AI PCs: Key Considerations for Procurement

Patrick Moorhead: The Six Five is On the Road at Intel’s corporate headquarters here in Silicon Valley. We are talking about everybody’s favorite topic and that is AI. Daniel, it’s great to see you. I’ve missed you. Great to see you, we’re talking AI again and it’s just amazing the depth of the conversations we can have. AI in the data center, AI on PCs, everything in between, real life benefits to enterprises, consumers. It’s great.

Daniel Newman: You took the words right out of my mouth, Pat. It’s been a big year and of course you and I love to talk about the 18 to 24 months that has changed the world despite the fact that there’s been four decades of AI algorithms and companies have been spending more than the last decade really doing AI. But I think generative AI has created this inflection point. It’s been really exciting to start to see the hype and the enthusiasm and the excitement reach us through our applications, through our tools, our devices. And I think that’s what we might talk a little bit about here today.

Patrick Moorhead: That’s right. It’s not just about data center AI, it’s also about AI PCs. It was really fun to get on Intel stage at CES. It had all the biggest PC makers on there. It was like the big unveiling of the first AI PC and it was fun. And at some point though, the entire ecosystem has to click in, endpoint management. All the enterprise ISVs and even things like procurement, right? How do people buy? What should they be looking for? And I can’t imagine two of the best people to talk about this. Carla, Mike, great to see you.

Carla Rodriguez: Thanks for having us.

Michael Monahan: Thanks for having us.

Patrick Moorhead: Absolutely.

Daniel Newman: Yeah, it is actually great, Pat, and you did actually point that out well. I shouldn’t say actually. It just was well pointed out that we are in this era though, where decisions are starting to be made. Companies are, they’re reflecting upon their fleets. They’re thinking about is it time to upgrade? Have we seen enough from the AI PC? Is this the moment? Everyone’s excited. We all love devices. We’re sort of a world, we’re waiting in lines around the corner to get that new thing. But Mike, I’d love to just get your reflections on the AI PC as a CTO, as a company that’s helping companies make these decisions. Where are you at with the AI PC? What’s got you excited?

Michael Monahan: Well, all the years, like you said, about 24 months ago, we saw Chat GPT come to the world and AI has been around for a long time, but generative AI is definitely top of mind. Customers are rushing to it. You look, 90 plus percent of our customers say they have an AI objective or their boards are putting an AI objective in front of them and they’re trying to figure out how do they take that to market. So the AI PC generation, 80% of that data is being generated at the edge on devices and it has to be processed because data gravity matters. You can’t wait for it to transmit back to a data center, try to process that data. So what we’re seeing is our customers are looking at ways to process the data on their devices. So having these NPU type devices, being able to shift to a device or a silicone component obviously allows them to be more efficient.

So I think we’ll start seeing measurements around ROIs. The good thing is Compucom is, as a managed service provider, we’re looking at that data today. We have processes for our managed platforms that we can tell them what are you currently doing? Things like boot times, how long does the application work, cycle refreshes. And then I think the other big part of this is the inflection point of a Windows upgrade, right? It’s a perfect time to marry AI PCs with a Windows 11 upgrade. So I think you’ll see that driving a lot of this over the next few years.

Patrick Moorhead: Yeah, I love the one two punch Windows 11, AI PCs. It just makes sense. And the way that Enterprise wants to buy too, they want to buy it. They’d love to get everything done at the same time. I mean sure, especially if you’re looking at an investment that they want to optimize for four to five years. So Carla, we’ve had some great conversations on The Six Five about software, and I think the last conversation we had to be specific was on how you’re enabling software. What I’d like to do, I’d like to narrow this question on what are some of the benefits in software that relate to AI PCs and business operations?

Carla Rodriguez: Great question. And yes, it’s great to see you in person. I know we’ve done a lot of this virtually. Yeah, I think the software ecosystem is just exploding right now with all things AI PC. Mike, you mentioned the ability to do a lot more locally. The software’s got to be able to leverage that local compute in areas such as security applications, which are very top of mind. And being able to do scans locally is one of the things that we’re hearing a lot from our customers and we’re engaging with the security software ecosystem.

Another thing being manageability, being able to reach that device anywhere in the world, which is actually something we’ve done at Intel for decades, and it keeps coming up more and more. And so the ability to choose where to run the workload so that the user and the enterprise and the decision makers managing the enterprise have the best, most secure experience is what’s really coming to bear in software these days.

Patrick Moorhead: I love it.

