Organizations adopting artificial intelligence soon discover the twin challenges of AI-scale data performance and global data orchestration. At that level, only a unified data platform can control capacity and policy across distributed environments.

Lack of centralized control forces organizations to rely on reactive capacity planning and siloed monitoring tools — introducing risk and escalating operational cost. What’s needed instead is both a high-performance data engine, as well as a unified operating and intelligence layer that delivers manageability on a global scale, according to Jonsi Stefansson (pictured), general manager of cloud at Vast Data Inc.

“There is [a] certain level of supply chain crisis in the world today,” Stefansson noted. “You have to have the flexibility to actually figure out: Where do I have compute availability? Where do I have GPU availability? Where does my data reside? [At Vast] we want to solve exactly that.”

Stefansson spoke with theCUBE’s Dave Vellante and Rebecca Knight at Vast Forward 2026, during an exclusive broadcast on theCUBE, SiliconANGLE Media’s livestreaming studio. They discussed how to address the performance and scale problems associated with raw data storage for AI through a unified data platform, as well as solving the fragmentation problem that emerges as infrastructure becomes distributed. (* Disclosure below.) 

From siloed storage to a unified data platform

Disconnected storage models were never built for AI scale. A range of enterprises have refocused their efforts on file and object storage to reflect the move away from siloed platforms. But what organizations really need is a unified platform uniting storage and the operating system, Stefansson explained. 

“Storage is the key, but ultimately it is the AI operating system capabilities of unifying or consolidating your entire AI pipeline into a single platform that is right next to the storage [that’s needed],” Stefansson said. “You can basically start contextualizing all your unstructured data within the same platform without having to migrate it.” 

But the shift isn’t just about reducing copies or chasing lower costs. Instead, it’s about giving enterprises the ability to dynamically place data and compute wherever opportunity appears — without losing control of context or blowing up storage footprints, according to Stefansson. But the focus is squarely on efficiency, not total reinvention.

“We are not breaching the laws of physics,” he noted, in regards to the realities of moving massive AI data sets across clouds and regions. “We’re just making it more efficient.”

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

(* Disclosure: TheCUBE is a media partner for Vast Forward. Sponsors of theCUBE’s coverage, including presenting sponsor Solidigm, do not have editorial control over content on theCUBE or SiliconANGLE.)

Photo: SiliconANGLE