On this episode of The Six Five – On The Road, sponsored by Intel, hosts Daniel Newman and Patrick Moorhead welcome Intel’s Anil Rao, VP, GM Systems Architecture & Engineering and Zscaler’s Ken Urquhart, Global Vice President of 5G Strategy for a conversation on how Intel is making aggressive plans to offer software, security, and services to its customers with initiatives like Project Amber, Confidential Computing and more.
Their discussion covers:
- Intel’s commitment to empowering developers with confidence, leveraging innovative hardware, developer software, and solutions
- Anil’s pivotal role in driving security technology and system-architecture innovation, including oversight of the Intel Trust Authority
- A spotlight on Intel’s client, Zscaler, and their successful implementation of Intel’s Confidential Computing, along with the achieved outcomes
- Exploring the concept of being “secured down to the silicon” and the consequential assurance of end-to-end data protection, instilling confidence in Intel’s clientele
Be sure to subscribe to The Six Five Webcast, so you never miss an episode.
Watch the video here:
Or Listen to the full audio here:
Disclaimer: The Six Five webcast is for information and entertainment purposes only. Over the course of this webcast, we may talk about companies that are publicly traded, and we may even reference that fact and their equity share price, but please do not take anything that we say as a recommendation about what you should do with your investment dollars. We are not investment advisors, and we ask that you do not treat us as such.
Patrick Moorhead: Hey, there. The Six Five is here at Intel Innovation 2023 in San Jose, and we’re having some super conversations. We’re talking all the way from the hyperscaler data center to the client computer and everything in between. We’re talking AI compute and we are talking security because quite frankly, if it’s not secure, nothing else really matters.
Daniel Newman: Yeah, once you realize it’s not secure, the problem’s probably…
Patrick Moorhead: Probably too late.
Daniel Newman: Way too late, but yeah, we’ve gone from the sileconomy, I just want to say that a few times the sileconomy…
Patrick Moorhead: Sileconomy.
Daniel Newman: To the AI PC. But whether it’s the silicon layer all the way to the PC, it does have to be safe, it does have to be secure. And that’s been a pretty important topic here I’ve noticed, I call it one of those horizontal topics that’s overlaying every different part, whether it’s data center, whether it’s edge, whether it’s PC, mobile device, whatever it is, has to be secure. And when you’re with developers, of course they want to build the killer apps, but it’s on their mind too.
Patrick Moorhead: Well, as we’ve always talked about too, securing things, it takes a village. One company can’t do it, and it takes partnership and it’s a multilayer approach. And while there’s buzzwords all over the place, probably one of the biggest things to hit the security conversation is confidential computing. And it’s with that, we can actually talk to our experts here, Anil and Ken. Good to see you guys. Thanks for coming on the Six Five. First time Six Five, and we appreciate it.
Anil Rao: Thank you for having us. Looking forward to the conversation.
Daniel Newman: Anil, let’s start with you. And before we even get too into the security side of things, let’s just get into the software side of things. This is a developer conference. I think sometimes, Pat, you and I do a lot of media. It’s almost like we’re teaching people from the beginning that, oh, it’s not just a chip company. They’re doing a lot more and software is part of it.
Patrick Moorhead: Right. Exactly.
Daniel Newman: And so, Anil, talk about that when people are like, “No, Intel’s a hardware company,” and you’re doing so much in software. Give us a little bit of the flavor of what Intel’s doing in the software realm.
Anil Rao: Yeah, it’s a great question, Dan. Intel is actually very well known for Intel Inside. This has been there for almost 20 plus years now. One of the things that we do in the company is not just silicon or platforms, but we invest a lot in terms of software as well.
In reality, we should be known as a software company too because a lot of work that we do in bios, firmware, and even going up the stack in terms of enabling operating systems and designing things in hypervisors, there is a lot of work that happens inside of Intel.
We take pride in saying that we develop not just silicon and platforms, but software in order to enable the industry to adopt some of the silicon solutions that we build. In this regard, we are also one of the unique companies in the industry, which has a full suite of portfolio.
You saw Pat talk about it. And in his conversation today, he spoke about client solutions, spoke about edge, spoke about cloud from a solutions as well. In order to make all of these things fly, there is a tremendous amount of software that goes into the picture and that’s why Intel is one of the companies that builds end to end from a solution portfolio perspective.
But I think one of the areas that we are less known for and we are working hard in order to make ourselves known, and tomorrow you’ll hear some more of this, is in the area of building software that customers can trust. Trust and security is an important element of what one needs to do in order to have a great, not just a developer experience, but also a worry-free developer experience.
