To understand 2026, let’s start with what happened in 2025. AWS, Azure, and Google Cloud didn’t just grow; they grew at scale. Combined, they crossed well north of $300B in annual cloud revenue, still expanding around 20–30% year over year. At that size, growth doesn’t come from basic migrations anymore. It comes from new kinds of workloads showing up in volume.

That workload is AI.

And AI has changed the rules of cloud in a very specific way: it made data the center of gravity. 

AI Didn’t Just Add Demand, It Changed Where Cloud Breaks

AI workloads differ from traditional cloud workloads. They are persistent, data-intensive, and demanding. Training pipelines move massive datasets. Inference runs continuously. Latency shows up in user experience. Downtime shows up as lost trust.

By 2027, more than half of new enterprise applications will include embedded AI inference, not as experiments, but as everyday product features. When that happens, the data layer stops being just an “infrastructure component” and starts becoming a strategic part of the product itself.

That’s why our cloud strategy at NetApp has been very intentional. We didn’t try to build a separate cloud. We went first-party with hyperscalers.

Azure NetApp Files, Amazon FSx for NetApp ONTAP, and Google Cloud NetApp Volumes exist because customers don’t want to assemble a data stack anymore. They want enterprise-grade storage that is native to the cloud they’re already using, same console, same APIs, same billing, same operational model, but with the performance, protection, and data management behavior they trust.

The Fastest-Growing Cloud Spend is Data-Driven, Not Compute-Only

One thing that became obvious in 2025 is that AI doesn’t just burn compute. It amplifies data gravity. Storage, snapshots, replication, recovery, all of it suddenly matters more, and it matters all the time.

Across hyperscalers, data services stopped being “attach rates.” They became core revenue drivers because AI fails fast when data isn’t fast, consistent, and reliable. This is exactly where first-party NetApp services show up. Customers running mission-critical workloads on Azure or AWS choose Azure NetApp Files or FSx for NetApp ONTAP because they deliver predictable throughput, low latency, and built-in data protection, without forcing teams to redesign their applications.

The focus isn’t on flashy features but on consistent data behavior at scale.

Multi-Cloud Isn’t a Strategy Anymore, It’s Just What Happened

Most large enterprises didn’t wake up one day and decided to be multi-cloud. They became multi-cloud because different teams made sensible decisions at different times.

AI accelerates this even further. Teams pick the cloud that fits the model, the service, or the geography. The result is fragmentation, unless data behaves consistently across environments.

The key question has shifted from “Which cloud should I pick?” to “How can my data behave consistently across all clouds?”

NetApp’s first-party cloud services are designed around that reality. Same ONTAP data semantics. Same snapshot behavior. Same replication and recovery patterns whether the workload runs on Azure, AWS, or Google Cloud. That consistency is what lets AI move from pilot to production without introducing new risk.

Open Source Runs AI, but Production Has No Patience for Fragility

If you look at any modern AI pipeline, you’ll find open source everywhere underneath it. That hasn’t changed.

What has changed is tolerance.

Once AI systems become customer-facing, outages aren’t just learning moments. They’re incidents. Enterprises still want open platforms, but they don’t want to operate them across regions, compliance regimes, and on-call rotations.

That’s where Instaclustr fits into the picture. It brings managed open-source data services into the same cloud reality, open foundations, delivered as reliable, production-grade services. It complements our first-party storage by addressing the broader data stack that feeds AI.

Open source didn’t slow down. It just stopped being experimental.

Cloud Platforms Split, and Half-Built Strategies Start to Show

By 2026, the cloud market draws a clear line. Customers don’t talk about it explicitly, but you see it in how they buy and deploy. They separate core from everything else.

Core workloads are where AI lives in the critical path always-on data, inference that affects users in real time, systems where latency, durability, and recovery are non-negotiable. For these workloads, customers don’t want abstraction. They want services that are native to the hyperscaler itself, same control plane, same APIs, same operational guarantees.

That’s why NetApp first-party cloud storage matters more now than ever. They are part of the hyperscaler fabric. That’s where customers anchor their data plane when failure isn’t an option.

Around that core, everything else is moving faster, and this is where marketplaces shine.

Cloud marketplaces are growing at 30%+ year over year because they remove friction. New teams spin up. New use cases emerge. Customers click, deploy, and move on, billing everything against existing cloud commitments. Marketplaces are absolutely the right vehicle for speed.

Here’s the problem we see with many competitors: they only exist in the marketplace layer. That works for expansion and experimentation. It breaks down when customers try to run their most critical workload and realize there’s no native anchor underneath.

This isn’t a knock on marketplaces. We’re investing heavily there too. But marketplace alone isn’t a complete cloud strategy in 2026.

What works now is both. First-party services anchor the core — where performance, resilience, and data semantics matter most. Marketplace offerings accelerate everything around that core — adoption, experimentation, and scale. One without the other leaves gaps customers eventually hit.

NetApp is deliberately built for this split. First-party where it counts. Marketplace where it moves fast.

That’s not hedging. That’s what cloud usage in 2026 actually looks like.

Who Actually Decides Has Changed, and Keeps Changing

Most new workloads don’t start with executive mandates anymore. They start with developers, platform teams, or increasingly automated and AI systems spinning things up on their own.

That has changed what “good” looks like in cloud services. Predictability beats cleverness. Automation beats customization. Boring, in the right ways, beats impressive.

First-party services matter here because they inherit the cloud’s operational model by default. They’re scriptable, automatable, and designed to be consumed by machines as much as people.

Where This Leaves NetApp Cloud in 2026

Net net, AI is driving cloud growth forward faster than migrations ever did. Hyperscalers are compounding at massive scale. Data, not compute, is the fuel driving cloud growth and transformation. Open source is foundational but must be managed. Complexity is being punished.

NetApp’s cloud strategy lines up directly with that reality: First-party cloud storage services embedded in hyperscalers, consistent data behavior across clouds, managed open-source platforms where it matters, and frictionless consumption through marketplaces.

This isn’t just marketing — it reflects the trends and customer demands of 2025.

And 2026 is when it becomes the default.