AI data readiness remains a hard enterprise problem because usable AI depends on more than fast storage: organizations must find, prepare, govern and activate data across fragmented systems. At NetApp Insight 2026, the company returned to that challenge with new infrastructure and data-service announcements. ITPro’s Ross Kelly argues that NetApp has reason to keep talking about readiness, even though its Novus architecture targets the high end of the market rather than the everyday data-preparation work most businesses face.
Why are enterprises still talking about AI data readiness?
Because the gap between an AI pilot and a dependable business system is not closed by a model or a storage upgrade alone. Organizations still have to identify which data is usable, address quality and consistency, establish ownership and controls, and make information available to the systems that need it. Kelly’s 1 October 2026 analysis describes data readiness as a recurring enterprise issue and a central message at NetApp Insight 2026.
| # | Preview | Product | Price | |
|---|---|---|---|---|
| 1 |
|
NETAPP 74672-10 NTEAPP Network Appliance 3.5" Drive Caddy | $97.00 | Buy on Amazon |
NetApp’s chief product officer Syam Nair frames the problem around four areas: scale, activation, control and return on investment. In the company’s account, enterprise data is scattered across on-premises systems, public clouds and edge locations, while custom pipelines and manually applied controls make preparation and governance costly. These are NetApp’s characterization of the challenges, not an independent assessment of every organization.
- Scale: Data estates span more locations and systems than a single AI pipeline can conveniently handle.
- Activation: Data must be discovered and made useful to models or agents, not merely stored.
- Control: Access, privacy, compliance and governance need to apply consistently as data is used.
- ROI: Technical work has to translate into business value, which requires more than a successful infrastructure deployment.
Nair’s vendor definition is that data readiness means data is reachable in place, governed and secure, and fast enough to be useful when a model or agent requests it. That definition captures important infrastructure needs, but readiness also involves data quality, classification, curation, ownership and organizational processes.
Free tools Windows power users keep installed
One-click scans. No signup required.
#1 Best Overall
What NetApp announced at Insight 2026
Novus: infrastructure for AI factories
NetApp presented Novus as a zettabyte-scale file system for large AI factories and GPU-cluster environments, powered by ONTAP. Kelly reports that NetApp claimed throughput of up to 100 Tbps. NetApp’s own 29 September 2026 post describes the system as designed for 100 TB/s. Those units differ, and the available accounts do not reconcile them; neither figure should be treated as an independently verified benchmark. NetApp also says actual features, functionality and timing may differ from its announcement.
Kelly’s analysis places Novus at the upper end of the market. A large, high-performance AI factory may have a specific need for that scale, but the architecture does not by itself solve more routine problems such as inconsistent data, unclear permissions or the work of preparing information for a useful AI application.
NetApp executive Arindam Banerjee, as reported by Kelly, said a stalled cluster of 100,000 GPUs could cost “tens of millions of dollars every day.” This is an executive’s estimate about a hypothetical large cluster, not a validated cost model that applies to all AI deployments.
AI Data Services and related platform announcements
NetApp says its AI Data Services can discover, understand, govern and operationalize data in place, including data on ONTAP, StorageGRID and non-NetApp storage. The company describes this approach as secure and zero-copy. The announcement also covers Console autonomous operations within customer-defined guardrails, Fleet Management, Keystone Sovereign and AI ChatOps. These are vendor descriptions of announced capabilities, not independent evaluations; NetApp notes that features and timing may change.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
In an October 2025 post, NetApp described AFX 1K as a disaggregated AI storage system and AIDE as a data-lifecycle service with metadata indexing, automated curation, privacy and compliance guardrails, and vectorization. NetApp also said AIDE included NVIDIA AI Enterprise licensing and NIM microservices, and discussed Keystone consumption, FlexPod AI with Cisco, and integrations involving NVIDIA, Domino Data Lab, Starburst, Microsoft and LangChain. Those descriptions show how the company has positioned its AI portfolio over time; they do not establish comparative performance or suitability for a particular deployment.
What NetApp’s approach does—and does not—answer
In-place discovery and governance could matter to organizations trying to reduce bespoke data movement and pipeline work, especially when information is spread across more than one storage environment. But a capability claim is not proof that a given estate can be integrated without substantial implementation effort, or that its data is accurate, well classified or properly owned.
That distinction is why Kelly’s broader point remains relevant: AI adoption is not simply a matter of buying tools. He reports NetApp CEO George Kurian describing it as “a business and leadership transformation program.” Leadership, teams and processes have to determine what data may be used, who is accountable for it, and how an AI outcome will be judged. Storage can support that work; it cannot substitute for it.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.How to evaluate an AI data-readiness solution
Compare solutions against the work your organization actually needs to do, rather than treating peak throughput or a long feature list as a proxy for readiness.
| Evaluation area | Questions to ask |
|---|---|
| Data readiness work | Can the solution help discover, assess quality, classify, add metadata and curate data? Who owns each step? |
| Governance and risk | How are permissions, privacy, compliance, sovereignty, protection and auditability handled? |
| Placement and movement | Can data be accessed in place, or must it be copied into pipelines? Which on-premises, cloud and edge environments are supported? |
| Performance and scale | What workload, concurrency, throughput and latency requirements matter? Is the architecture designed for ordinary enterprise work or a GPU-cluster-scale AI factory? |
| Operations and business fit | What implementation effort and staff skills are required? How will costs and return on investment be measured, and what process changes are necessary? |
The announcements and reporting do not provide a neutral benchmark of NetApp against competitors or comparable pricing and deployment data. A buying decision therefore requires deployment-specific evidence: fit with the existing estate, validated performance for the intended workload, governance requirements, implementation effort and a credible way to measure business value.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