As you design and develop your software, you shouldn’t be worrying about, “Oh, is my IP safe? Is my data safe? Is my IP secure? Is it adhering to all the privacy and confidentiality rules?” These are some of the things that we do. There is a tremendous amount of software and there is a tremendous amount of services and solutions that you’ll see from Intel.
Patrick Moorhead: Yeah. It’s interesting the new security threats that come along with generative AI, if nothing else, it’s one of the first times that different types of data all the way from ERP to CRM to legal data and everything throughout the entire enterprise to get the best result from generative AI needs to be activated on.
Now you don’t have to do it. It can be front and center, but that plus the threat of using some sort of rogue LLM just opens up a new opportunity for both Intel and Zscaler. Anil, in your bio, I mean you’ve got a really impressive side. The things that you do, I mean you have to come up with the right architecture and you and your team come up with the right safeguards throughout the company. And this has a lot of implications to your customers, especially in this new world of generative AI. Can you talk a little bit about your role and how you safeguard, how you make decisions and how you approach this?
Anil Rao: It’s a fantastic question, Pat. If you look at the focus that I put on and my team puts on at Intel is going down to the root of the underlying hardware. And Pat spoke about it today as well, where security starts with Intel. At the core of it, you have to be able to trust the underlying hardware and rely on the underlying hardware and not just trust and rely, but make sure that the hardware provides certain level hooks in order for you to run your software and your operational environment.
In this regard, I’m extremely excited with a new technology that we are bringing to the market and that technology is confidential computing. And most of us are familiar today about how data is always encrypted and secure when it is stored on devices or data is secure when it’s in transit, but we don’t protect data in use.
Very few amount of computation these days is done in solar machines. It’s always done in shared, machine shared infrastructure. Especially when you go into the public cloud, you typically go and rent out virtual machines and when you rent out virtual machines, you are running in a shared infrastructure. There are a lot of people who are operating on that machines who possibly have access to data, and you don’t want that data to be in the clear.
You also want to make sure that whatever data is sensitive to your operations, because we are all taking advantage of the cloud, whether it’s public or private cloud, you operate in a way that you know who has access and if someone is authorized to have access. And if anyone is not authorized to have access to it, then don’t provide access.
The way it is done is through confidential computing. We create an enclave typically called the trusted execution environment. And any application which is running inside of the enclave is trusted. And how do you know it’s trusted? You trust it by going through an attestation mechanism, and this is where we have the Intel trust authority, attestation service. The Intel trust authority attestation service is giving you proof. It is proof that can be audited and that proof is what you use in order to make sure that the entire infrastructure is secure.
Now, the beauty of going back to AI that you’re talking about, you can do this for AI models, AI data. And you clearly have a fantastic understanding of what exactly is the data, the model and the software that went into either the training process or inference process. It makes the entire use of AI a lot more interesting and a lot more compelling because now you can rely on what’s happening in the underlying infrastructure.
Daniel Newman: It’s like the root of trust, but it provides a very significant way to watermark every action that’s taken place.
Patrick Moorhead: A root of trust per application.
Daniel Newman: Right. Yeah.
Anil Rao: Yeah. In fact, one of our customers says that it’s almost like you’re creating a private cloud in a public cloud infrastructure or creating a secure cloud in a private cloud infrastructure.
Daniel Newman: We spent a lot of time researching confidential computing. And it’s interesting because obviously there’s the umbrella confidential computing, and then of course there’s Intel and others that have created versions based on certain set of standards. I want to get Ken involved here. Ken, thanks for being patient.
Patrick Moorhead: Ken’s being very patient here.
Ken Urquhart: I’m enjoying the conversation.
Daniel Newman: You’re in. It’s about to happen here right now.
Ken Urquhart: I’m ready.
Daniel Newman: You’re a partner, user of Intel’s confidential computing at Zscaler. Talk a little bit about why you went down that route and how you’re using the technology.
Ken Urquhart: Well, we’re customer driven. Customers want end-to-end cybersecurity. We do zero trust. Now, if I ask you what does zero trust mean to you? Is it a marketing phrase? Does it mean something in particular to you?
Patrick Moorhead: I would say yes, yes and no. There is reality behind it. There’s an architecture behind it. Yes, but there’s architecture in everything.
Ken Urquhart: Yeah, old products cast in new clothes. Hey, we’re with it. We’re happening. Fine. We’ve been around since the dawn of what is now accepted as zero trust. Around the time John Kindervag put his stake in the sand and say, “We need a better way to do cybersecurity.”