Daniel Newman: Yeah, I love that you said that. Pat and I talk about it all the time, like this isn’t new.

Carla Rodriguez: Exactly.

Daniel Newman: You hear me talk about AI. I do love anything that creates buzz. We sort of live and die for the next big tech moment. And of course they come fast and furious and AI has accelerated that, but some of these things are new, some of these things aren’t new, and some of these maybe just inflection Carla to your point, it becomes the opportunity. Things that should have been taken advantage of by these enterprises for a long time starts to be taken advantage of.

Mike, I want to talk about the criteria, the decision making a bit more. You’re evaluating different partners. You mentioned different silicon architectures, different designs. When you go to your commercial customers, if you go to them and say, here’s 700 designs from three, four different silicon providers with different GPUs, MPUs, CPUs, you’re just going to confuse them. And they’re all technical and they’re capable, but they’re looking for you to help. How are you thinking about that when you’re taking all these different opportunities to them?

Michael Monahan: Well, I mean cost is obviously a point. The article I read the other day, there was an 800-hour inflection point is kind of the cost model that a lot of companies are looking at. So that’s kind of where they’re starting in that range. So cost matters. To Carla’s point, security. How do we bring these security postures to the customer? I read the other day, most customers have 12, 15, 20 different security products that they use. And the problem becomes is how do you integrate all this? So what you’re going to see is a shift in the mentality S and B enterprises. There’s probably going to be three or four companies that rise in both and getting away from security point products into a platform. So security platforms are going to kind of drive the next generation. So how do we push security platforms down to the PC is obviously going to be important to them.

And then obviously when we start looking at predictive analytics, machine learning, what is involved with those systems? So internally, obviously we do a lot of that. We’re evaluating our customers’ data that they’re feeding into us today, and we can tell them based on your systems running slow because you need to upgrade to a new system that has an NPU, or here’s recommendations around just startup boot times are really slow. So I mean, I think in an AI world, data is the king. So the more data we have, the easier we can create reflection points against that to say, here’s why you want to look at these type PCs. Now, like you said, there’s a lot of vendors. Everyone’s using Intel as their key go-to manufacturer today. I think that’s going to be important for us to be able to differentiate that for our customers.

Patrick Moorhead: So Mike, the way that IT typically sets up a wave of a new fleet is they come up with a kind of checklist. And you’re providing value too, which is, “Hey, here’s what we think it should be, here are their requirements”, because you’re adding value along the way. Invariably, particularly the higher the level you get is the question about return on investment, right? That’s for everything that gets spent, any type of major purchase. And I’m curious, ROI specifically for AI PCs, what are your early thoughts on payback? You already talked about maybe the expected price point. Maybe it’s 800. I’m seeing right now $1,000 is that added expense for AI PCs, but what are your current conversations like on ROI?

Michael Monahan: So we’re looking at it from a workforce perspective. And you’re right, if we’re starting to integrate things like copilots and other applications, now there’s an additional cost to pay for software. So how do you show an ROI if I push out a copilot PC with AI and integrate it, and how do I justify that additional 20-25 hours ahead? You got to look at what that productivity returns. So we collect all those data points. Today we can say, Hey, based on… And AI PCs obviously are newer to the market, so we don’t have a lot of data points yet from our customer base. I think over time that’s going to grow. But we’re looking at things like what is your productivity increase or decrease depending on how you look at it. For our customers, like today, we could say we see a 10% productivity decrease because you’re doing this, where if we integrate or upgrade your fleet, you could increase your productivity.

So we’re measuring those types of statistics to be able to say, if it takes you 10 minutes to write an email and AI can do it in two, there’s productivity we return back to you where you can go focus on something else. I think the big shift has also become, it’s less about cost. When I was in IT, it was always about how do I reduce cost. Today it’s about how do I generate revenue. So I think the measurement there is, okay, I’m not paying for you to do something, but you’re generating because now you’re more efficient, you have more time to go focus on making the company money. That’s where I think the return on investment becomes, because now instead of taking six hours a day doing this, you take two and I can go focus on new task.

Patrick Moorhead: And I’m really interested. I mean the software, the on-device AI software is moving so quickly. I’m hearing, and we talked to enterprises too, and what they’re looking at as well from the cloud-based AI services is they actually see potential cost reduction possibilities by keeping it all on their systems and being able to do some very important types of applications. There’s files on the PC and the ability to query those in an intelligent manner without actually having to go up to the cloud. And what I’m seeing is customers who do on device and Cloud, finding a more of a hybrid type of opportunity where they do as much as they can on the PC and then they farm out the rest to the Cloud. So it’s a really constantly moving target across all the ISVs. There’s going to be a little bit of push and pull because all this investment into the NPU on this device actually saves the Cloud providers money because they can do more of it on here.

