In this episode of the AI Intelligence Lab, NetApp’s Tore Sundelin, Senior Director of AI Product Management, and NVIDIA’s Jacob Liberman, Director of Enterprise Product, dive into how the AI Data Engine turns massive, messy enterprise data estates into something you can actually build apps and agents on. The platform plugs data compute nodes directly into existing NetApp storage clusters, creating a unified, secure path from raw data all the way to AI applications. Instead of juggling a dozen tools and copying data six or seven times, teams get one integrated stack that handles discovery, indexing, search, and RAG-style access to live data.
Making life easier for storage and security teams
A big theme is “don’t make storage admins become data scientists overnight.” Storage teams use the same ONTAP System Manager interface they already know to deploy GPU-powered data compute nodes, manage them like any other controller, and point AI Data Engine at the right volumes and buckets. As it indexes everything in the background, the system builds a global, searchable view of the data estate spanning on-prem and cloud, which Jacob calls the critical first step in AI adoption. On the security side, NetApp adds more than 185 classifiers and policy controls at the storage layer so security officers can decide what’s in or out, who can see what, and how sensitive fields are masked, all while shrinking the attack surface for GenAI workloads.
Giving data scientists and engineers a real runway
Once the foundation is in place, data scientists and data engineers move into Data Curator, a workspace built into the NetApp console. There they can explore governed datasets, build collections, and have them transformed into AI-ready embeddings in place—no lifting and shifting data out of sovereign regions or secure zones. This is where NVIDIA’s tech shows up most clearly: models like NeMo Retriever embed unstructured data as vectors for fast, large-scale semantic search, all tuned to run efficiently on NVIDIA GPUs via NIM microservices. A standout feature Jacob highlights is dynamic collections, which automatically stay in sync as files or permissions change so AI agents don’t drift away from the source of truth.
A “better together” platform
Throughout the conversation, both sides stress how complementary the partnership is: NVIDIA brings accelerated computing and cutting-edge AI models; NetApp brings battle-tested data management, governance, and global data fabric. NVIDIA is not just a partner but also an early internal customer, running AI Data Engine at scale and appreciating how deeply their own software is embedded into ONTAP, to the point where it “almost disappears” behind the familiar storage experience. Tore and Jacob close by looking ahead to more industries, more pre-release customers, and a rapid cadence of new capabilities, all aimed at letting enterprises tap into their data with trusted, production-ready AI—not just more pilots and proofs of concept.
Explore More of the AI Intelligence Lab
In Episode 4, the conversation continues about how enterprises invest heavily in GPUs but data bottlenecks keep them sitting idle and how AFX is purpose-built to solve the data throughput challenge that prevents AI workloads from scaling to production.
