NTAP

NetApp builds the storage systems that hold the data AI models need — training datasets, model checkpoints, inference indices, and the vast unstructured data lakes that enterprises are organizing for AI applications. The company has been in enterprise storage for over 30 years, and its ONTAP storage operating system is the industry standard for enterprise data management, running on both on-premises arrays and in public clouds (AWS FSx for NetApp ONTAP, Azure NetApp Files, Google Cloud Volumes).

The physical products are rack-mounted storage arrays containing dozens to hundreds of NVMe flash drives, connected to server clusters via high-speed Ethernet or Fibre Channel. NetApp's AFF (All Flash FAS) systems deliver millions of IOPS with sub-millisecond latency — the performance tier that GPU clusters demand for feeding training data. A single AFF A900 array can hold up to 1.5 petabytes of flash storage. An AI training cluster might deploy 10-20 of these arrays to keep thousands of GPUs fed without stalling.

NetApp's AI positioning is enterprise-focused rather than hyperscaler-focused. Approximately 300 customers selected NetApp for AI data preparation and storage as of Q3 FY2026. The company's value proposition is data management — not raw storage performance — offering unified storage that serves AI workloads alongside traditional enterprise applications from a single platform with ONTAP software. Keystone Storage-as-a-Service (growing 65% YoY) reflects the consumption model shift.

Q3 FY2026 revenue was $1.71B (+4% YoY), with all-flash array revenue hitting a record $1B (+11% YoY). Non-GAAP operating margin was 31.1%. FY2026 guidance is ~$6.75B with EPS of $7.92-8.02. The company generates strong free cash flow and returns capital through buybacks and dividends. This is a profitable, stable enterprise infrastructure business with moderate AI exposure, not a hyper-growth AI play.