Realizing value from AI investment has become the ultimate prize for most enterprises today. This requires the ability of data and compute to work together, a key mission for the partnership between Nvidia Corp. and DataDirect Networks Inc.

Both companies have been collaborating on solutions that facilitate consumption of AI across an organization, with a focus on optimizing the use of GPUs inside the orchestration layer.

“Nvidia is now an AI infrastructure company, so it’s all about building infrastructure and creating value out of that infrastructure,” said Alex Bouzari (pictured), chief executive officer of DDN. “I think the value creation and monetization, making GPUs productive, making GPUs profitable is what it’s all about. That’s what we’re razor-sharp focused on.”

Bouzari spoke with theCUBE’s Dave Vellante for theCUBE + NYSE Wired: AI Factories – Data Centers of the Future interview series, during an exclusive broadcast on theCUBE, SiliconANGLE Media’s livestreaming studio. They discussed how DDN and Nvidia are focused on building new architectures to realize value from AI.

Maximizing AI investment for agents

To maximize AI value, Nvidia and DDN have been working on solutions that strengthen GPU efficiency across AI factories. Earlier this month, DDN announced new advancements in its AI data intelligence portfolio aligned with BlueField-4, the storage processor utilized within Nvidia’s Vera Rubin AI platform.

“Vera Rubin was developed specifically for agentic use cases,” Bouzari noted. “A chatbot sends one request; an agent sends 30 requests. That’s 30 times the power, 30 times the compute, 30 times the data. Unless you have a framework and AI infrastructure where the compute is highly efficient and the data is highly efficient, well, your agents cannot function. You have to bring in novel approaches, novel architectures developed specifically for this kind of scale for AI to shift into this new world of agentic enablement.”

Another factor in the drive for value creation involves cost per token, a pricing metric employed by model providers to quantify the expense of processing text tokens in AI workloads. As the cost of AI deployment becomes more of a concern for enterprises, infrastructure builders such as Nvidia and DDN are focused on lowering token costs by a meaningful amount.

“The economics have to pencil out, and that’s significantly lowering the cost per token,” Bouzari told theCUBE. “Nvidia is talking about improving cost per token by a factor of 10, by a factor of 20. We’re doing it day in, day out across industries and customers. You connect these two things together, and that’s how rapid adoption of enterprise AI happens. When Elon [Musk] talks about the addressable market of SpaceX being close to $30 trillion, that requires enterprise adoption of AI. And for that, you need the data layer to be enabling. That’s the job that DDN does.”

Here’s the complete video interview, part of SiliconANGLE’s and theCUBE’s coverage of theCUBE + NYSE Wired: AI Factories – Data Centers of the Future interview series:

Photo: SiliconANGLE