From Personal Computing to Personal Intelligence: HP on Reimagining the Role of the PC

AI agents are no longer a future workload for the PC. With 72% of organizations already deploying or piloting them, they are beginning to redefine what the device needs to do.

At the Six Five Summit: AI Unleashed 2026, Ketan Patel, President of Personal Systems at HP, joined Patrick Moorhead and Daniel Newman to open the AI Devices track. Patel made HP’s case that the PC is evolving from a tool people operate into a personal intelligence device that works alongside them.

That vision depends on hybrid AI. The cloud remains essential for training and deep reasoning, but everyday inferencing is shifting toward a distributed model spanning cloud, on-premises infrastructure, and the device itself.

The economics are accelerating that shift. One HP customer saw its annual AI model spending climb from $20 million to between $200 million and $300 million. HP estimates that moving basic queries to local models can save approximately $650,000 per 1,000 employees—savings that could grow as on-device models advance from roughly 20 billion parameters today to as many as 120 billion on next-generation architectures.

HP’s strategy spans three layers: context-aware devices, a connected ecosystem of PCs, printers, and peripherals, and an IT control plane designed for agent-scale governance. That last layer becomes critical when teams accustomed to managing hundreds of thousands of devices must also oversee millions of AI agents.

‍

Key Insights:

🔹 AI adoption is redefining the PC: With 72% of organizations deploying or piloting agents, intelligence is moving closer to the user and the device.

🔹 Hybrid AI is becoming the operating model: Cloud will handle the heaviest training and reasoning workloads, while more enterprise inferencing runs on-premises or directly on the PC.

🔹 Token economics are now a boardroom concern: AI spending can scale from tens of millions to hundreds of millions of dollars in a single year.

🔹 Local models could materially lower costs: HP estimates approximately $650,000 in savings per 1,000 employees when routine queries are processed on-device.

🔹 Governance must scale from devices to agents: IT organizations will need to manage millions of agents alongside their existing hardware fleets.

🔹 The market could be enormous: HP estimates that agent-based AI could represent a nearly $6 trillion total addressable market by 2028.

Patel’s bet is that the PC will become the platform where personalization, governance, and AI cost control converge.

Watch the full video at sixfivemedia.com, and subscribe to our YouTube channel so you never miss an episode.

Catch the rest of the AI Devices track and full Six Five Summit: AI Unleashed 2026 coverage at sixfivemedia.com/summit.

Disclaimer: Six Five Media is for information and entertainment purposes only. Over the course of this video, we may discuss companies that are publicly traded, and we may reference their equity share prices. Nothing discussed during this webcast should be considered investment advice or a recommendation to buy or sell any security. We are not investment advisors, and you should not rely on this content as financial advice. Six Five Media collaborates with technology companies and industry leaders to produce research-driven interviews and multimedia programming for enterprise technology audiences.

‍

Ketan Patel:
The PC of the future will have more context awareness, the surroundings across audio, video, the sensing capabilities which can have. It's the only device which actually will have both personal and professional context in the most optimized way.

Patrick Moorhead: Welcome to the Six Five Summit 2026. The theme of this year is AI Unleashed. Daniel, it is amazing over the last three years, all the twists and turns, you know, I guess three years in regular is like 30 years in AI time.

Daniel Newman: 

Yeah, it's seven years of Six Five Summit and how AI became just the epicenter for three years. And yet we're still unleashing it. And it's funny because despite the fact that we're so far into this process, it still feels very early to me. I don't know.

Patrick Moorhead: 

For sure. One of the biggest trends that we're seeing is, you know, the conversation used to be with AI was all about data center. Right. But now it's very clear. You're looking for sovereignty. You're looking for token efficiency. You're looking for spreading all the love out in the enterprise. And quite frankly, with consumers, that is going to be the client device and devices is going to be as important. And we are here with the AI Devices Track Opener. And I would like to introduce Ketan Patel, President of Personal Systems at HP. Great to see you, Ketan. It's been a pleasure to get knowing you outside of the video and our meetings. Anyways, welcome. Thank you, Pat.

Ketan Patel: 

And congratulations for the Six Flags Summit. I appreciate that.