Personally, I go way, way back. I’m actually one of the people probably who can claim I installed one of the world’s first firewalls at a university at a time when cybersecurity breaches meant that, yeah, we don’t see a need to put a password on the root account.
Patrick Moorhead: Yeah. Because it’s all perimeter, you’re going to stop it there.
Ken Urquhart: Well, I have a wall. I’m me and there’s them, and we connected this thing called ARPANET up. And people went, “Oh, what do you mean someone else can get into our machines?” It’s like, yeah, that’s why you need to put passwords on them. That was how easy it was to breach. I just say admin, password, you hit return. Oh, you’re in. Yeah.
Daniel Newman: We’ve all done that at least once to somebody, right?
Ken Urquhart: There was a time when that was the way. You put the firewall in. Great. Okay, I got to communicate with other people. Fine. What do you do? Punch holes in it. Ah, okay. I’ve got this wall, I’ve got this water on the other side, and I start punching holes in it and wonder why my feet are wet.
Okay, fine. Zero trust said, “Okay, look.” How should I put it? I think it’s like what I call the beehive model. Wasps attack the beehive at the entrance. They get swarmed, they get killed, as much as everyone will attack them. Wasp eats his way in through the back of the hive and gets in, they’re pretty much assumed you’re allowed to be there and they leave them alone and they wreak havoc in the hive. That’s the model we were living with because it was based on a physical metaphor, us, them, wall, protection because I had my data center. Cool. Now we live in a world where computing happens everywhere.
Patrick Moorhead: Oh, it’s fractalized. It is literally everywhere.
Ken Urquhart: The point is anywhere. You say, “Well, what’s zero trust?” At its heart, one sentence. I want to make sure you can operate securely when you have to work in environments you cannot verify are secure. When you travel over networks you can’t verify are not breached. That’s the heart of it.
Single sentence. How do you do that? Well, we do it with five things. There’s identity. Let’s go back to what does everybody want to do in this industry? You take a user, using a device, connecting over a network to a workload to exchange data securely. Five things.
Patrick Moorhead: You make it sound so simple.
Ken Urquhart: It is. You’re a user. Zero trust means I don’t trust anything. Who are you? You say, “Hi, I’m user X.” You go, “Who are you?” “I’m user X.” “Cool. What’s your password?” “That.” “Fine. Here’s some biometrics authentication. Do you pass that? Here’s some MFA. Enterprise gets to choose.” And you got to go through all that to say, “Okay, I trust you to my idea of what trusting you is.”
Okay, I’m at a device. Fine. Do I own that? Do you own that? Is it BYOD? If I own it, is it running the version of operating system? Is it the hardware I want? To my satisfaction over a network? Any network today will get you to me to a workload. Is that workload what I said it is? It’s I’m talking to Oracle database. Cool.
Is it acting like a database? Are the only ports open what I would expect from a database? Is it only trading data using protocols that database should use? Wait, that’s the accounting database. Why is it accessing our GitHub? Well, maybe it’s not really an accounting database. Is it behaving normally?
And finally, the data has to be secured at rest and in motion. We take care of all that. As a developer, you don’t have to worry about it. It’s just there. In some ways, you can take an app that may not be fully secure and we can wrap it and make it secure because we don’t let anyone else see it. It becomes invisible just like the users become invisible. And you can’t attack what you can’t see. Cool. That works great.
I’ve got a way to do from cloud to user over any network. We have 40% of the Fortune 500, 12 of the 15 cabinet level posts, DOD, DOJ trust us with their cybersecurity. Developers, do you have to do anything different to integrate with us? For the most part, no.
Patrick Moorhead: You’re answering all my follow-up questions here, man.
Ken Urquhart: They’re just trying to get this baseline.
Daniel Newman: He’s actually asking the questions and answering them.
Patrick Moorhead: I know.
Ken Urquhart: Now you say, “Okay, what’s the deal with Intel?” First of all, we’re an all intel shop. Why? And I mean we have a global communications network. We are the world’s largest hyperscaler that does not host, does not offer apps, does not offer storage. We offer throughput and communication, 300 billion transactions per day. A trillion data points put through AI, ML to keep our customers secure. Great.
Why are we an all Intel shop? Because of the technology they have, the software, they have the boundaries they push for our customers because our customers demand throughput. They demand efficiency. They demand five nine’s, uptime. And I can often say to people is that we’re usually testing and using your new innovations before our customers get it because we’re always being pushed more, more, more. More volume, more throughput, more efficiency, more security.
