Everpure Expands Enterprise Data Cloud Strategy Via New AI-Focused Software Advances

Everpure unveils software-focused data management advances, including Universal Data Intelligence and Data Stream, aimed at helping enterprises discover, govern, contextualize and vectorize data for AI while giving partners new opportunities around data-centric modernization.

High-performance storage and data management technology developer Everpure Wednesday used its Pure Accelerate 2026 conference in Las Vegas to introduce its newest architecture aimed at enabling discovery, context and governance for data used in advanced AI applications.

Everpure, which until February was known as Pure Storage, also introduced Everpure Data Stream to accelerate enterprise AI projects by helping remove cost and complexity barriers.

Chadd Kenney, vice president of product management, told CRN that the news follows last year’s introduction of Everpure’s Enterprise Data Cloud, a unified data plane that effectively allows a business to run all of its workloads from archive to AI, as well as its Intelligent Control Plane, which helps build autonomy around operations including data governance, data controls and automation.

Everpure this year is introducing its Universal Data Intelligence layer, which brings applications and data from inside and outside Everpure infrastructure into a single layer that stores the shared context of all of those applications, he said.

“This lets customers understand their data and make meaning of it for a world of AI, as well as other things for governance and compliance and the like,” he said.

For Everpure’s unified data plane, which virtualizes and connects data across every location into a single fabric, the company added several new capabilities, Kenney said. One of those, Purity Turbo for the Everpure FlashArray//XL190, adds performance by taking advantage of a secondary controller.

“We’ve never really done that before, but it’s useful for read-centric workloads,” he said. “As an example, let’s say you're running a backup or an ETL [extract, transform and load] process at night. If it potentially would mess with the SLA of a production workload, it’ll actually move those workloads to the secondary controller to make sure those SLAs are met. This turbo capability provides incremental headroom.”

The second performance addition is OverDrive, which Kenney said is part of Everpure’s Evergreen//One Storage-as-a-Service technology.

“OverDrive provides incremental performance, and users only pay for that performance during a certain 24-hour period time frame,” he said. “You only pay for it when you need it. You get performance agility when you need some incremental performance.”

The company also added Azure-native virtual machines to its Everpure Cloud technology, formerly known as Pure Storage Cloud, Kenney said. This lets customers run native VMs in Azure on top of its Everpure Cloud, he said.

For object storage, Everpure is also building new capabilities around object storage including life-cycle policies, notifications, tagging and strong consistency, Kenney said. “Strong consistency” ensures that data read in one location is consistent with the same data read from another location, he said.

On the file storage side, Everpure is unveiling ActiveCluster for File, similar to what it previously offered for block storage, Kenney said.

“The big win there is it’s at a fleet level, so we’re now managing it across the unified data plane,” he said. “And we’re making it policy-oriented so it can move across systems as you change the service levels within it. It also works across multiple data centers. We’ll show on stage a demo between Mountain View and Sunnyvale. Fully synchronous, active-active. Then we’ll kill one of the data centers, but the data will be fully up and running on the other side with no delay or disruption.”

For its Intelligent Control Plane, Everpure is introducing the ability to do compliance reporting and remediation to simplify policy changes, Kenney said.

“Let’s say you had a ransomware attack, and the environment needed to change to a 30-day retention policy for snapshots,” he said. “Many environments are all ad-hoc-built. There are no data controls, so you’d spend months trying to find all the things you needed to update. In a world with data controls where you imply intent, you can just go change the policy, and it updates all of them and then monitors to make sure they stay that way. We call this Compliance Reporting And Mediation.”

Also new is dynamic rebalancing and mobility, which lets customers get predictions of performance challenges if a workload is moved, Kenney said.

“It will predict a performance challenge before it occurs and move that workload to an alternate system without disruption,” he said. “This not only helps ensure SLAs are delivered on, but also efficiently manages capacity across environments.”

Everpure has also made its February acquisition of 1touch a component of its new data management platform, Universal Data Intelligence, Kenney said.

“Universal Data Intelligence gives the ability to understand the meaning of data, whether it’s on Everpure or on SaaS Infrastructure as a Service or even on other people’s storage arrays,” he said. “It discovers what data sources are out there, and then it classifies the data to understand the context and then contextualize it, building a knowledge map on the meaning of data.”