Daniel Newman: It moves the economics around. And so companies of course are having that push and pull, and have for some time about where does a workload belong. Where should that data, does that data need to all be processed and then does it ultimately need to be sent back to the Cloud? And of course, one of the things we have to all accept, the exponential growth of data isn’t going to slow. It’s only going to get bigger. And these apps get bigger. The amount of data gets bigger, what we’re going to want to process, what we’re going to want to do. And Carla, that takes me very nicely-

Michael Monahan: Thank you, Pat.

Daniel Newman: To kind of the last question. Can you imagine me hitting a golf ball? It’s terrible, terrible to see. But I wrote a book called Future Proof. I want to think about future-proof and how to future proof your investment. We talked about ROI. We talked about the investment people buying AIP PCs, it’s about figuring out the right hardware. It’s about identifying those right software applications that both of you talked about. How are you thinking about and how are you enabling your partners customers to future proof with the AI PC?

Carla Rodriguez: Yeah, great question. I was going to refer to a comment you just made. A lot of the conversations we have with our ISV partners are precisely early on where things need to run across CPU, GPU, NPU, plus the hybrid component on Cloud. That is probably the number one decision that we face with our ISV partners a year or two out from product launches and their application launches as well. Because that dictates a lot of the benefits that we see downstream. And this is moving very fast. And it is beautiful to see the ROI in terms of not just what you’re going to reduce, but the other way around. How are you freeing up your employee base to do more creative strategic things that are going to bring in that extra source of revenue?

And so that’s a big part of how the ROI is playing out currently with a lot of our ISV partners. But back to the question you were posing Daniel. Yeah. We’re seeing a lot of the future proofing in how we’re testing, how many applications can be loaded up on NPU, how many applications can be loaded up on GPU. And that is sending a lot of measured early data back to managed service providers, large enterprises to say I really need to step up. Because we all know with data exploding, with more apps utilizing MPU, GPU and CPU, you don’t want to be left behind. The users are also going to start demanding this. They’re already seeing the benefits of AI helping them with mundane tasks and the fear of this is going to take parts of my job away, is starting to be replaced with, oh my gosh, I can do so much more now.

Michael Monahan: I think the statistics we’ve seen is probably the number one issue is password reset. So if you can have an intelligent gen… Well think about it.

Daniel Newman: It’s the reality.

Michael Monahan: I know it’s not easy, but it’s like that’s the reality. Yeah. That’s different challenges. And inside Compucom we build a gen AI chat bot, which we can do stuff like that. We can pull in service tickets, we can do password resets, we can ask how’s my PC running today and make recommendations. So they’re the type things where if, to Carla’s point, if we can make our workforce more efficient, take away those mundane tasks. I saw a study one day, like two hours a day spent administratively. If we can eliminate some of that and make them more productive, that’s huge.

Patrick Moorhead: I’m in.

Daniel Newman: Absolutely.

Michael Monahan: And Dan, to your point, I mean part of this is siloed data, living at the edge, living in data centers, living in data lakes, Cloud, right? Public private Clouds. How do we create those fabrics, data fabrics to be able to expand across all that to process the data where it lives instead of pulling all that data together, because data is the lifeblood. Who wants to give their data away to a Cloud provider or data lake or data house? So I think I see a lot is they’re asking, how do I leave my data process where it is, utilize the resources, but then pull all that together to make intelligence from it? And that’s where we’re looking at things like data fabrics to be able to suck all that data ML pipelines into it and then make intelligence from it.

Daniel Newman: Yeah. Mike, I think you’ve done a nice job of previewing another show that we need to do down the line to talk about how this moves from device to edge to cloud to really enable enterprises to maximize the value of their data. But we have to go. So Carla, Mike, I want to thank you both so much for taking the time here. Very interesting conversation. Let’s keep on this topic.

Michael Monahan: Awesome. Thank you so much.

Carla Rodriguez: That’s great. Thank you.

Patrick Moorhead: And thank you for joining us for this Six Five On the Road. We’re here at Intel headquarters. Another great conversation about Intel AI PCs, and what these two companies Compucom and Intel are doing. Stay with us for more, but we got to go. We’ll see you all later.