Home / Technology / NetApp Unveils World's Fastest Storage for AI Factories
NetApp Unveils World's Fastest Storage for AI Factories
29 Sep
Summary
- NetApp Novus offers unprecedented speed for large AI GPU environments.
- It is designed to prevent bottlenecks and maximize GPU utilization.
- The architecture supports zettabyte-scale file systems and 100 TB/s performance.
NetApp has introduced NetApp Novus, a new storage architecture engineered to be the fastest on the planet for AI infrastructure providers managing extensive GPU environments. This cutting-edge solution is designed to support zettabyte-scale file systems, ensuring that large AI factories can maintain continuous data flow to millions of GPUs.
The Novus architecture specifically targets the bottlenecks that traditional storage systems create in massive GPU deployments. By separating metadata from the data path, it allows performance, capacity, and concurrency to scale independently. This approach aims to maximize GPU utilization, which can otherwise drop significantly due to insufficient data delivery.
NetApp states that Novus can achieve throughput exceeding 100 terabytes per second. It combines NetApp Novus Data Director, the metadata software, with NetApp ONTAP data services on AFF A90 systems. This disaggregated design is built to support hundreds of thousands of GPUs while preserving trusted enterprise data services.
Key benefits for adopters include keeping every GPU productive by reducing storage-related bottlenecks, scaling data under a single namespace without fragmentation, and improving overall AI infrastructure efficiency by allowing operators to scale only necessary resources. This solution is orderable today and provides a path toward software-defined deployments.