Universal Data Intelligence is a fantastic move for Everpure, said Ned Engelke, CTO of Evotek, a San Diego-based solution provider and Everpure channel partner.

“Data discovery, data management and data control is a hot space because it is critical, and what we’ve seen in the past is most storage vendors mostly have wanted to focus on talking with customers about IOPs [I/Os per second] they can get out of their gear,” Engelke told CRN. “And that’s fine. But what our customers really need is, ‘How do I find things? How do I categorize them? How do I understand if I’ve got liability around this? How do I move it if I need it in a different location? How do I do metadata tricks to make my application think that it’s been moved without actually moving it because of the time it would take to move across the network?’ All of those elements are a part of that technology.”

Everpure’s vision is spot on, Engelke said.

“I care about why do you have storage?” he said. “Well, it’s because I want to have data. Why do I have data? Well, I would need to find it to have an advantage for building an application that can talk to it. I need to go through this whole life cycle. So I really love the idea that they’re integrating it with turning this into a platform for data and data management to expand on their value proposition of being a great storage platform.”

A key component of that is Data Stream, which Kenney said is the ability to vectorize data to make it ready for AI.

“We built a platform that will look at the data, chunk and vectorize the data, and then store that in a vector database used for retrieval augmented generation, or RAG, for AI deployments,” he said.

Applications today are for the most part siloed with their own context, data and workflows, Kenney said.

“As agents start interacting with these things, they don’t have a shared context or shared understanding of what’s going on in each of these,” he said. “Humans do. You know customer data is in here, and that invoices are over there. You may know the context of each of these. But no one’s really been able to nail how we’re going to actually get understanding broadly, and so this Universal Data Intelligence layer effectively takes the context from all of these locations and allows agents of next-gen applications as well as analytics to use that as the new substrate. I think that’s one of the more cool visions that we have, that application centricity is going to transition to data centricity, and data will become the center of gravity and be primary versus the app, and then context will become the new substrate in which applications are developed, agents consume, and analytics environments get access to manage those things.”

Engelke said Data Stream looks to be very important as the use of GPUs, which have big memory spaces that handle data orders of magnitude faster than any other storage media, continues to grow.

“The name of the game for going fast with within a training or inferencing environment is to have the dataset you’re trying to work against in memory and then do intelligent pre-fetching to get it from what is effectively archival storage, like NVMe,” he said. “How do you go through and ask the question to get it? How do you transit it? How do you make sure that you’re pre-fetching? How do you collaborate with it? It requires a design partnership between Nvidia and the storage platform, and that’s been something that we’ve seen from a few other vendors in the past, but Data Stream seems like a very effective way of connecting storage for those workloads that are particularly performance-sensitive.”

Data scientists tend to be very expensive and hard to get, and so the last thing a customer wants is to have that person staring at a beach ball while waiting for a model to get trained, Engelke said.

“So the faster you can go, the more you get out of your investment in people,” he said.

While Everpure is best known as a developer of high-performance all-flash storage arrays, the company is not discussing new hardware at Accelerate, and Kenney said that was by design.

“We’re kind of in our software year,” he said. “Last year we did a complete update of the entire platform, the FlashArray//X and FlashArray//C. The FlashArray//XL got an r5 last year, and the XL190 Flash Blade got an r2, and there were more updates. We updated the entire portfolio. You will start seeing us looking at storage as the resource which delivers outcomes for customers, with less centricity around the particular boxes and more around what use cases we can help them be able to deliver.”

When asked about new training or certification for partners on the new technology, Kenney said it related to where the company is hoping to take customers, including evolving how people think about things like policies and operations, he said.

“We created the Enterprise Data Cloud Maturity Model, and we’ve been training both our internal field and partners to help customers along that maturity model so they can look at the values and outcomes they can get delivered,” he said. “There’s training and certification, and a bunch of services that are great for our partners to help customers rationalize what they do. I joke that it’s somewhat like giving somebody who’s been very disorganized for a long period of time a bunch of boxes to organize things, and all of a sudden it’s, ‘What am I supposed to do with these things? I’ve never had to do that before.’