To determine the power, reliability and scalability of the next generation of artificial intelligence systems, AI infrastructure upgrades are essential because they determine performance, reliability and broad usability.

To accelerate AI infrastructure upgrades, Solidigm launched an AI Central Lab that enables advanced solid-state drive benchmarking through a strategic partnership with MinIO Inc., according to Avi Shetty (pictured, right), senior director of AI enablement and partnerships at Solidigm, a trademark of SK Hynix NAND Products Solutions Corp.

“We’ve had a fundamental shift in the way we do SSD benchmarking,” Shetty said. “Our goal with investing in this AI Lab was to have our engineers, as well as our partners access to a real world AI infrastructure with the state-of-the-art GPUs from Nvidia and the state-of-the-art networking, 800 gigabit networking where the GPUs are talking amongst each other at 800 gigabit. The storage is talking to the GPUs at 800 gigabit with Solidigm SSDs and then with our partners like MinIO will manage that cluster for us.”

Shetty and Garima Kapoor (left), co-founder and co-chief executive officer of MinIO, spoke with theCUBE’s Dave Vellante and Savannah Peterson at SC25, during an exclusive broadcast on theCUBE, SiliconANGLE Media’s livestreaming studio. They discussed the critical role of AI infrastructure upgrades and how the Solidigm–MinIO partnership is turning this vision into reality. (* Disclosure below.)

AI infrastructure upgrades that maximize GPU efficiency — Inside the Solidigm–MinIO partnership

The Solidigm–MinIO partnership marks a major turning point in AI infrastructure upgrades. By uniting Solidigm’s high-performance SSDs with MinIO’s enterprise object storage, the companies deliver a unified, data-centric platform that removes bottlenecks and boosts GPU efficiency. As AI workloads become increasingly data-intensive, this collaboration enables faster deployments, lower costs and truly scalable performance, Shetty pointed out.

“We announced the world’s highest performing test cluster with our PS1010, Gen5 TLC SSD and also the world’s most dense test clusters, the 122 terabyte drives,” he said. “We have 192 of those in 16 1U slots getting you 23 petabytes, and we have MinIO software now managing that, thus giving MinIO [a] representation of actual TCO as well as our joint customers access to the lab where they can test out this infrastructure before they make their choices. For anyone who’s interested in accessing the lab, just go to storageforai.com and you have all the instructions.”

The Solidigm–MinIO collaboration enables organizations to create actionable AI plans by rigorously testing and validating AI architectures. This process turns abstract concepts into data-driven, execution-ready strategies by revealing real-world performance, bottlenecks and compatibility gaps before deployment, according to Kapoor.

“What our customers are looking for is a very actionable plan,” she said. “This testing and reference architecture gets exactly that in terms of what kind of networking you need and what kind of compute do you need, what kind of performance you can expect, and that makes it very actionable for our end customers to use it. That’s why partnering with Solidigm was extremely important for us.”

AI performance hinges not just on GPU speed but on how fast data reaches it. Solidigm’s AI Central Lab prioritizes storage as a core component, not an afterthought, marking a transformative leap in AI infrastructure, Kapoor noted.

“You want a system in place that is scalable, that works on the TCO for the customers, and that gives them the performance that they are expecting for their AI application workloads,” she said. “That’s where this lab is extremely important or exciting for us to partner with Solidigm, along with Intel and Supermicro to make sure the customers get what they expect.”

Here’s the complete video interview, part of SiliconANGLE’s and theCUBE’s coverage of SC25: 

(* Disclosure: Solidigm sponsored this segment of theCUBE. Neither Solidigm nor other sponsors have editorial control over content on theCUBE or SiliconANGLE.)

Photo: SiliconANGLE