NetApp, the intelligent data infrastructure company, unveiled an advanced generative AI data vision and end-to-end integrated solutions that combine NVIDIA AI software and accelerated computing with NetApp intelligent data infrastructure for enterprise retrieval augmented generation (RAG) to power the future of agentic AI applications.
This new offering will bring enhanced capabilities to the NetApp ONTAP unified storage operating system. It leverages a new NetApp global metadata namespace to unify data stores for the enterprises that rely on NetApp for their data infrastructure. This development opens up exabytes of enterprise data stored across cloud and on-premises infrastructure, driving RAG capabilities to accelerate next-generation agentic AI applications.
The solution integrates the NetApp AIPod architecture with NetApp ONTAP and the NetApp BlueXP unified control plane, complemented by NVIDIA NeMo Retriever and NIM microservices, part of the NVIDIA AI Enterprise software platform. This combination aims to enable customers to discover, search, and curate data on-premises and in the public cloud, adhering to policy-based governance criteria.
Harv Bhela, Chief Product Officer at NetApp, commented, "To power AI applications and drive transformative progress for their business, enterprises must unlock the potential of their data. Combining the NetApp data management engine and NVIDIA AI software empowers AI applications to securely access and leverage vast amounts of data, paving the way for intelligent, agentic AI that tackles complex business challenges and fuels innovation."
Manuvir Das, Vice President, Enterprise Computing at NVIDIA, added, "Data is fundamental to the evolution of generative AI. By combining NVIDIA AI software and accelerated computing with NetApp intelligent data infrastructure, enterprises can turn their data into knowledge, and AI agents can turn that knowledge into action."
With the new NetApp AI capabilities integrated into NetApp AIPod, certified for NVIDIA DGX BasePOD infrastructure and NVIDIA OVX solutions, and managed through BlueXP, NetApp customers will be able to easily discover, search, and curate data across on-prem and public cloud environments. This is expected to honour existing policy-based governance criteria.
Once data collection is established through NetApp BlueXP, it can be dynamically connected to NVIDIA NeMo Retriever for dataset processing and vectorization. This process ensures the data is accessible for enterprise GenAI deployments with appropriate access controls and privacy guardrails. The aim is to create a foundation for a generative AI flywheel that powers agentic AI applications capable of autonomously and securely accessing data to complete various tasks supporting customer service, business operations, financial services, and more.
The end-to-end integration prioritises security and policy guardrails throughout the AI data and model lifecycle. Initially presented as a proof-of-concept by Huang in his NVIDIA GTC 2024 keynote address, this secure and compliant GenAI integration will be showcased at NetApp INSIGHT and is slated for a technology preview release later this year.
Furthermore, NetApp has commenced the NVIDIA certification process of its ONTAP storage on the AFF A90 platform with NVIDIA DGX SuperPOD to enable organisations to leverage leading data management capabilities for extensive AI projects. This certification will complement and build upon NetApp ONTAP's existing certification with NVIDIA DGX BasePOD. NetApp ONTAP aims to address data management challenges for large language models, thus eliminating the need to compromise data management for AI training workloads.
