Nutanix has announced that its Nutanix Unified Storage (NUS) solution has achieved NVIDIA-Certified Storage status at the enterprise level, a milestone aimed at helping organizations accelerate production-scale AI deployments while maximizing infrastructure performance and efficiency.
The certification validates Nutanix Unified Storage for interoperability with NVIDIA-powered AI infrastructure, enabling enterprises and cloud providers to deploy AI environments with greater confidence. As organizations increasingly invest in AI factories to support large-scale training, inference, and agentic AI workloads, the ability to move and process vast amounts of data efficiently has become critical to achieving optimal GPU utilization and operational scalability.
“As organizations across the Middle East and Africa accelerate their AI ambitions, the conversation is rapidly shifting from experimentation to production-scale AI deployments. Enterprises need infrastructure that can deliver consistent performance, efficient data movement, and simplified operations at scale.” — Ahmed Rashad, Sr. AI Specialist, Middle East & Africa, Nutanix
Nutanix said the certification addresses key challenges facing AI deployments, including fragmented infrastructure, data silos, integration complexity, and storage bottlenecks that can limit AI performance. By providing a validated reference architecture, the company aims to help organizations reduce deployment risk while ensuring reliable, high-speed access to data across AI environments.
Built on a 10-node all-NVMe cluster, the certified solution leverages enhanced parallel NFS (pNFS) and GPUDirect Storage over NFS with RDMA to create a low-latency, high-throughput data path between GPUs and storage systems. The architecture is designed to support demanding AI workloads while maintaining predictable storage performance as deployments scale.
According to Nutanix, the platform delivers linear scalability from 10 GB/s read and 5 GB/s write throughput for 32 GPUs to 160 GB/s read and 80 GB/s write throughput for deployments supporting up to 1,024 GPUs. The solution incorporates NVIDIA Spectrum-X Ethernet networking, including Spectrum-4 switches and BlueField-3 DPUs, to optimize data movement across AI infrastructure.
The certified architecture supports a broad range of AI use cases, including model training, fine-tuning, inference, and retrieval-augmented generation (RAG) workloads. It is compatible with multiple compute environments, including NVIDIA RTX 6000 PRO Blackwell GPUs, NVIDIA H200 NVL systems, NVIDIA HGX servers, and NVIDIA GH200 Grace Hopper Superchip platforms.
Nutanix also revealed plans to support NVIDIA BlueField-4 STX technology in the second half of 2026, further advancing its AI-native storage strategy. The NVIDIA-certified Nutanix Unified Storage reference architecture is available immediately, providing enterprises with a scalable foundation for next-generation AI operations.
