Walk the show floor at any infrastructure conference in 2026 and you’ll hear two letters more than any others. But when I sat down with Dan Kogan, Everpure‘s VP of Cloud and New Products, while at the Pure Accelerate event in Las Vegas this week, the conversation that stuck with me wasn’t really about AI at all. It was about money—specifically, the math most companies get wrong when they move storage to the cloud, and why the loud industry narrative about everyone fleeing the cloud doesn’t match what he’s actually seeing.
Kogan has been around the cloud and virtualization story for a while, and I’ve been covering both since 2004, so we got to talk shop without much warm-up.
The cloud cost math nobody runs
Here’s the pitch in its simplest form, and it’s a good one. Everything you put in the cloud is thick-provisioned. No data reduction, no snapshots, no efficient copies. If you have a petabyte of data, you pay for a petabyte of cloud storage—full stop.
Now run that same workload through Everpure’s data reduction software sitting in front of Azure or AWS. Kogan says you routinely see four-to-one or five-to-one reduction, sometimes higher. So your petabyte becomes 250 terabytes. Instead of paying the cloud provider for a full petabyte of premium SSD or Ultra disk, you’re paying for a quarter of that, plus some Everpure licensing. The net, in his telling, is 50 to 60% off the storage bill, and in some cases more.
He was honest about the caveat, which I appreciated. This works beautifully on the right data—virtualized databases, SQL workloads, the bread-and-butter stuff Everpure has always handled well. It does not work on already-compressed or encrypted data, where there’s nothing left to squeeze. So this isn’t a magic wand for every byte. But for the workloads where it applies, the savings are real and meaningful.
“Save money here because you’re going to spend it there”
The reason this matters right now is AI, and Kogan connected the dots in a way that should land with any CIO staring at a cloud invoice. When you get sticker shock from the cloud, it’s usually because you’re overspending on the boring plumbing—storage, compute, networking—not on the high-value services you actually went to the cloud to use.
His framing was blunt: “Save money here because you’re going to spend it there.” Trim the storage bill on the core infrastructure, and you free up budget for the AI Foundry, Fabric, and analytics services that were the point of the move in the first place. He used a CPG customer as a real example, one that laid out specific reduction numbers and the savings that came with them, then redirected that money toward higher-value data services.
So is this finally making customers care about storage efficiency again, after years of treating cloud capacity as something you just buy more of? Kogan’s answer was a flat “that’s our value prop, exactly.”
The repatriation narrative gets a reality check
I asked the question everyone’s debating: what’s harder, convincing customers to move to the cloud, or convincing them to bring workloads back on-premises? His answer was refreshingly unsentimental.
For Everpure, he said, it honestly doesn’t matter much either way—they play on both sides. But he pushed back hard on the fashionable idea of mass repatriation. Once a company has truly committed to the cloud and closed its data centers, going back is genuinely difficult, because you’ve left the data center business entirely. More to the point, he thinks the full-scale repatriation story is mostly told by vendors with an agenda. “That’s a hopeful story of vendors that probably don’t have a cloud play,” he said. If you can’t help a customer in the cloud, of course you’ll argue everyone should come home.
That said, he didn’t dismiss it completely, and this is where the nuance lives. He sees pockets of it—cheap-and-deep archive workloads that eventually make more sense on-prem, and a real resurgence in sovereignty initiatives. Azure Local and AWS Outposts, which he described as products on “life support a couple of years ago,” are now genuinely useful again in private and sovereign cloud scenarios. So the repatriation story isn’t fiction, it’s just far narrower than the marketing suggests.
The VMware question, from a storage angle
You can’t have this conversation in 2026 without Broadcom coming up. Kogan’s read tracked with what I’d been hearing all week, but his vantage point is storage and cloud rather than compute, which gave it a different flavor.
He thinks the exodus talk is overblown and that things have stabilized. Some customers are leaving outright. Plenty are staying and optimizing their licenses. A meaningful number are moving a percentage of workloads to alternatives, with Nutanix and Red Hat OpenShift coming up most often—Everpure’s Nutanix integration, he noted, has been smooth enough that customers light up in the telemetry without the team even prompting them. He also floated a candid theory about Broadcom: that they may have overplayed their hand, pushing pricing on customers they actually wanted to keep, then quietly coming back with deals to hold onto them.
The throughline is that there’s no one-size-fits-all answer, and Everpure has made a point of being storage-ready across all the destinations—VCF, Nutanix, OpenShift, Azure, AWS—so it doesn’t much care which way a given customer jumps.
Wearing two hats: cloud and new products
Kogan’s title includes “New Products,” so I asked what’s newest on his plate. His answer was OEM—a motion he said Pure has never really had before. Everpure recently landed a large Telco edge network operator that selected it as the storage layer for a full 5G edge network sold to the world’s leading operators. Turning that into a repeatable channel means building fully integrated products, commercial offerings, and supply chain pricing from scratch, with hopes of winning more OEM deals from there.
And his pick for the product people will be talking about over the next year? Data Intelligence. Not Everpure’s first software product, he was quick to note, but probably the biggest leap, because it reaches beyond the Pure ecosystem and beyond storage-array concepts into the data management stack itself.
What an IT pro should take from this
If you remember one thing from Kogan’s view of Everpure’s evolution, he said it’s this: the move from Pure Storage to Everpure is an expansion of the value stack, not a pivot away from storage. AI is all about data, the data lives on storage, and the company is widening the aperture to cover more of what customers actually struggle with.
But the practical takeaway, the one you can act on Monday morning, is the cloud cost argument. If your cloud storage bill gave you sticker shock, the answer probably isn’t a painful, full-scale retreat back to your own data center. It’s looking hard at how efficiently that data is stored in the first place. Run the numbers on data reduction before you run from the cloud entirely. As Kogan put it, the unhappy cloud customer is usually the one paying premium rates for plumbing while getting none of the value—and that’s a fixable problem.
##
