Whether it’s training large-scale artificial intelligence models or building applications that implement AI in financial services, the success of these and other AI-driven initiatives depend on the ability to store, move and process massive volumes of data at extreme speeds.

A major challenge to achieving this performance is that today’s graphics processing units are now much faster than the data systems feeding them. This mismatch between compute and data creates a bottleneck, driving demand for AI factory solutions, according to Alex Bouzari (pictured), chief executive officer of DataDirect Networks Inc.

“[An AI factory] is a way to generate and create business value, business outcomes and financial outcomes for organizations who are looking at making investments in AI,”  Bouzari said. “It is not about technology. It is a tool that has the flexibility to handle a whole pipeline of AI from models and model training, to analytics, to inference.”

Bouzari spoke with theCUBE’s John Furrier and Dave Vellante at the SC25 event, during an exclusive broadcast on theCUBE, SiliconANGLE Media’s livestreaming studio. They discussed how AI is situating itself in today’s enterprises — including examples of the deployment of AI in financial services — and the increasingly important role of AI factories. (* Disclosure below.)

Marrying data and compute for AI in financial services

Companies such as Nvidia Corp. are providing the underlying infrastructure AI needs in order to generate outcomes that drive value in banking, investing and other industries. The challenge is getting data and compute to work together to facilitate the consumption of AI across the enterprise, from the software stack to networking to storage, according to Bouzari.

“It’s an orchestration layer that needs to facilitate all of this,” Bouzari explained. “There have been debates about, ‘Well, should it be file system? Should it be parallel file system? Should it be objects? Should it be S3?’ It’s not about that. It needs to be a variety of technologies that come together so that customers can consume AI resources in order to get business and financial benefits.”

The critical thing is for enterprises to remain focused on the ultimate value that the customer needs to achieve, Bouzari noted. That focus should guide how they approach AI projects from infrastructure to applications. 

“You have to deliver value to customers,” he said. “For that, you have to understand their software stacks, their frameworks, their pain points and bring in an AI capability that enhances the value that they’re delivering to customers.”

The focus on AI factories is similar to the attention placed on sovereign AI, Bouzari noted. Both concepts aim to align compute and data in ways that produce tangible outcomes for organizations.

“AI factories is really for enterprises and businesses,” he explained. “Sovereign AI is for government, but it’s really kind of the same thing in that you have to add value and it all means tying in and marrying the computer and the data together.”

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

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

Photo: SiliconANGLE