NetApp AFX and the AI Data Engine explained: what they are and when they fit
NetApp’s answer to enterprise AI data is AFX, a disaggregated version of ONTAP, and the AI Data Engine that runs on it, announced in October 2025. Alongside it sits AIPod Mini with Intel for smaller inference projects. Here is what each one is, who it is for, and what to check before you buy.
NetApp AFX separates performance from capacity within ONTAP, so controllers and NVMe flash enclosures scale independently, up to 128 controllers in a cluster, for high performance file and object workloads such as AI. The AI Data Engine runs on optional DX50 compute nodes inside the AFX cluster and prepares data for AI: it keeps a live view of the data, classifies sensitive content, and handles vectorisation and search. AIPod Mini with Intel is a smaller package for departmental inference. AFX suits organisations that want AI data on ONTAP with its existing data services; it is not the default for a simple midrange refresh.
C4C Group is an independent UK technology consultancy and an official partner of Dell Technologies, NetApp, HPE and VAST Data, tied to none of them. Nick Watson and Paul Smith spent over a decade at EMC selling enterprise arrays to finance, banking and telco customers, so we read a launch from the buyer’s side. We supply, implement and support what we recommend, with hardware break fix support from the vendor, and we will tell you when a different platform fits better. Talk to us about it.
What is NetApp AFX?
AFX is NetApp’s disaggregated ONTAP platform, announced at NetApp INSIGHT in October 2025. Instead of controllers and disks scaling together, AFX controllers and NVMe storage enclosures scale independently, which Blocks and Files describes as up to 128 AFX storage controllers per cluster, sharing a single pool of storage over Cisco Nexus 400G switches, with pNFS, NFS, SMB, S3 and NFS over RDMA. Source: Blocks and Files, 14 October 2025.
NetApp’s documentation lists the AFX 1K controller, NX224 NVMe shelves and the Cisco switches as the AFX components from ONTAP 9.17.1, and its supported platforms list adds the AFX 2K from ONTAP 9.19.1. AFX keeps ONTAP’s data services and works with existing AFF and FAS systems through SnapMirror, SVM migrate and FlexCache. Sources: NetApp AFX release notes, NetApp supported platforms, Blocks and Files, 14 October 2025.
What is the NetApp AI Data Engine?
The AI Data Engine is a set of data services for AI that runs inside the AFX cluster on optional DX50 data compute nodes, so it does not take resources from primary workloads. Blocks and Files describes four parts: a metadata engine that keeps a current, queryable view of the data estate; data sync to keep AI copies current; data guardrails that classify data and find sensitive content to redact or exclude; and a data curator for discovery, search, vectorisation and retrieval, with a built in vector database aligned to the NVIDIA AI Data Platform reference design. AFX is also certified with NVIDIA DGX SuperPOD. Sources: Blocks and Files, 14 October 2025, Techzine.
Where does AIPod Mini fit?
AIPod Mini with Intel is a separate, smaller offer for running AI inference at department or team level, pairing Intel Xeon 6 processors with NetApp all flash storage and Kubernetes integration. NetApp made it available in July 2025. It suits a team that wants to run inference on its own data without building a large AI platform. Source: NetApp, July 2025.
Who is AFX for?
AFX fits organisations that already trust ONTAP, want their AI data to keep the same snapshots, replication, security and governance, and need performance and capacity to grow separately as AI workloads expand. The AI Data Engine is aimed squarely at the hardest part of enterprise AI: getting current, governed, searchable data to models and agents without making uncontrolled copies.
When might it not be the right answer?
- A straightforward midrange refresh. If you are replacing an ageing AFF or FAS system for general workloads, current AFF or ASA systems are the simpler path. See NetApp end of support and tech refresh.
- Very large AI training estates on other platforms. Compare AFX with other AI data platforms on your workload. See storage for AI: VAST vs Pure vs NetApp vs Dell.
- Capacity claims. The headline effective capacity depends on data reduction, and some commentators question how far that applies to AI data. Size on usable capacity for your data. Source: StorageMath.
- New architecture. AFX is a newer platform than classic ONTAP clusters, so check the features and integrations you need are available on it today, not on the roadmap.
What should you check before you buy?
- Which AI workloads you are actually running: training, retrieval, inference or data preparation.
- Whether you need the AI Data Engine, which requires the optional DX50 nodes.
- How AFX connects to your existing ONTAP estate and data protection.
- Usable capacity and growth, not the effective headline.
- The commercial model: purchase or a consumption programme such as Keystone. See storage as a service compared.
Planning AI data on NetApp?
Tell us what your AI workloads are and what you run today. We will tell you honestly whether AFX, a current AFF system, AIPod Mini or another platform fits best.
Prefer email? Reach us directly at hello@c4cgroup.co.uk.
Frequently asked questions
What is NetApp AFX?
NetApp AFX is a disaggregated version of ONTAP announced in October 2025, in which controllers and NVMe storage enclosures scale independently, up to 128 controllers per cluster. It serves high performance file and object workloads such as AI over NFS, pNFS, SMB, S3 and NFS over RDMA, while keeping ONTAP data services.
What is the NetApp AI Data Engine?
The AI Data Engine is a set of AI data services that run inside an AFX cluster on optional DX50 compute nodes. It keeps a current view of the data estate, syncs AI copies, classifies sensitive content and handles discovery, vectorisation and retrieval, so models and agents get current, governed data.
What is NetApp AIPod Mini?
AIPod Mini with Intel is a smaller NetApp and Intel package for departmental AI inference, pairing Intel Xeon 6 processors with NetApp all flash storage and Kubernetes integration. NetApp made it available in July 2025.
Does AFX work with my existing NetApp systems?
AFX runs ONTAP and works with existing AFF and FAS systems through ONTAP features such as SnapMirror, SVM migrate and FlexCache. Check the specific integration and protection workflows you need with NetApp or a partner before you design around it.
Which ONTAP version does AFX need?
NetApp’s documentation lists AFX 1K systems from ONTAP 9.17.1, and its supported platforms list shows AFX 2K from ONTAP 9.19.1.
Is AFX the right choice for a normal NetApp refresh?
Usually not. AFX is designed for high performance, independently scaling workloads such as AI. For a general midrange refresh, current AFF or ASA systems are typically simpler and more cost effective.
Is C4C a NetApp partner?
Yes. C4C is an official NetApp partner and also partners with Dell, HPE and VAST Data, tied to none. We supply, implement and support NetApp, with hardware break fix from NetApp, and we will tell you when another platform fits better.