Daniel Newman: 

Yeah, it's just getting every year and it's moved so fast and you're in the middle of it. And that's it's a really exciting time. I mean, your job here is really talking about this shift that's going on. You know, there's been some stops, some starts, some acceleration. I think we're all realizing that we've entered this era, this time, where we're going to want to bring some workloads onto the device. And there's an economic story there. There's a convenience story there. A lot of us have been doing it for a long time, just kind of subconsciously, just on our phones, like, you know, on our PCs. But, like, you've been talking about this. for some time, the personal intelligence device. I'd just like to kind of get your broader thoughts on this. What's going on here and what's shifting?

Ketan Patel: 

What's driving the big shift right now? This is an amazing time in the world of personal computing. PCs, as all of us know for the last two decades, have been a tool you operate. Fundamentally, click, type, give some commands, and it executes. That's how it has been, and it has really served the purpose very well across enterprise, small businesses, and consumers. But the world which we are now seeing is completely different in the world of personal computing, moving to personal intelligence. I say this because there are three fundamental shifts which are driving this. Number one, as of today, almost 72% of the organizations are almost deploying or piloting agents. Now, clearly, as agents start getting deployed, you will see intelligence increasing on the devices more than ever because agents will be autonomous. It will really help user try to have work, which is like a companion, for themselves and agents work autonomously for them rather than user giving commands to the machine. So a lot of intelligence will be built just because of agents coming into the environment. So that's trend number one. Number two, In the world of agents, humans and agents still will have to work together because eventually it is about how the agents are actually driving personalized experience for individual because you and me will have different needs, different context, different requirements, and how they work with each other becomes super critical. And most important is about governance and security associated because how I want to operate my agents with a certain governance is different than probably yours. So that's where PC is the right platform where you have ability to interact through right level of governance and security. And third is PCs are becoming more and more powerful. Now you see the latest silicon architecture and the way we look at system design, the data center is coming to the PCs. As you see, some of this new generation architecture can actually even host 120 billion parameters models locally on the device and really take care of a lot of this token efficiency topic, which we can talk more. But this three fundamental shift is also now giving software players a lot of investments they are prioritizing of developing capabilities directly on the PCs, which was just 2% in 2024. So there's a huge shift happening. So this world of personal intelligence is exciting as these three vectors are coming through, and we are at the center of it and feel very good about how we want to take it forward.

Patrick Moorhead:

Yeah, so, Kaethan, you've talked about, you know, you've said the future is hybrid AI. And I think for the most part, you know, if you would have said that a year ago, people would have looked at you cross-eyed. But I think those, you know, kind of inside the industry and those who are talking to customers realize that this is true. Will you kind of break down break this down for us, and in the end, it's about unlocking value, right? It's less about the cool factor, although I like cool factor and whiz-bang technology, but how does it deliver better results?

Ketan Patel: 

Absolutely. I think that The way things started on AI is about all model training. And of course, it was all in the cloud, and it was a very, very important starting point. We believe that that trend will continue as the models will get continuously refined. But as you think about workloads coming into the real user environment across enterprises. Inferencing is now the important thing in terms of how you use the models and actually utilizing for your use cases, for your workflows, for different personas within an environment. As you think about how people started deploying initially, The tokens just maxed out as token maxing and token economics is a big topic nowadays. And I was with some customer roundtable just a few weeks back and we got inputs that, hey, A certain customer spent $20 million previous year as they started implementing some of this model's access to their employees. This year, the bills are up to $200 to $300 million and increasing by the day, and it is becoming a CFO topic. in terms of how to really manage this entire equation. Because on one side, you want employees to get the latest and greatest. On the other side, you want to see the economics of it and what's the ROI the company is getting. And this is about really from experimentation or pilots to maybe moving to profits is something which is going to be super important as customers look at the overall cost associated. And that's why essentially we believe the future of inferencing is going to be distributed or hybrid. Cloud will remain essential for large-scale model training, deep reasoning work, and certain personas who will have to go more to access that. But a lot of workloads which are in the enterprise by industry type or personas can be managed on-prem as you can actually now load a lot of open-source models on-prem as well as on-device. Just to give you an example, a laptop which we have today can have a 20 billion open-source model hosted onto it. Then as you go to on-prem, it can actually even host up to trillion parameters of models. How can now you as an enterprise deploy solutions across the enterprise which can access model on-device or on-prem or go on Cloud seamlessly, but the user gets the best experience, the organization gets the best economics, and you are able to effectively run in a right, secure, and governed environment. That's where the trend is going in terms of how we see this architecture coming through, just combination of economics. Then the second big area apart from economics is security. and governance because one of the other challenges a lot of customers are dealing with is how data flows across on-prem or cloud. And there is certain data one is okay to move and certain data they want to make sure that remains within the premises. And that's another trend which we believe is going to be super important as we think about distributed AI, where the data remains on-prem and then the signal goes to the cloud and then you are able to process the right query for a specific customer.

