Pure Accelerate 2026 marked the first time the company formerly known as Pure Storage showcased its recently revised name and showcased its evolution from a hardware storage vendor to an AI-driven data management platform.

Customer feedback is important at the best of times, but amid a period of change for the company, comments on both technology and commercial models are arguably more crucial than ever.

SDxCentral sat down with a handful of customers at Accelerate in Las Vegas to get a sense of where the now-named Everpure is heading. And for the most part, consensus seemed positive.

Pharmaceutical giant Sanofi, for example, is using Everpure's FlashBlade//S 200 to meet the storage needs of its cryo-electron microscopy (cryo-EM) proof of concept.

Pradeep Bandaru, head of platforms and AI workflows at Sanofi, detailed that the firm put the storage solution through its paces with its cryo-EM projects – which are used to help determine structures of molecules in drug discovery – generating around 15 terabytes of data per run.

Bandaru noted that as microscope hardware improves and data capture speeds increase, the resulting data volume creates a severe storage and networking infrastructure bottleneck.

“Historically, we have many petabytes, actually, of cryo-bank data largely sitting in the cloud that we're trying to take off of the cloud and onto on-premises storage, and we do more intelligent data management types of things with that data that we're obviously taking off of the cloud," Bandaru said, adding that the Sanofi team made the switch to Everpure in a bid to save costs.

“The cloud-first workflow for this typically took about 40 hours [to] just push the data after it was collected to the cloud, and then process it … and that processing obviously used high-performance storage, like FSX or Lustre, as well as expensive cloud GPUs, and that led to many hundreds-of-[thousands] in cost, not to mention the cost of actually running these models in the cloud on historical data," Bandaru said.

The move to bring more of those workloads on-premises also aligned with the pharma giant’s spending when it comes to tokens – the economics of AI infrastructure. Bandaru said he was referring to the rising large language model (LLM) and token costs, more specifically, a “J-curve of token costs in the cloud,” and that a broader industry trend of shifting to on-premises was in response to some of the limitations of token spending.

Cost wasn’t the only factor behind Sanofi taking on Everpure, however, with Bandaru outlining the need for real-time signal from experiments and physical AI that cloud-only can’t deliver.

“Typically, when people run these experiments they would like some level of signal as to whether the experiment was actually correct or of good enough quality," Bandaru said. "Sometimes there's also the chance to pick up bad experiments as they're running, and then flag those, and then discard those, or stop the run. That’s the embedded feature of this real-time infrastructure that we built around the cryo-EM, and then integrated, obviously, with FlashBlade.”

Sanofi does use Snowflake in addition to Everpure’s offerings, but Bandaru positioned the rival offering as part of Sanofi’s cloud analytics and data products layer, distinct from the edge and FlashBlade workflows. Specifically, Snowflake is used for structured, meta-analysis workloads on historical data, not for the high-velocity cryo‑EM edge pipeline itself.

“We're trying to achieve a healthy balance of data that lives in the cloud and lives on the edge, and what we typically see is that in a world where we're confronted with increasing token costs, as well as … agents and large language models, don't really operate in physical environments, they hit a wall there. So we need to have different infrastructure that actually meets the needs of physical environments, like labs or manufacturing plants,” Bandaru explained. “That's where we start to deploy things more intelligently on the edge and bring the models directly to the data itself.”

Bandaru concluded by proposing the concept of an “edge data lake” or some kind of system that would allow Sanofi engineers to “bridge the gap on the data management side between the edge and the cloud, as well as ultimately the compute side too” to create some kind of hybridized compute fabric across the edge.

From à la carte to all-you-can-eat

Another customer who talked about their Everpure needs was Chris Weiss from RC Willey, which operates as much as a financing and credit company as it does a home furnishing company.

Weiss was bullish on the reason why the firm joined Everpure: cost. Having previously been long-time users of Dell Data Domain (now PowerProtect Data Domain), the exec admitted that when it came time for a hardware refresh point, the tech giant treated every capability as an extra SKU.

“We needed encryption. Dell's like, ‘we can do that. Cha-ching, here's the price.' We needed deduplication and compression. ‘Sure, dah-ding, here's the price.’ We want immutability, ‘Not a problem for this price.' And everything was à la carte," Weiss said.

If Dell were treating RC Willey like a visit to Morton's The Steakhouse, with each entree priced individually, then Everpure was like an all-inclusive buffet where everything is bundled into one largely predictable price. Weiss revealed that signing up with Everpure represented the same price for three units as one unit from Dell.

Cost considerations aside, like Sanofi, RC Willey switched to Everpure to meet performance needs. As a credit lender, the firm needs a fast recovery platform, not just a cheap, durable backup. Weiss described its original Dell offering as “a great appliance for backing up” but a “horrible” recovery solution.

“I tried to do an instant-on on Data Domain, and after 40 minutes I still couldn’t log in,” the exec told SDxCentral. “On the Pure solution, we could have it up, accessible, and logged in in two minutes. That was the difference.”

That need for a fast instant‑on data recovery solution pushed RC Willey toward combining Veeam with Everpure as a high‑performance backup target.

“We use Veeam, which can do an instant‑on, meaning you can spin it up on this compute running off the storage, and nothing's restored; it just runs in memory,” Weiss explained. “People can work off that, and then you can migrate that right into your production environment. … They don't know anything, but the system was down for a few minutes.”

Weiss said he was happy running Veeam on Everpure for instant recovery today, but he ultimately wants the pair to be smarter together.

In his view, Pure should be able to say, "It's not here, but Veeam’s got five copies of it," while Veeam can confirm those copies are accessible and malware‑free. Almost like turning backup plus storage into an end‑to‑end data‑intelligence layer rather than two point-products – a concept that ties neatly into Everpure’s pivot from solely storage to what the vendor describes as “data primacy,” with it building secure, context-driven layers to get the utmost from disparate sources.

“Someday somebody's going to have to develop just the dashboard that talks to everything and gives you back the information, because if not, we're still siloed all this stuff that we're talking about fixing,” Weiss added.