What’s Next is a monthly conversation between SDxCentral CEO Matt Palmer and a senior-level executive from the technology industry. Each month, Matt has an informal but in-depth video chat with a fellow thought leader to uncover what the future holds for the enterprise IT and telecom markets – the hook is each guest is a long-term acquaintance of Matt’s, so expect a lively conversation.

This month, SDxCentral CEO Matt Palmer spoke with Kevin Deierling, SVP of networking for Nvidia. Known for both his technical and business expertise, Deierling has led both research and development in areas including networking, cloud, security, big data and DNA sequencing, among many others.

Editor's note: The following is a summary of the conversation between Deierling and Palmer, edited for length. To hear the full conversation, be sure to watch the video.

Networking as the foundation for AI

When asked what artificial intelligence (AI) means to your network, Deierling said, “The fuel for AI is data, and the starting point is getting that data into the system.” Networking, he said, is a critical part of getting data into the data center. Going further, calling the network the backbone of the data center, Deierling said networking was “the foundational element of these new generations of AI computing.”

At the same time, he said AI is being applied to networks, primarily to optimize the network. Especially now, where networks have become so complex, with giant systems of connections, AI is needed to keep track of all the moving parts. “Automation is really important,” he said. “And you need to apply AI [to the network], then, to actually make sense of everything that's happening.”

Nvidia’s platforms, frameworks and Morpheus

Kevin Deierling: So we are building platforms for security and monitoring. But Morpheus takes in massive amounts of telemetry data. So rather than just sort of spot-checking things, we look at everything going on in the network. And then we can identify security issues and anomalous behavior. Because maybe something's degrading and a device is about to fail. And so we can use AI to actually generate digital fingerprints of machines, of individuals, of devices. And when they start to deviate from normal behavior, then we can detect that automatically.

Matt Palmer: This Morpheus — is this actually a product that you're shipping? Is this the code name inside your matrix, sort of, without going too far down the pun?

Deierling: Yeah, you got the pun. What we see is that threats are morphing all the time, like in “The Matrix” — changing shapes. It's not really a product in the sense that we don't build things like security solutions that go to end customers. But it's a platform or a framework (we have lots of frameworks). So we have frameworks for large language model (LLM). We have frameworks for virtual worlds or omnibus platform digital twins.

Morpheus is a platform that our partners can consume and then utilize inside of their products that they take to market. One of the things that we see is that the threats that are actually impacting data center and enterprises are getting more and more sophisticated. And one of the reasons is because of AI — people can use it. And so we need AI to actually address those types of threats.

On getting started with AI

Palmer: If you're building an enterprise-class data center today and you're looking to start solving big problems. How do you recommend that somebody get started?

Deierling: Often, the learning curve of AI is just too big. It frightens people away. Obviously, the easiest is just to get started in the cloud. And so we've stood up what we call DGX Cloud, and it's available from Oracle, or GCP. Or Azure, or smaller vendors like CoreWeave. We have the entire Nvidia AI Enterprise stack, which has a whole bunch of vertical workloads. That's really where you can leverage all the work that our R&D team has done to set up these frameworks. That's what they're built for is to get customers to market faster, using AI for all sorts of different verticals.