Three executive takeaways from theCUBE's NetApp INSIGHT coverage
AI is moving from dashboards and batch jobs to agents that act on your behalf. That shift makes data strategy a board-level priority. theCUBE's analysts and NetApp leaders made one point clear: winning with AI is a data problem first, an infrastructure problem second.
Below are three insights that matter for executives who own growth, risk and operating leverage.
Insight 1: AI data strategy is uniting infrastructure and innovation
Agentic AI isn't hypothetical. NetApp CEO George Kurian described agents serving as a digital twin for every employee - executing tasks, making decisions, and creating new data as they work. To make that safe and useful, two conditions are non-negotiable: curated, high-quality data and context-aware guardrails.
Kurian also pushed a practical principle: act where data is created. Moving less data means lower latency, lower cost and fewer failure points. In other words, bring compute and controls to the data, not the other way around.
- Make "data quality and meaning" a requirement, not an afterthought. Define gold datasets, lineage and context.
- Codify guardrails from domain knowledge: entity relationships, permissions, and allowed actions for agents.
- Shift from copy-heavy pipelines to in-place processing and policy enforcement at the data source.
Insight 2: AI returns come from unified, intelligent data
The conversation at INSIGHT moved beyond bigger models. Nvidia's Tony Paikeday underscored that models only deliver when they're fed enterprise context - customer nuances, IP, vocabulary, and culture. NetApp's new AI Data Engine, built on Nvidia's reference design, points to a clear playbook: unify your data, simplify access, and scale efficiently.
NetApp leaders Russell Fishman and Syam Nair emphasized integrated tooling so teams focus on outcomes, not plumbing. That same thinking shows up in the NFL's approach with NetApp: prepare the data first, or AI projects stall.
- Tie AI ROI to business workflows, not model size. Measure cycle time saved, conversion lift, and error reduction.
- Unify data across on-prem and cloud with one namespace, one set of policies, and built-in observability.
- Prioritize data readiness: classification, deduplication, governance, and feature access patterns for GPUs.
Learn more about NVIDIA AI Enterprise for enterprise-scale AI stacks.
Insight 3: Secure foundations and data hygiene come first - everywhere
The message from global leaders was consistent. In APAC, Andrew Sotiropoulos highlighted government-backed ecosystems balancing data sovereignty with AI adoption. In EMEA and LATAM, Giovanna Sangiorgi stressed turning policy into practice through a clear data strategy, or systems won't hold up over time.
Sandeep Singh put a fine point on it: AI runs on data, and that data must be secure across multi-tenant, hybrid environments. GPUs need seamless, governed access to unified data - with zero trust principles baked in.
- Adopt a shared control model: identity, encryption, isolation, and auditability across on-prem and cloud.
- Standardize data hygiene: retention, PII handling, model input/output logging, and drift monitoring.
- Operationalize risk: map AI use cases to policies and controls using recognized frameworks.
Reference the NIST AI Risk Management Framework to align policy with day-to-day operations.
A simple plan for the next 90 days
- Week 1-2: Name your top three agent use cases per function. Define the decision rights and guardrails for each.
- Week 3-6: Stand up a unified data layer with clear ownership, lineage, and a gold dataset per use case.
- Week 7-10: Integrate policy and access controls end-to-end. Validate with red-team prompts and audit trails.
- Week 11-13: Pilot with a small group. Track business metrics, not model metrics. Iterate or kill fast.
Executive checklist
- Do we have curated, contextual datasets that agents can trust?
- Are guardrails defined from data meaning and user intent, not just system rules?
- Can GPUs reach the right data securely across on-prem and cloud, with full auditability?
- Is ROI tied to workflow outcomes and cost-to-serve, not parameter counts?
- Are we acting where data is created to reduce copies, latency, and risk?
Upskill your leadership team
If you're aligning AI initiatives to business outcomes and need targeted upskilling by role, explore curated programs at Complete AI Training.
(* Disclosure: TheCUBE is a paid media partner for the NetApp INSIGHT 2025 event. Neither NetApp, the sponsor of theCUBE's event coverage, nor other sponsors have editorial control over content on theCUBE.)
Your membership also unlocks:
