NetApp Unveils 100 TB/s Storage Architecture as AI Infrastructure Scales

NetApp has unveiled a storage architecture designed to exceed 100 TB/s of aggregate throughput, as the company looks to address the infrastructure demands of large-scale AI deployments. 

Called Novus, the architecture is aimed at neoclouds and GPU-as-a-service providers and was announced alongside new data, cloud, and hybrid infrastructure capabilities at NetApp’s INSIGHT 2026 conference.

The announcements span different parts of the AI infrastructure stack, from moving data fast enough to keep large GPU clusters running to finding and governing enterprise data for AI applications. NetApp also announced a planned managed storage service with Oracle Cloud Infrastructure and new autonomous management capabilities for hybrid environments.

Novus separates metadata from the data path, allowing performance, capacity, and concurrency to scale independently under a single namespace. Its initial configuration combines Novus Data Director with ONTAP data services running on AFF A90 systems and qualified Supermicro infrastructure.

NetApp said Novus is designed to exceed 100 TB/s of aggregate throughput and support high utilisation across hundreds of thousands of GPUs. The system is orderable today. The company also cited Omdia modelling that projects 100 TB/s of sequential read throughput based on testing and modelling of the architecture.

“AI factories struggle and GPU economics collapse when data can't keep up,” Syam Nair, Chief Product Officer at NetApp, said in a statement.

The company has expanded its AI Data Engine to discover metadata across NetApp and non-NetApp storage, including NFS, SMB, and S3 repositories. The platform is designed to help organisations understand, govern, and operationalise enterprise data for AI workloads, while NetApp is previewing AI-powered data understanding, governance intelligence, and agentic services.

“The agentic enterprise will be defined by how quickly and safely organisations can turn their data into intelligence,” Asad Khan, Senior Vice President and General Manager, AI, at NetApp, added.

NetApp has also expanded its integration with Commvault for threat detection, cleanroom validation, and recovery workflows.

Separately, NetApp and Oracle announced a planned fully managed storage service native to Oracle Cloud Infrastructure (OCI), bringing ONTAP capabilities to OCI. General availability is planned within the next 12 months.

The company also introduced Console Autonomous Operations, Fleet Management, and AI ChatOps, alongside expanded Keystone Sovereign capabilities and a planned NetApp-Nutanix solution for virtualised workloads.