HP's Ketan Patel on Turning the PC Into a "Personal Data Center" for AI Agents
TBS CROSS DIG with BloombergIn this episode of 1on1 Tech, TBS CROSS DIG business editor Masahiro Nakagawa interviews Ketan Patel, president of HP's Personal Systems business. The conversation starts with HP's new machines built around NVIDIA's RTX Spark chip, which Nakagawa describes as a laptop that "becomes sort of a miniature data center." The central question is how much AI work will move from the cloud to devices people own. Patel's position is that the PC is becoming a "personal intelligence" device shared by one human and many agents. In his view, AI inference will become a hybrid of on-device, on-premises, and cloud computing, driven by token costs, security, and data sovereignty.
From Tool to "Personal Intelligence": One User, Ten Agents
Patel argues that for many years the PC has been an input-output tool: you give it input and it returns output. He sees a new opportunity in what he calls a shift from personal computing to "personal intelligence."
The main driver, in his view, is agents. "Today one PC is getting used by a single user but tomorrow a PC will be used by one user and 10 agents onto it," he says. Agents will move through different applications and issue commands to reach outcomes for the user. Patel says this changes how compute is consumed. The machine has to support multiple actors working in the same environment and manage their tasks intelligently, so local hardware has to become much more powerful.
He presents the RTX Spark machines as HP's answer to this need. The flagship he shows is the OmniBook Ultra, which he says launched "last week." Its key feature, according to Patel, is a unified silicon architecture in which CPU, GPU, and memory share one pool. The machine has 128 GB of unified memory. Because of that, he says, workloads run locally with efficiency "like of a class of data center." He adds that the point is not agents for their own sake. What matters is how they change workflows, productivity, and experience for freelancers, developers, creators, small businesses, and enterprises.
"NVIDIA Inside"? Patel Says Multiple Architectures Will Coexist
Nakagawa asks whether the old "Intel Inside" PC is becoming an "NVIDIA Inside" PC, with a GPU in every machine. Patel does not accept that framing. He lists Intel, AMD, NVIDIA, MediaTek, and Qualcomm as silicon providers and says HP works with all of them. In his view, each architecture is more efficient for particular use cases and price points, so a range of options matters to users. "There are multiple architecture which will coexist bases the use cases," he says.
On GPUs specifically, Patel predicts that every PC will become more capable and will include strong graphics capability. Whether that comes from a unified memory architecture or a discrete graphics card is, in his words, "secondary."
Why Developers Are the First Audience
Asked what NVIDIA's entry means for a PC maker like HP, Patel calls RTX Spark "a great addition to the category" and names developers as its classic user. In the past, developers had to go to the cloud for many of their tools. Patel says a model of roughly 120 billion parameters can now be hosted locally on this machine. He lists the benefits he sees:
- practically no token cost for tasks that can be done locally;
- more security, adjusted to whether the user is a freelancer, gig worker, or enterprise;
- no internet requirement, so a developer can keep running AI workloads while traveling with a weak connection.
Patel says this makes development cycles faster and more efficient and calls it "a breakthrough for developer segment." He also stresses NVIDIA's CUDA stack, which he says is very important to the developer community. Together, he says, it makes the product a fully integrated solution rather than hardware alone.
From 95/5 to 50/50: HP's Estimate of the Local–Cloud Split
Nakagawa asks how much data center computing can actually move to personal devices. Patel answers that the question is not about one device. HP's range, he says, will host models from about 4 billion to 20 billion, 50 billion, and 120 billion parameters across different products. He notes that even a mini desktop could run a 4-billion-parameter model for a specific task.
He then gives HP's assessment. Today, the majority of AI workloads run in the cloud, roughly 95% by his figure. Over "a couple of years," HP believes the split between local and cloud will be about 50/50. He describes this as HP's projection, not a measured result.
The main economic argument is token cost. Patel says token economics is becoming a big topic, especially in large organizations. He asks the listener to imagine giving AI access to all 10,000 employees and paying for it. Nakagawa adds that some companies spend a billion dollars, and Patel agrees. Organizations want every employee to have AI, Patel says, but they also care about cost and ROI. Balancing these, he concludes, the world "is going to be essentially hybrid," with inference spread across small local models, 120-billion-parameter local models, and cloud models.
On-Premises AI and Data Sovereignty: The Z Fury Workstation
Patel describes a third tier between the personal device and the cloud: on-premises AI. Many customers, he says, worry about security and data sovereignty. They do not want their company's intelligence to enter a public domain where it "becomes an intelligence for somebody else." But they still want the power of a strong model.
For that, he points to HP's Z Fury, part of its workstation line. He says it can host a model of up to one trillion parameters. Units can be stacked, and he gives the example that ten of them could host a 10-trillion-parameter model, all running locally. His summary of the future architecture is on-device, then on-prem, then cloud.
Later in the interview, Nakagawa returns to this workstation. He notes that it runs on NVIDIA's GB300 and can run a very large model on device, and Patel confirms the GB300. Nakagawa suggests these machines will appear in workplaces first but may eventually reach homes.
