NetApp AFX’s disaggregated storage architecture addresses customers data needs for AI

Unveiled at NetApp Insight 2025, NetApp AF delivers proven, enterprise-grade data management and security features of NetApp ONTAP while integrating seamlessly into organizations’ hybrid multi-cloud data estate.

NetApp AFX, which was unveiled at NetApp Insight 2025, delivers proven, enterprise-grade data management and security features of NetApp ONTAP while integrating seamlessly into organizations’ hybrid multi-cloud data estate. This allows AI in that organizations are using to only access the data that is intended, addressing the concerns companies have when it comes to using data for AI.

It is the industry’s first disaggregated storage architecture purpose-built for enterprise AI. Given that legacy storage architectures can’t scale to meet the high-performance demands of modern AI workflows, organizations continue to struggle with fragmented point solutions. These not only introduce silos but also lack enterprise-grade security and resilience as well as increase operational complexity. Data mobility challenges also disrupt AI pipelines, making it harder to move, manage, and curate data across environments.

According to Sandeep Singh, SVP, GM Enterprise Storage at NetApp, customers are continuing to modernize their underlying infrastructure because they're buried in silos, and they really need to be able to unify their data across on-prem and cloud and across workloads.

“Where we have built the NetApp data platform is to be able to help them be able to have a data platform that is tied to any app, any data, any cloud. What that fundamentally means for customers is that they can seamlessly deploy their application workloads and have it optimized and accelerated. They're able to seamlessly have their data stitched together across on-prem and cloud. They have the world's best security and cyber resiliency that is just built into it, and they can truly unleash the power of AI with AI-ready data. That's fundamentally the journey that we have been on,” said Singh.

NetApp AFX

For Singh, NetApp AFX gives customers the levels of performance that they need to be able to keep their GPUs fed wherever the GPUs live.

“It is a disaggregated architecture so that customers can scale the performance and capacity independently. It gives them linear scaling as well as efficient scaling. It can scale into terabytes per second of bandwidth. That is critically important not only for inferencing use cases but also for the training and checkpointing use cases there. And it basically can deliver all the way into 128 node scalabilities for customers. Fundamentally, AFX is built on NetApp ONTAP,” Singh said.

Singh also highlighted that this means that AFX will be a plug and play. Early access customers were able to install AFX and all of the automation with ONTAP that they had in place and were able to just go and manage AFX with it.

“Because it's ONTAP, it brings enterprise-proven reliability, availability, the multi-tenancy. When you think about the centers of excellence that customers are creating AI, ultimately it requires multiple data science, data engineering teams and eventually agents to have access to the right data sets. And this is where you need that secure multi-tenancy and QoS built in. It also brings along the secure and efficient data mobility while preserving all of the security permissions. So, when you think about the use cases where customers' data might be in one data center and their GPU cloud might be in a different data center or it might be in the cloud,” Singh explained.

As customers need to be able to have a zero-copy access to that data while preserving all of the security permissions, Singh added that with ONTAP, it is just built in and available to them.

At the same time, Singh also stated that AFX is cloud-connected. As ONTAP is the backbone that unifies on-prem and cloud, Singh said customers are able to seamlessly have their data accessible to the AI tools and frameworks in the public cloud.

The partner ecosystem

As NetApp AFX is on ONTAP, Singh explained that channel partners are going to be a big part of going and taking this to market across the board.

“All of our channel partners natively understand ONTAP technology and because it's plug and play, it's going to be super simple for customers. The second part, we have invested over the last year in what is called AIPod. These are essentially reference architectures that are validated to bring overall compute, networking, and storage alongside the AI software stack together,” Singh said.

For the channels specifically, Singh mentioned that NetApp has also designed AIPod Mini to invest, working with the channel partners in all of the proper enablement as well as helping them to go and succeed.

“There's a ton of excitement with the channel around AIPod Mini and this will continue. Secondly, Flexpod continues to be a fantastic overall converged infrastructure solution that's jointly between NetApp and Cisco. We are adding AFX over to Flexpod as well. So that's another part that's just going to tremendously help our channel partners because they are the key in bringing FlexPod to customers,” he concluded.