NetApp Launches Novus Storage to Unlock AI Factory Scale
Event summary
- NetApp introduced Novus, a storage architecture designed for AI factories, claiming 100TB/s throughput and zettabyte-scale file system support.
- Novus separates metadata from the data path to maximize GPU utilization, addressing bottlenecks in large-scale AI infrastructure.
- The initial release combines NetApp Novus Data Director with ONTAP data services on AFF A90 systems, available for order today.
- Analysts project Novus can scale to 100TBps sequential read throughput with dozens of exabytes of effective capacity.
The big picture
NetApp's Novus targets a critical bottleneck in AI infrastructure: storage architectures struggling to keep pace with GPU-intensive workloads. As AI factories scale, traditional storage solutions create inefficiencies, with GPU utilization dropping below 30%. Novus aims to solve this by enabling independent scaling of metadata, bandwidth, and capacity under a single namespace, potentially reshaping the economics of AI data centers.
What we're watching
- Adoption Pace
- How quickly neocloud and GPU-as-a-Service providers integrate Novus into their AI factory operations.
- Performance Validation
- Whether independent testing confirms NetApp's claims of 100TB/s throughput and zettabyte-scale capabilities.
- Competitive Response
- How existing storage providers react to Novus's disaggregated architecture and potential market disruption.
Related topics