Transcript
A (0:00)
Foreign.
B (0:05)
Welcome to Reshaping Workflows with Dell Pro Precision and Nvidia, where innovation meets real world impact in high performance computing.
A (0:19)
This is Logan, reshaping workflows, GTC 2026. I'm Logan, your host. You know me here with Gregory of F5. So, Greg, before we get started, I'm going to hand you the mic, tell everyone your role, kind of what you do, and then a brief overview of F5.
B (0:33)
Sure. So, Logan, Greg Dahl here. I'm director of Product management for AI Solutions. And so what we do is we're bringing solutions for, in terms of traffic management, application delivery for those big AI clouds. And so we are presenting a bunch of new things this week at gtc. And so it's not as sexy as the kung fu robots, but we are actually showing how to optimize AI factories and AI clouds in general, running traffic management, security, and trying to optimize the number of tokens that these factories can generate.
A (1:11)
All right, the podcast audience, some are very technical, know exactly what you just said. Some maybe on the ITDM side understand, but maybe not at a Leopard. So we say AI factory, right? We're thinking cloud, we're thinking data center, we're talking multiple GPUs kind of clustered together. But let's talk, let's say one, you said kind of helping the speed and the orchestration of that. What exactly does that mean? And what is F5's responsibility in that?
B (1:33)
Yeah, so basically think now let's focus on inferencing, right? And so you need, you get a request and it comes from a user somewhere on the Internet. You need to route it to the proper spot and to the proper nodes where the models are running. And so we optimize from that point up to the worker nodes where the models run, the application run. And we do load balancing. That's where we started about 30 years ago. But we optimize that for AI. We do LLM routing, intelligent AI load balancing. So for example, being able to send the traffic to the right servers based on the load of the GPUs.
A (2:11)
So load balancing makes sense, especially if you're handling like, you know, multiple requests across distributed. But you mentioned, you know, there's a couple, you said not as, maybe as sexy as robots, which, I mean, maybe it is, maybe it isn't, depending on who's listening. But you said you had several announcements, maybe give or, you know, new product things that F5 came out with this week. Maybe give me your top two.
B (2:29)
Yeah, so, and you'll have to come to, to my session tomorrow. But we're going to talk about the AI grade in particular and how to do this distributed inference and routing the traffic from the user to the application. So that's, that's one thing. And we'll show the integration with LLM routing in particular for that. But also very basic stuff like DNS, which is becoming cool again for AI. And then. So that's one thing. The other thing that we are talking about a lot is AI guardrails. Right. And I think security is obviously becoming more and more important as people really adopt AI. And so we have a. We had the acquisition a few months ago of company called Calypso AI and so we're presenting that this week and some demos on this.