Patrick Moorhead: Data sovereignty conversation was discussed a lot about three years ago. And that kind of like, I don't know, it went to bed, it kind of hung out. And then even recently in the last six months, it's become this major conversation that's at the boardroom. Right? It's beyond the CIO when there's this potential conversation about intellectual property leaking into a model. And then it gets very, it's very strategic conversation at this point.

Ketan Patel: 

Oh yeah, and just to add on that, because a lot of customers don't want to, to your point on the sovereign data, their own IP and intelligence to become like a public information, and how do they otherwise differentiate while they definitely want to access the latest and greatest, but they want to keep their proprietary information to themselves?

Daniel Newman: 

That presents a big challenge. It also, you know, opens up opportunities for whether it's on device, whether it's open source, whether it's, you know, of course the frontiers are having a lot of fun and they're building really great technology and we're thankful for it because the open source wouldn't be coming and we wouldn't be building around it. But at the same time, you know, the economics, to your point, Don't necessarily make sense. You guys sponsor the Ferrari. I like the analogy that you don't need to drive the Ferrari F1 car to go pick up your coffee from Starbucks. There's a lot of workloads, and I think I saw a figure that you guys had come out with something like $650,000 per 1,000 employees. could be saved, and I would actually argue it might be larger, because that's the assumption kind of based on almost linear thinking of what we know they do now, agentically. But I know when we were talking, we said maybe it's one human employee for 10 agents, 20 agents, but that number's gonna be thousands and maybe eventually more. And so companies that are making decisions at the board level, CFO level, then at the technology level, can probably create an incredibly compelling TCO to have a hybrid and distributed strategy.

Ketan Patel: 

Oh, yes. And I think the example which we gave of 650,000 for 1,000 employees is just for very basic queries, which can be run by a very basic local model. But as models get more sophisticated for local hosting, including the example I gave up, Imagine a 100 billion parameter model sitting there. It can even start running other tasks that $650,000 per 1,000 employees can just move in terms of savings to a very different level. And hence, this entire TCO or economics across the distributed layer is going to be critical for business leaders and CFOs to look at what they get and at what cost.

Daniel Newman: 

Let's just pivot a little bit here, is every company across the tech stack right now is trying to tell the customer how they help. The model companies, the info companies, the device companies, the middleware, everyone, the applications, they all are kind of, here's how to apply AI, right? And one, it's both incremental, it's additive, hopefully it's productive and efficient to the business, but you need to figure out your own strategy. What is HP's to help Customers harness the power of AI, and I love that the word harness is in there because then it becomes a really dorky thing we can talk about, too. That's a nice nerd insert. Let's keep it at the high level. How are you guys helping?

Ketan Patel: 

