The agreement gives Everpure another major customer opportunity as data centers look for denser, more power efficient ways to handle growing volumes of data 

[For more news, click here.] 

Everpure has secured a design win and supply agreement with a second top five hyperscaler, extending its footprint in a data center storage market that is being reshaped by the rapid expansion of AI infrastructure. The deal underscores how hyperscale operators are rethinking storage architecture as AI workloads demand faster access to ever-larger volumes of data. 

Everpure has not identified the new customer, and the company has not disclosed the value or expected deployment size of the agreement. Everpure expects the deal to become a meaningful contributor to revenue beginning in fiscal 2028, so the near-term significance lies less in immediate sales and more in what a second major hyperscaler is willing to adopt. 

The announcement follows Everpure's first hyperscaler design win, signed in late 2024. Shipments for that deployment began in fiscal 2026, and Everpure said it exceeded its annual shipment target for the segment. 

Why Data Center Storage Is Changing 

The growth of AI is putting pressure on more than computing capacity. Data centers also need to store increasingly large datasets and make that data available to systems that must access huge amounts of information quickly, a challenge that is pushing operators to reconsider how storage infrastructure is built. 

Meta outlined one part of the problem in a 2025 engineering paper on QLC flash. Hard drives have continued to grow in capacity, the paper noted, but their input/output performance has not kept pace. As a result, the bandwidth available per terabyte has declined, creating difficulties for workloads that need to access data frequently, a common pattern in AI training and inference. 

QLC flash is emerging as a potential middle tier between traditional hard disk drives and more expensive TLC flash. It offers substantially greater storage density while providing better power efficiency than HDD-based infrastructure for certain workloads. Meta expects QLC-based storage to eventually deliver meaningful reductions in both acquisition and power costs as flash density continues to improve, a trend with direct implications for data center storage economics. 

For operators running data centers at enormous scales, that shift matters. More capacity within the same physical footprint can translate into fewer servers, less rack space and lower overall infrastructure requirements, all of which affect the total cost of running AI-scale storage systems. 

Meta Has Already Tested the Approach 

There is already a real-world example of where this storage strategy is heading. Meta has said its storage teams have been working with Pure Storage, Everpure's former corporate name, using its DirectFlash Modules and accompanying software to bring QLC storage into its data centers. 

According to Meta, those modules can scale to as much as 600TB using the same NAND package technology, with a target of six times the byte density of Meta's densest TLC-based server. The technical challenge extends beyond simply fitting more flash into a server. Meta's engineers also had to adapt their storage software to handle the unusually high density and data-processing demands that QLC systems introduce. 

For Everpure, that collaboration provides a concrete example of its technology being evaluated in a hyperscale environment, rather than relying solely on the company's own claims about performance and efficiency. 

The first major deployment is also moving beyond its initial design stage. Everpure said shipments exceeded its forecast in fiscal 2026, with additional shipments expected in fiscal 2027. The company expects its integrated hardware-and-software approach to reduce power and space requirements while lowering maintenance costs and failure rates compared with legacy disk storage systems. 

A Second Customer Changes the Picture 

The latest agreement matters because it gives Everpure a chance to show that its approach to AI-era storage can extend beyond a single hyperscale deployment. 

“Our growing success in hyperscale environments is based upon a foundationally more advanced architecture and technology than legacy storage solutions. Securing a design win with a second hyperscaler for our hyperscale solution signals strong recognition of the economic, operational, and performance advantages of our DirectFlash technology. DirectFlash sets a new benchmark for efficiency, density, and performance for data storage at scale,” said Charles Giancarlo, Chairman and CEO of Everpure. 

Still, much about the agreement remains unknown. Everpure has not named the new customer, disclosed the size of the deployment or provided financial terms, and the expected revenue contribution remains more than a year away. 

That makes the second win notable for a reason that extends beyond the deal itself. One hyperscaler adopting a new storage architecture can be read as an experiment. A second suggests the underlying approach may have broader market traction among the largest operators of AI data centers. 

As AI systems continue to demand more data, the next phase of infrastructure competition may not be defined only by who has the fastest chips or the largest data centers but to how much information those data centers can store in the smallest footprint, using the least amount of power. 

Related Articles 

VAST Data Is Now Worth $30 Billion, The Real Story Is Why 

VAST Data and AMD Expand AI Infrastructure Alliance to Tackle the Industry's Inference Bottleneck 

Core42 Is Building the AI Compute Network That Doesn't Pick Sides Between AMD and NVIDIA