Enterprise storage from leading NVIDIA partners, certified to deliver performance, security, and scale.
Overview
With NVIDIA-Certified Storage, enterprises and cloud providers can confidently deploy storage solutions that deliver the performance, security, and scale required for today’s agentic AI and physical AI workloads.
Benefits
NVIDIA-Certified Storage is rigorously benchmarked against critical input and output (IO) profiles. Simulating loads up to 10,000 GPUs, it ensures the strict throughput and latency demands of modern AI models are met without becoming a bottleneck.
Certified storage platforms go through a suite of functional tests to ensure they meet the highest standards of scalability; quality of service; reliability, availability, and serviceability (RAS); and multi-tenancy.
By verifying that the storage layer can efficiently feed data to accelerated computing resources, users avoid "starving" their GPUs. This leads to faster model training, lower latency inference, and better overall infrastructure utilization.
Certification Types
The NVIDIA-Certified Storage program includes certifications for both general-purpose storage, suitable for a variety of AI and accelerated computing applications, and special-purpose systems designed to address the specific storage needs of an AI pipeline.
Partners
NVIDIA-Certified Storage is available globally from leading data platform and storage providers.
Find an NVIDIA-Certified Storage solution.
AI Factories
An AI factory is specialized infrastructure designed to meet the massive compute demands of processes like data ingestion, training, fine-tuning, and high-volume inference. NVIDIA-Certified Storage systems are key to achieving the best performance of AI factory infrastructure, whether used as part of an enterprise reference architecture or an NVIDIA Cloud Partner.