Our approach is pretty simple. We want to think about three fundamental layers we want to help customers with. Number one, we have devices clearly across PCs, printers, rooms, workstations, peripherals, and many other products which HP has. So there is a device. But on device, we want to make sure how we drive our unique HP experience layer through which customers can access all of this on-device thing which I was talking about. That's layer one. Layer two is about workplace ecosystem, because along with the device, there's a workplace ecosystem which is around, and how do we make that seamless for the customers? And the third is about, for IT, how do you make it a good control plane? Because as you imagine, and IT was managing, let's say, 300,000 devices in the past. Tomorrow, they'll be managing 300,000 devices or even more, but also manage three million agents at the same time. How do you really provide that unified experience from an IT perspective also of ensuring some of these things are seamlessly working? Let me give you our approach across the three layers. From a device layer, we want to make sure that we are able to bring intelligent devices. And when I say intelligent devices, what do I mean? The current PC has a certain capability as we started the conversation on, but the PC of the future will have more context awareness, the surroundings across audio, video, the sensing capabilities which can have. It's the only device which actually will have both personal and professional context in the most optimized way. And we want to make sure that that context becomes a pretty critical pillar for any outcomes which the customers will get delivery of. So that's number one. Number two is about the ecosystem experience, which we want to make sure as PC gets connected to the entire workplace, it could be all products which we do like printers and peripherals, but any other device in the ecosystem, how does it seamlessly operate? Because it is not about an agent on the PC alone. An agent on a PC should operate seamlessly across the infrastructure, to have a seamless experience for the user. So that's another layer of ecosystem where we want to make it super smooth for the user to have a unified experience. And third, as I spoke about, the Workforce Experience Platform and the Wolf Security Stack, which we have, are two big properties we want to make sure IT is able to use to, as I said earlier, deploy all the devices, manage all the devices, drive the right level of security policies, because one of the biggest pains IT is going to deal with is about governance of these agents and driving the right level of policy. That layer is supercritical, and then the whole security stack which we have in order to help how IT will manage the entire infrastructure. So those are the three areas how we're thinking about. making sure that what role we play in the entire infrastructure. And eventually, our fundamental role also is going to give token economics TCO to our customers saying, hey, bringing these loads into the distributed or hybrid AI gives you this kind of an advantage.

Patrick Moorhead: 

So, great conversation. I want to go macro again and talk about the opportunities that agentic AI brings to the PC industry. And it's interesting. Historically, it comes in waves. right, and whether it was, you know, multimedia, DOS to Windows, the internet transition, you know, the sub-$1,000 PC, so many, you know, gaming drove, PC gaming, and, you know, when we were all working from home, right, that was another piece, but I'm curious, and let's put the memory things aside right now, what is the long-term opportunity here that you're seeing for the PC industry?

Ketan Patel: 

I think it's a phenomenal opportunity as some of the trends which we spoke about because the category itself is transforming at a very different level versus a regular, I would say, addressable market which you would have seen in the past. I look at it from a perspective that agent-based AI market is almost a $6 trillion TAM. by 2028 as a total agent-based AI opportunity. Now, within that, there is a certain area where the PC industry can play. And I think the devices and some of the ecosystem thing, which I spoke about, will drive huge premiumization in the category. So we spoke about the example of $650,000 for 1,000 employees likely to move to a $2 million, let's say, for 1,000 employees, and so on. If you start saving, as a customer, almost $1,000 to $2,000 per user on a PC, the premiumization in terms of getting some of these agentic capabilities and the local models, which I spoke about, it will only increase premium on the PCs. So this entire opportunity of premiumization of the PC is only going to go and give tremendous opportunity to all of us. The other big area which we believe we have a huge role to play is in this entire local on-prem deployment and driving cost economics ROI based outcome for the customer. So that's another area, apart from the devices play, that's an opportunity unlock for everyone who is in this place. And we are, of course, the front runner and we want to make sure we are leading in this space. And lastly, because we are going to unlock a lot of new experiences through agentic and the personalization thing which we spoke about earlier. It gives opportunity to the industry to get into a new software solutions and other revenue and margin unlock, which the category has not seen before. So if you look at all these three layers, the opportunity is going to be now unlocking value beyond selling hardware and making transactional margins, but at the same time making money and profit for you and your investors, but you are now adding tremendous value to the customer as they are going through their AI deployment transformation. And that to me is going to be a huge value which we can add as an industry as customers go through this journey. Good answer.

Daniel Newman:

I agree. Vote of approval, the seal of approval. Patrick Moorhead seal of approval here.

Patrick Moorhead: 

Whatever that's worth, folks. We'll see.

Daniel Newman:

AI devices track here at the Six Five Summit. Caitlin, I want to thank you so much for making the time to be with us here. Thanks for having us into your home.

Ketan Patel: 

Thank you.

Daniel Newman

 Thank you, Dan. This isn't really his home, but it's the home of HP.

Ketan Patel:

 Yes.

Daniel Newman: 

I don't know. We don't know how many hours he works. Maybe it is his home. Homish, and thank you everybody for being part of this year's Six Five Summit. You just heard it here. Ketan Patel, President of Personal Systems at HP, kicking off our AI devices track. Be part of our Six Five Summit. Stay tuned with all of our other content. Check us out at sixfivemedia.com. We appreciate you. More coming up next.

‍