NetApp CEO Outlines Data Strategy for Production AI With Novus Architecture

Kurian said 93% of organizations pursuing production AI identify data as their leading challenge. Addressing the associated safety risks will require action from model builders, security vendors and enterprise operators, he said.

"When you want to make AI scale across your enterprise, you need to integrate it into the operational systems and the workflows and how work gets done," Kurian said, according to SiliconANGLE. "That creates ... new requirements on the infrastructure, like governance, security, control and so on."

NetApp introduced Novus this week, the report said. The architecture combines NetApp's established data layer with an independently scalable metadata layer. It is designed to deliver more than 100 terabytes per second and support zettabyte-scale capacity within one namespace.

Kurian said the combination is intended to meet the performance requirements of demanding enterprises. The metadata service becomes a place for data discovery, data cataloging, and data governance and control, he said. A highly scalable metadata service, he added, allows organizations to implement the guardrails needed for production AI.

According to Kurian, organizations can start small and expand to large AI factories on a consistent platform. He described intelligent data infrastructure as essential to creating trusted knowledge.

"I think that the AI era requires organizations to be really good stewards of their data, because data is the foundation for knowledge and knowledge is the intellectual property," Kurian said. "To have knowledge, you need really good intelligent data infrastructure that not only allows you to extract memory and context from your data, but also to keep it secure, well governed and a resource that you can trust."

The discussion covered production AI, data governance and NetApp's infrastructure strategy, according to SiliconANGLE. TheCUBE is a paid media partner for NetApp INSIGHT, and neither NetApp nor other sponsors have editorial control over content on theCUBE or SiliconANGLE, the report said.