Which brings us to this little problem that you’ve already said. Great, network secure, apps secure. There’s that final nanometers of the data in your app goes down through hypervisor to the silicon to actually do the compute and comes back up. If that hypervisor is breached, someone’s watching, someone can modify, someone can block.
And that’s where Intel’s confidential computing comes in, in that final layer. We go right now, we could do cloud to device to user. Now we go down to the silicon. Now it’s true end-to-end. I’ve been saying I protect you. Across the networks, across places you can’t verify are secure. And Intel stops that final layer where you can breach on the hypervisor, which most people don’t think about. That’s now secure, so now it’s cloud to silicon closes that path. And for most developers that don’t even need to understand that’s happening, you just know it’s secure. It’s the easy button for end-to-end. Now I’ve said a lot of stuff.
Patrick Moorhead: Yeah, you have, and by the way, the good news, well, as we’ve seen over 30 years is that hackers go for the soft spot. They go for the ROI. And with hacking as a service out there and nation state budgets, it’s all getting more sophisticated.
Guess my follow-up question for you would be, in this world of generative AI where I can spoof something a lot better than I could ever spoof before, I can come in, I can look like a transaction that you would expect an accounting system, but I’m not.
I can come in and I can look like a certain type of device, maybe in a certain country behaving like a real human, not like some script that was written five years ago. How do you combat that with this new technology?
Ken Urquhart: Well, for one thing that’s been happening already, we see it every day. Why do you think I say we’re parsing 1 trillion metadata points each day looking for this behavior.
Patrick Moorhead: You are seeing generative AI based intelligent threats coming in and hitting you already.
Ken Urquhart: Even before that. We were having… It’s like you want to attack a neural network, use a neural network. Go home and take a break, come back. Hey, there’s your exploit. We see this all the time. There’s even things like you can attack with an AI, that’s been going on for a while.
You can even do things like if I breach an organization, I can poison their training data. I can insert back doors into neural networks that are activated when I say certain things. You talk about generative AI. How about having a generative AI where I poison the data such that when I ask a certain question, it’ll spit out the answer I want, not the right answer?
This is happening now. In telco. You can look like noise coming from our RF signal to a cell phone tower to convince the tower it’s overloaded when you’re the only device on it. These are real things happening now with AI. Your data needs to be extremely clean and you need to watch out for rogue attacks that can make your AI do something completely different.
And data quality now takes on a huge new importance to organizations. You need confidential computing, you need zero trust networking. The old ways of doing are not enough. That’s been breached.
Patrick Moorhead: Very powerful duo here.
Daniel Newman: I was going to say. We spend another 10 or 15 minutes with Ken and the world’s going to start to feel like it’s closing in.
Ken Urquhart: No, no, seriously.
Daniel Newman: No, I’m having fun.
Ken Urquhart: It’s a world where we’re here to innovate, Intel Innovation 2023. I’m telling developers there is an easy button for this.
Daniel Newman: That was almost where I was getting, I was like, if you keep telling the stories of every… It’s because a lot of times we’re in this euphoric state of all the cool things we’re doing. And it was like those people that were entering all their proprietary data into ChatGPT, weren’t even paying attention.
I mean, we’re our own worst enemy a lot of times when it comes to security. But I think what point you made that was really, really compelling here, Ken, is when you get past the point of the human error and you do everything else right, there are these very small one and two degrees that make the difference between truly securing the entire environment and missing by this much, which can open up to everything being breached. And I think that’s a great testament for what Intel’s doing with confidential computing.
Ken Urquhart: Oh, absolutely. We’re right there with them.
Daniel Newman: Yep. And I would like to say, because I’d love to talk more because I want to hear some more of this what’s on fire, we’re going to talk afterwards here.
Ken Urquhart: We’ll do that in the green room afterwards.
Daniel Newman: But Ken and Anil, I just want to thank you so much for joining us here on the Six Five at Innovation 2023.
Ken Urquhart: Thank you for having us.
Anil Rao: Appreciate having us.
Daniel Newman: It was fun. That was fun.
Ken Urquhart: Absolutely.
Anil Rao: Thank you much.
Daniel Newman: All right, everybody, hit that subscribe button. Join us here for all of our Intel Innovation 2023 segments on the Six Five. Patrick, it’s time to say goodbye.
Patrick Moorhead: Bye-bye.
Daniel Newman: See y’all later.