By Steve McDowell
Publication Date: 2026-03-31 17:49:00
NVIDIA CEO Jensen Huang
AFP via Getty Images
Nvidia unveiled two initiatives at its recent GTC 2026 event that disrupt the enterprise storage landscape. The Nvidia Data Platform and STX incorporate the entire storage ecosystem into Nvidia’s framework by standardizing data transfer to GPUs.
These efforts further reinforce Nvidia’s leadership in AI infrastructure design and are also changing how storage companies compete in that market. For IT buyers and storage vendors alike, the implications are substantial.
Nvidia AI Data Platform and STX
Nvidia’s storage strategy operates across three tiers, starting with a basic certification program to guarantee interoperability. Nvidia-Certified Storage verifies minimum performance standards and became essential for AI storage systems in 2025.
The Nvidia AI Data Platform provides a reference design that extends into data services, integrating Nvidia software such as NeMo Retriever microservices for RAG, the AI-Q Blueprint for query agents, and the Dynamo inference library directly into the storage stack.
The third and most consequential tier is STX, which standardizes the data path between storage and GPUs, going beyond interface compatibility to provide a complete architectural framework. Its rack-scale reference design, built on BlueField-4 DPUs, Vera Rubin accelerators, and Spectrum-X Ethernet networking, is designed for deploying agentic AI inference workloads at scale.
Nvidia has nearly all traditional enterprise storage…