Renen Hallak

"once you want to start inference, then it needs even faster access because you have trillions of agents eventually all needing to access all of this data in parallel. And it's not one hop inference, it's multi hop for reasoning."

"The world shifted from training to inference. We're now moving from inference to agentic workloads. The organizations that are doing AI are changing from experts, the model builders, the big AI clouds to enterprises, all of that means that the weight is shifting from compute to data."

"What is the operating system? It's my analogy for that software infrastructure layer. We're not hardware people, we're not AI model people, but we want to provide that middle of the sandwich software infrastructure stack that includes storage, that includes database, that includes scheduling of jobs and orchestrating compute."

"when you build that operating system, I like that analogy because it's right in the middle between the hardware and the application. And NVIDIA obviously built all of the hardware pieces, the GPUs, the DPUs, the networking for these new AI workloads, and they're also in that enterprise application layer."

"It includes security. All of the parts that you need to make it simple, such that regular enterprises can start writing and using AI apps to make it secure, such that they can comply with regulation."

"Governance is super important. I think one of the big reasons why the enterprise isn't moving faster is because they have compliance and regulation and other things that they need to adhere to. For example, access control lists, who is allowed to see what information?"