practices and configuration steerage for deploying NVIDIA-Licensed servers, NVIDIA Spectrum-X networking and NVIDIA AI Enterprise software program.
Options primarily based on NVIDIA Enterprise RAs can be found from the world’s main methods suppliers to cut back the time, price and complexity of enterprise AI deployments. Enterprise RAs at the moment are obtainable for a variety of NVIDIA Hopper and NVIDIA Blackwell platforms, together with NVIDIA HGX B200 methods and the brand new NVIDIA RTX PRO 6000 Blackwell Server Version GPU.
These NVIDIA applied sciences and companion options are the constructing blocks for enterprise AI factories, representing a brand new class of enterprise infrastructure for high-performance AI deployments at scale.
Enterprise AI Wants Scalable Storage
Because the tempo of AI innovation and adoption accelerates, safe and dependable entry to high-quality enterprise knowledge is turning into extra vital than ever. Knowledge is the gasoline for the AI manufacturing facility. With enterprise knowledge creation projected to succeed in 317 zettabytes yearly by 2028*, AI workloads require storage architectures constructed to deal with large, unstructured and multimodal datasets.
NVIDIA’s expanded storage certification program is designed to satisfy this want and assist enterprises construct AI factories with a basis of high-performance, dependable knowledge storage options. This system consists of efficiency testing in addition to validation that companion storage methods adhere to design greatest practices, optimizing efficiency and scalability for enterprise AI workloads.
NVIDIA-Licensed Storage will likely be integrated into NVIDIA Enterprise RAs, offering enterprise-grade knowledge storage for AI manufacturing facility deployments with full-stack options from international methods companions.
Licensed Storage for Each Deployment
This certification builds on present NVIDIA DGX methods and NVIDIA Cloud Companion (NCP) storage packages, increasing the information ecosystem for AI infrastructure.
These storage certification packages are aligned with their deployment fashions and architectures:
- NVIDIA DGX BasePOD and DGX SuperPOD Storage Certification — designed for enterprise AI manufacturing facility deployments with NVIDIA DGX methods.
- NCP Storage Certification — designed for large-scale NCP Reference Structure AI manufacturing facility deployments with cloud suppliers.
- NVIDIA-Licensed Storage Certification — designed for enterprise AI manufacturing facility deployments with NVIDIA-Licensed servers obtainable from international companions, primarily based on NVIDIA Enterprise RA pointers.
With this framework, organizations of all sizes — from cloud hyperscalers to enterprises — can construct AI factories that course of large quantities of information, practice fashions sooner and drive extra correct, dependable AI outcomes.
Be taught extra about how NVIDIA-Licensed Programs ship seamless, high-speed efficiency and attend these associated classes at GTC:
*Supply: IDC, Worldwide IDC World DataSphere Forecast, 2024–2028: AI In every single place, However Upsurge in Knowledge Will Take Time, doc #US52076424, Might 2024