NetApp has agreed to acquire PEAK:AIO, a Manchester software company whose parallel file-system technology is designed to keep large GPU clusters supplied with data. The September 25 announcement adds a specialist metadata layer to NetApp’s push into storage for artificial intelligence and high-performance computing.

 

The purchase price and expected closing date were not disclosed. NetApp said the transaction remains subject to customary closing conditions and regulatory approvals, making this a planned acquisition rather than a completed integration.

 

Further Reading

 

How NetApp’s PEAK:AIO Acquisition Targets Metadata Bottlenecks

AI infrastructure is often discussed in terms of accelerators, but storage can become the constraint when thousands of processors request training files, model checkpoints and intermediate results at once. A GPU that waits for data is expensive capacity sitting idle.

 

PEAK:AIO focuses on the metadata operations that tell systems where files are, who may access them and how data is distributed. Its architecture separates metadata services from stored data so each layer can scale independently as workloads grow.

 

NetApp says the combination is intended to provide four capabilities:

  • Independent scaling for metadata services.
  • A global namespace across large storage environments.
  • Parallel NFS access for concurrent workloads.
  • Support for trillions of files and multi-exabyte deployments.

 

Those capacity targets are architectural goals, not independently verified results for a finished NetApp product. The company’s announcement also states that future integrations, features and delivery dates remain subject to change.

 

PEAK:AIO Brings Parallel NFS and the Lattice Project

PEAK:AIO’s key contribution is parallel NFS, or pNFS, which lets clients reach multiple data servers through a standards-based file-access protocol. This can distribute traffic instead of forcing every request through one metadata endpoint.

 

StorageReview identified pNFS Lattice as an important part of the deal. The open-source project separates metadata authority, protocol coordination and data placement, using multiple metadata-server processes and standard NFS referrals to direct clients.

 

Lattice was developed with Los Alamos National Laboratory and Carnegie Mellon University and published with Linux Foundation support. It is still described as an early release for research, development and testing, so production-readiness should not be inferred from the acquisition announcement.

 

PEAK:AIO also has deployments at Los Alamos, NHS AIDE, Oxford Robotics Institute, Carnegie Mellon, the University of Liverpool, the University of Strathclyde MediForge Hub and the Zoological Society of London, according to NetApp.

 

NetApp Plans to Connect PEAK:AIO With ONTAP

NetApp plans to combine PEAK:AIO’s metadata and parallel-namespace technology with ONTAP, its core storage operating system. The intended design keeps ONTAP’s data management, resilience and security functions while allowing metadata services to expand separately.

 

That approach also gives existing ONTAP customers a potential migration path toward larger AI environments without replacing every established data-management process. CRN reported that PEAK:AIO’s parallel NFS client fills a capability that NetApp did not previously have in its portfolio.

 

The acquisition follows NetApp’s July purchase of DataPelago, which brought GPU-accelerated data processing closer to the storage layer. StorageReview described PEAK:AIO as NetApp’s second AI-focused acquisition in roughly two months, indicating a strategy built around both data movement and metadata scale.

 

The Deal Extends Competition Beyond AI Chips

The transaction reflects a broader shift in AI infrastructure spending. As accelerator clusters grow, buyers must also invest in networking, storage software and orchestration that can keep those processors occupied across training, inference and agentic workloads.

 

For operators, that shifts planning from raw capacity toward metadata throughput, file concurrency and standards-based access. These less visible systems can determine whether costly AI compute remains productive during sustained, data-heavy jobs.

 

NetApp is betting that established enterprise controls can be paired with architecture developed for research laboratories and high-performance computing. PEAK:AIO gains access to NetApp’s sales reach and installed customer base, while NetApp acquires technology aimed at a problem becoming more visible as GPU deployments scale.

 

Several questions remain open. NetApp has not disclosed the purchase price, a closing timetable or which product will receive PEAK:AIO technology first. It has also not committed publicly to the future governance of the open-source Lattice project.

 

The next concrete milestones will be regulatory clearance, transaction close and a product roadmap. Until then, the strategic direction is clearer than the delivery schedule: NetApp wants metadata and parallel file access to become core parts of its AI storage platform.