Pure looks beyond storage to data management

AI is having a significant effect on enterprise storage expectations and strategy, which includes a deepening relationship between how data is stored and how it is managed.

Today, many storage vendors talk about governance, permissions and data access almost as easily as they talk about IOPS, bandwidth and throughput. And some are striving to build closer relationships between data management in storage and data management elsewhere in the enterprise.

At this week's Pure//Accelerate 2025, Pure is highlighting a raft of new products to customers, including its Enterprise Data Cloud, an architecture that turns disparate data silos into a single fleet and pushes storage deeper into services territory.

Here, John "Coz" Colgrove, Pure's founder, talks with Informa TechTarget about the convergence of data management in the enterprise, the benefit of storage copilots and the future of the storage admin.

Editor's note: This interview has been edited for length and clarity.

Let's start with a future-looking question: Is storage's final state invisibility?

John 'Coz' Colgrove

John Colgrove: The problem is there are tradeoffs around energy and access time, where more energy generally means faster access. If I take data and put it on archival storage it uses less energy, but it might take hours to get it back. … I don't see how storage gets to be invisible, because I think the only way it would truly be invisible is if we broke that energy equals performance and access flexibility tradeoff. And I don't know that that is possible with any storage technology -- or ever will be.

The enterprise talks about data management in terms of storage. But there's another part of the enterprise that talks about data management often in relation to analytics. These two pieces seem to be moving closer together -- converging. Is this the direction Pure is headed?

Colgrove: I think that is a direction that we're trying to go -- and others are trying to go there. There's this term that people use for a dumb group of discs called JBOD: 'just a bunch of discs.' It doesn't do anything, and that kind of storage was much more popular years ago. But now you look at how cheap a bunch of the computing has gotten, and how much denser and more efficient the storage systems have gotten. It's natural to try and bring more of the whole end-to-end problem together.

Because it is really about the data. Think about everything you store -- every file, every photo, every medical record, financial record. There are things about how that data is managed that make sense to you, and that's the way you think about managing your own personal files. And yes, everybody should be doing that with all their enterprise data.

But as you get rules and regulations designed to protect people -- the European right to be forgotten or personal information privacy and healthcare privacy -- and you combine it with things like [retention policies] … people do need to do more.

So, yes, storage has been moving in this direction, and the two are coming together and becoming more intertwined.

What's the benefit of seamlessly connecting these data management pieces together?

Colgrove: One of the biggest problems that companies such as Pure face is that we have all this confidential financial data. We can't just plop it into an AI learning bot, because then we're exposing it to everybody else. We have all this HR data about our employees. We can't just plop it into some large learning model because then it gets exposed to everyone else.

What companies have to do is focus on how they can build models and get learnings, but do so while not exposing the data beyond where it should go. There's a lot of data governance that has to go into that. There's a lot of data access [and] permissions that go into that. We've run into things like that where we have an AI-enhanced search engine that our [sales team] will use to search for past experiences with competitors. We occasionally find, through that search engine, a folder that has public permissions but shouldn't. AI helps expose the security flaws -- and faster.

It puts a huge premium on data governance, data management, data access, permissions -- all of that kind of stuff.

How does Pure's Enterprise Data Cloud fit into this convergence trend?

Colgrove: The easiest way to think about the Enterprise Data Cloud from my point of view -- a customer today, they buy a product and they use it. They then buy a separate product and they use it independently. They configure it independently, and they use it independently. And then they might buy five different products as a result, adding complexity to their environment. That complexity then makes it harder to manage your data, control your data, understand your data.

What the Enterprise Data Cloud is trying to do is to say as you acquire these products, you put them together into a fleet that makes a simple service -- one seamless service across all your storage assets -- and that allows you more simplicity, more control, better planning and more uniformity [in how you] access the data, control the data, apply where it's used and how it's used.

So the Enterprise Data Cloud is aimed at managing data sprawl?

Colgrove: It's dealing with that, but it's also helping you organize your data better. With the Enterprise Data Cloud, you have a few experts in your organization who define the levels of storage services you want to deliver. This goes back to what I was talking about -- where there are tradeoffs between performance and access. … They define the kinds of storage services you want to offer out to the organization, and then everything's organized into those few groups of tradeoffs.

These experts who define storage service levels, are they different from the storage administrators of today? Or are they the next evolution?

Colgrove: They're sort of the next evolution. … Today, the storage admin says, 'Oh, you need 10 TB of storage, I will configure it like this.' And they make an independent decision. Three weeks later, you say you need another 5 TB of storage, and they say, 'I'll provision it like this.' And they make another independent decision. What you get is, even with the best intentions, a mess of different things.

The senior-most storage admins, let's say they become storage architects, and they define Class A storage -- how often it will be backed up, how it will be replicated, what level of performance it will have. Now, when you come along and say, 'I need 10 TB of Class A storage,' you can get that. You can still go through someone in IT to give it to you, but they can also automate this -- and should.

It changes the way the jobs work. There are people who define the service levels. There are people who physically run the data center -- because you still need somebody who can decommission machines. … And then you have the people who use the actual results of [automatically] provisioning storage, but they do it within the parameters you laid out.

Do you think there will be a time when storage is run by AI agents?

Colgrove: I think that's very possible. We're starting to put AI copilot interfaces into our product. As we're doing with the Enterprise Data Cloud, we're talking about how you now can take predefined classes and put them into your AI-driven workflows -- whatever workflows you have. There's no reason you still have to have people go and do that. … The copilot interfaces are designed to let you interact with the fleet that you have in a better, easier, more natural way.

It also levels the playing field. If I can use natural language to interact with the system, then I don't have to have special programming, coding, education or training.

Colgrove: People don't have to become experts in the nitty-gritty details of how to do something. This lets them focus on what to do.

Let's do a round of quick-fire questions. Is Pure looking to make acquisitions for AI skills?

Colgrove: We're always looking at what we could acquire or not, and we'll make sensible decisions whenever. But we're not counting on anything like that to build our AI. We have people who want to learn it. We go and we hire new people all the time.

CIO budgets are in flux due to the tariffs and economic uncertainty. Is Pure seeing that reflected in discussions with customers or in buying decisions?

Colgrove: We do have customers who are talking about it. We think maybe in Q1 we saw some extra sales because of it -- or think of it as people rushing to get their purchases in before the tariffs [hit]. I suspect it's now dragged on long enough and it's been uncertain long enough that that effect has gone away, and it's almost business as usual in an uncertain environment.

I know Pure and Meta have worked together on AI projects. Meta just announced plans to build a new AI lab. Will it bring Pure along for the ride?

Colgrove: Meta would not like us to make any comments on specific uses.

Nicole Laskowski is the editorial director for Informa TechTarget's enterprise IT news team. She drives coverage for news and trends around AI, enterprise applications and IT infrastructure.