Pre-validated, NVIDIA-certified architectures eliminate months of integration work. Go from data to inference-ready pipeline in weeks. 

One ONTAP data fabric spans on-premises, hybrid cloud, and edge — eliminating silos that slow AI model training and RAG pipelines. StorageGRID extends that fabric with an S3-native object tier, so exabyte-scale training corpora sit inside the same namespace rather than in a separate data lake.

Disaggregated architecture lets you scale storage performance and capacity independently — match resources to AI demand without over-provisioning. 

Autonomous Ransomware Protection, multi-tenancy, RBAC, audit logging, and automated data masking built in from day one — not bolted on.

From raw data discovery and governance through vectorization, RAG, and agentic AI endpoints — NetApp covers the full pipeline. 

Trusted by the world's largest enterprises. Co-engineered with NVIDIA, Cisco, and Intel — validated at every layer of the stack. 

NetApp AFX is a revolutionary disaggregated all-flash storage system that decouples compute from capacity. Every controller node accesses every drive across a shared Ethernet fabric — enabling linear performance scaling that no traditional shared-nothing architecture can match. 

Purpose-engineered for the demands of AI workloads: massive sequential throughput for LLM training, sub-millisecond latency for real-time inferencing, and nondisruptive scale-out as data volumes grow. 

Key Capabilities

• Scales to 128 nodes with linear performance gains — delivering terabytes per second of throughput
• Full ONTAP data services: SnapMirror, FlexClone, FlexCache, Autonomous Ransomware Protection from day one
• NFS over RDMA for ultra-low latency AI/ML data pipelines and GPU cluster connectivity
• NVIDIA DGX SuperPOD certified — co-engineered for DGX GB300 deployments
• Secure multi-tenancy and nondisruptive scaling for shared AI infrastructure
• Deployed live in WWT's AI Proving Ground (8-node cluster with 4 shelves)
• Tiers cold blocks to StorageGRID via FabricPool — flash stays reserved for what the GPUs are actually reading

Use Cases

• LLM training
• AI/ML pipelines
• GPU cluster storage
• HPC workloads

An AI factory needs two tiers of storage. AFX delivers the throughput a GPU cluster consumes while it trains. StorageGRID holds everything else — the raw corpus, archived checkpoints, inference history, and the images, documents, and sensor output that make up the overwhelming majority of an enterprise's unstructured data and that no organization can afford to keep on flash.

StorageGRID is NetApp's S3-native object storage platform: a single global namespace spanning sites and geographies, with a metadata-driven policy engine that places every object on the right medium in the right location automatically. AI and analytics frameworks address it exactly as they address public cloud object storage — same S3 API, same access patterns — while the data stays on infrastructure you own and inside boundaries you set.

Key Capabilities

• S3-native object storage at exabyte scale — the API AI and analytics frameworks already target, without egress charges on every training run
• Single global namespace across sites and regions, with geo-distributed erasure coding for durability without full replication
• Metadata-driven information lifecycle management places objects by policy — performance media for active training data, low-cost media for archive
• Object Lock (WORM) immutability protects training corpora and model artifacts from ransomware and accidental deletion
• Serves as the FabricPool capacity tier for ONTAP and AFX — cold blocks tier off flash into the same object store the AI pipeline already reads
• Secure multi-tenancy lets multiple AI teams share one repository under separate credentials, quotas, and audit trails
• Business outcome: keep the entire data estate available to AI, not just the slice that fits on flash. Training sets, model artifacts, and inference history stay online, addressable, and governed at a cost per terabyte that makes retaining everything defensible — which matters, because the dataset you delete is the one the next model needed.

Use Cases

• AI data lakes
• Training corpora
• Model and checkpoint archive
• FabricPool object tier
• Multi-site AI

AI models are only as good as the data they are handed, and this is precisely where enterprise AI initiatives stall. The model works. What nobody can say is whether it was given the right data, or whether the answer it produced can be trusted.

AIDE sits between your data and your models. It continuously organizes, protects, and adds context to information as that information changes — so when a model or an agent asks a question, the answer comes from the current, complete, permitted set of data rather than a stale export somebody made six months ago. Right data means relevance and freshness. Trust means provenance: knowing where an answer came from, that the source was cleared for use, and that sensitive content was masked before it ever reached the model.

That is the difference between an AI pilot that impresses a steering committee and a production system the business is willing to run on.

Mechanically, AIDE (AI Data Engine) is NetApp's storage-integrated AI data service that spans the entire AI lifecycle — from discovering and preparing raw data through powering RAG, agentic AI, and generative AI endpoints. Built on the NVIDIA AI Data Platform reference design.

AIDE collapses multiple data preparation and management steps into a single, storage-native service — eliminating the data engineering overhead that delays most enterprise AI projects.

Key Capabilities

• Global metadata catalog continuously analyzes content and enriches context beyond basic file labels
• Semantic search and data vectorization collapse multiple data prep steps into a single service
• Data change detection and synchronization eliminate redundant copies and keep retrieval indexes fresh
• Governance: RBAC, automated redaction, masking, audit logging, and policy-driven enforcement
• Native RAG and GenAI endpoint integration — connects directly to AI applications
• Built on NVIDIA AI Enterprise software and accelerated compute stack
• Business outcome: remove the data engineering that sits between a dataset and a working model, and give the business a defensible reason to trust the output — answers traceable to a governed source, sensitive content masked before the model sees it, permissions enforced at retrieval time. Governance stops being the thing that blocks deployment and becomes the thing that makes deployment approvable.

Use Cases

• RAG capabilities
• GenAI applications
• Data governance
• Agentic AI
• Trusted retrieval

Announced at Cisco Live 2026 (June 3, Las Vegas), FlexPod AI is the evolution of the industry's most trusted converged infrastructure platform, now validated end-to-end for enterprise AI workloads. Combines Cisco UCS compute and Nexus networking with NetApp storage and AIDE data services. 

New tiered configurations support every stage of AI adoption: large enterprise training clusters, departmental inferencing, RAG workloads, and edge environments running containerized and virtualized workloads. 

Key Capabilities

• Pre-validated, pre-tested stack from edge to enterprise — reduces deployment risk and time
• Tiered configurations for enterprise training, departmental inferencing, and edge AI
• Splunk SOAR storage response integration for AI-driven cyber resilience
• Support for RAG, semantic search, and containerized AI workloads out of the box
• Cisco security + NetApp ARP = defense-in-depth for AI data protection
• Backed by a joint Cisco + NetApp solution engineering and support organization

Use Cases

• Enterprise AI
• Department interfacing
• Edge AI
• Hybrid cloud

Co-developed with Intel, AIPod Mini brings enterprise AI inferencing and RAG to the department level. Purpose-built for teams on the front lines of innovation — manufacturing floors, retail operations, legal departments — where AI value needs to be realized quickly without a data center build-out. 

AIPod Mini offers an affordable, secure, and easy-to-deploy path to AI that remains adaptable as organizational requirements evolve — and connects seamlessly to enterprise AFX when demand outgrows the appliance. 

Key Capabilities

• Powered by Intel Xeon 6 processors with integrated AI acceleration
• Prepackaged RAG workflows for immediate time to value — deploy and run on day one
• Affordable, compact, and easy to deploy — no specialized infrastructure team required
• Integrated NetApp data management with ONTAP security and data services
• Adaptable to evolving AI requirements — scales up or connects to enterprise AFX
• Available as a WWT lab environment for hands-on validation before purchase

Use Cases

• Department RAG
• Edge inferencing
• Manufacturing AI
• Retail/Legal AI