Anthropic & OpenAI Are Hardening Cisco Infrastructure | Jeetu Patel

Jeetu Patel (0:00)

7 to 9 minutes.

Sophia (0:01)

7 to 9 minutes, I was going to say an hour. 7 to 9 minutes is way worse.

Jeetu Patel (0:05)

Then you start thinking about, okay, so how long does it take for a patch to be applied to infrastructure from the time that a vulnerability is announced? About 45 days.

Sophia (0:15)

Oh gosh, really, 45 days.

Jeetu Patel (0:17)

The exploit starts happening in 7 minutes. The bad actors, the adversaries, are already attacking a system within 7 to 9 minutes, and it's taking you 45 days to go out and do a patch. So the current system simply does not work. What you have to do is fundamentally re-architect it.

Sophia (0:34)

Hello, everyone, and welcome back to MTS. Today, I am joined by Jeetu Patel, who is the President and Chief Product Officer at Cisco, and he oversees Cisco's product portfolio across networking, security, compute, Splunk, Silicon, and the infrastructure companies are now rebuilding for AI. But one of the exciting news is that Cisco is also working with both Anthropic and OpenAI on frontier AI systems designed to find security vulnerabilities before attackers do. Jeetu, welcome to MTS.

Jeetu Patel (1:03)

Thank you for having me, Sophia, and it's great to be here.

Sophia (1:06)

Excited to have you. I would love to learn a little bit about what you guys are working on with Anthropic and OpenAI. Can you tell me a little bit more about that project?

Jeetu Patel (1:14)

Sure. It's actually a multitude of projects, but we've been working with the Frontier Labs for a while now where we are also investors in Anthropic, but with OpenAI, Codex was, we were the first design partner for Codex. We have 32,000 engineers. They all use Codex. We will have about half a dozen products that will be fully written with AI by the end of the year.

Seventy percent of our portfolio will hopefully be fully written with AI by the end of next year. When you think about $63 billion of revenue at that scale, having AI so broadly proliferated, it wouldn't have happened without the partnerships we've had with these companies and a lot of debt of gratitude from us to them.

Sophia (1:57)

That's pretty exciting. I know you guys are working directly with Anthropic on the project class wig, and you're working with OpenAI on daybreak, right? And so what my understanding is, is that you actually get to put this frontier AI against your own software and you're using it to find vulnerabilities before attackers actually do that. What's been happening lately in AI cybersecurity that made Cisco decide this is super necessary now?

Jeetu Patel (2:22)

Yeah, I mean, look, the thing is, is we are, the way you should think about us is we are critical infrastructure for the critical infrastructure. So what happens if you have a cyber attack with a hospital or a power plant or, you know, a water supply, a water company is that lives can get lost.

And if your hospital goes down, before you know it, you can't get dialysis. If your power goes out in three, four days, you know, a city runs out of food supply. And so there's meaningful dire consequences that can happen as a ripple effect of a cyber attack. And these cyber attacks now are happening at machine scale. And when Methos first came out Anthropic, you got to give them a lot of credit. What they did was they said, let's not roll this out to everyone. Let's make sure that we take it out to just a select few people so that we can really learn because it knew how to go out and identify vulnerabilities at a much, much faster pace than what we've ever seen before. And so what we did was we hardened our infrastructure as a result of being part of Project Glasswing, which by the way, is not just good for Cisco. It's good for the entire world because a substantial percentage of traffic in the world goes through Cisco infrastructure. So it's important that we don't have that get compromised.

And so we've done that with not just with Glasswing with Anthropic, but we've also done that with Project Daybreak with OpenAI. And I think both of them are great partners that are actually, sometimes they might get a bad rep, but they're really concerned about security and the people that we talk to. Yes, we talked to Sam and Dario and all those folks, but the people we talked to within those labs, they really care about making sure that the world is safe and secure.

Sophia (4:12)

Yeah, that's really interesting to understand from the inside, because both Anthropic and OpenAI, of course, as you alluded to, deeply, deeply care about security. One thing I want to get from your perspective though, because you do get to work with both of these companies and you get to work with them on these different projects, do their systems have different types of approaches to finding these vulnerabilities and problems?