NetApp Seeks to Buy PEAK:AIO to Scale Storage for AI Workloads

NetApp wants its storage platform to serve AI systems that need to reach vast numbers of files at once. On September 25, it said it intends to acquire PEAK:AIO and bring the company’s metadata and parallel file technology into ONTAP. The deal is still pending, and the scale NetApp describes is a target rather than a delivered capability.

A different way to handle the file map

The technical focus is metadata: the information a storage system uses to keep track of files. NetApp says its planned architecture separates metadata services from the data itself, allowing those services to scale independently. PEAK:AIO would contribute metadata technology and a parallel file architecture to ONTAP, NetApp’s data-management software.

That distinction is central to NetApp’s pitch. AI clouds can have growing GPU clusters asking shared storage for data at the same time. Increasing the number of files those clusters can find and access is not simply a matter of expanding the space that holds them. NetApp says PEAK:AIO’s approach adds a global namespace—a shared view of files—and parallel access based on the NFS file-sharing standard. The aim is to help keep GPUs from waiting for data.

An upgrade path, not a replacement pitch

NetApp is presenting the acquisition as a way for existing ONTAP customers to gain more scale without abandoning the platform they use to manage data. It says the combined design is intended to preserve ONTAP’s resilience, security and operational simplicity. PEAK:AIO’s technology would address a more specialized demand: shared file access for large, simultaneous AI workloads. Whether customers get both benefits in one usable system depends on the integration NetApp has yet to deliver.

The technology has roots in PEAK:AIO’s collaborations with Los Alamos National Laboratory and Carnegie Mellon University, according to NetApp. The company also lists deployments of PEAK:AIO’s platform at research and other institutions. Those details describe PEAK:AIO’s existing work, not results for a PEAK:AIO–ONTAP product. NetApp has not supplied a comparative customer benchmark showing how the proposed combination performs.

The distance between design and delivery

NetApp describes an architecture intended to support trillions of files and deployments spanning multiple exabytes of data. That is a statement of product direction, not a measured result for the combined system. Its announcement explicitly says the company makes no commitment to develop or deliver the integrations or features it describes. The timing and final form of any customer-facing product remain at NetApp’s discretion.

The acquisition itself is not complete. NetApp disclosed neither a purchase price nor an expected closing date, and the transaction remains subject to customary closing conditions and regulatory approvals. The next concrete milestones are therefore different from the architecture’s ambitions: closing the deal, defining what NetApp will deliver, and showing how the integrated system performs for customers.