A Sales Professional's Day With Local AI
Nakagawa notes that much of the audience is general business professionals, not engineers, and asks for a concrete example. Patel describes a salesperson's day. The core job is to interact productively with clients, but much time goes elsewhere: more than 100 emails, at least about 30 minutes of preparation per meeting, note-taking during meetings, and then follow-up such as emails and task assignments.
In Patel's scenario, a local model already on the salesperson's machine accesses their applications, such as a CRM, before an 11:00 a.m. meeting and prepares the client briefing. He emphasizes that this is "all free all local." Confidential client data does not leave the network, and the salesperson skips the emails and preparation. During the meeting, a meeting agent powered by the local model captures the discussion. The salesperson can focus on the client instead of action items. Afterward, the agent sends emails and tasks to the relevant people. Patel says the same pattern applies to other roles in enterprises, small businesses, and consumer settings. He presents this as a vision of how the products could be used, not a deployment he reports.
Does Expensive Hardware Widen the AI Divide?
Nakagawa points out that high-end RTX Spark machines cost over 1 million yen, while cloud AI is also getting more expensive. He asks whether this could widen the gap between people who can afford AI and people who cannot.
Patel replies that the RTX Spark machine is not the only route to local AI. It is the highest-performing device at its price, aimed at users like developers and game creators. For them, he argues, the cost pays off through productivity and avoided cloud spending. A sales professional does not need that much capability. Patel says such a user could buy a $1,000 product and still get local AI for basic tasks. He names the OmniDesk Mini, a mini desktop, as HP's entry-level product with local AI capabilities. The range then goes up to RTX Spark and further to workstation-class on-prem devices. HP's position, he says, is to have solutions across price bands for different users.
Mac vs. Windows: The "Open Ecosystem" Argument
Nakagawa observes that many AI developers use MacBooks and keep the lids open so their AI workflows don't stop. He asks why someone should choose an HP Windows machine. Patel says every user can choose the ecosystem that suits them. HP's pitch is openness: users can pick any model or application and work across operating systems without being locked into one ecosystem.
This matters for agents, he argues, because someone may have a phone on one OS and a PC on another. In an agentic world, they may give a command from the phone and expect the PC to carry it out. Patel says HP focuses on interoperability and continuity across devices.
Nakagawa pushes back: Apple has smartphones, laptops, and PCs, while HP has no smartphone. Patel repeats that HP is "an open ecosystem player." Users can choose operating systems ranging from Android to Windows and are not locked in.
Memory Prices and the Total Cost Argument
Will RTX Spark-class machines get cheaper? Patel says scale tends to bring down technology costs over time. But he says, "I can't predict that this is going to be a cheaper cost after few years." He expects that wider adoption of unified memory architecture products and scale across silicon architectures will help users get better technology at better value over time.
Nakagawa then asks about the recent surge in RAM prices. Patel says hyperscalers' buildout has made memory the most constrained commodity and "the most important commodity in the world of AI." He notes that memory costs have risen significantly year on year over the past 12 months. HP's response, he says, is to design memory-efficient systems that deliver maximum output with a smaller memory footprint. HP is working on its own architecture and with Microsoft on optimization.
He also argues on total cost. According to HP's internal analysis, he says, using local AI on a PC saves about $650 per user for basic tasks and up to about $1,200–$1,300 per user for advanced tasks. He presents this as larger than the extra hardware cost, so the total cost still works out despite recent commodity price increases. These are HP's internal figures as reported by Patel.
Young Smartphone Users and the "Sleeping PC That Keeps Working"
Nakagawa says he has heard that many young people grow up on smartphones and start using PCs only when they join a company. He asks whether AI gives them a new reason to own a PC. Patel says HP believes so. He agrees the phone will remain everyone's primary device. But as more agents are deployed, he argues, they need "a little more powerful surface area" to run on, which makes the PC critical. He adds that the PC has always been a strong creation device and more local power improves that. He describes the relationship as complementary: the phone becomes a "command center" that gives instructions, and the PC executes them.
Patel predicts that time spent on the PC will rise, but not the user's own time. Agents will work on the user's behalf in the background. Today people use a PC for anywhere from 2 to 12 hours depending on who they are. In the future, he says, "when your PC is sleeping the agents are still working." That would use the machine's full capability around the clock, which he sees as a big advantage since PCs are not used 24/7 today. Asked whether "command on the phone, run locally on the PC" will become normal, Patel says device continuity will be part of normal life: commands given on one device, executed on another, and results checked on the phone.
Containing Unpredictable Agents: Wolf Security
Patel raises device-level security next. HP's Wolf Security stack, he says, includes work on a BIOS-level secure root of trust. This matters in the agentic world, he says, "because agents can operate in a way which can be unpredictable." HP is using its security stack to contain agents and apply governance. He frames HP's offering as more than a product at a price: it is a user experience, the ROI from the sales example, and security together.
Nakagawa mentions a recent rise in cyberattacks in Japan over the past few weeks and asks whether security is the reason to run AI locally. Patel says it is "one of the reasons," not the only one. He positions local AI plus HP's container technology as a reason to choose HP, calling security and trust a core problem that every customer in the agentic world will face. He notes that HP built an endpoint security controller, which he says was first in the industry, because it recognized that hacking can happen through the endpoint.
Managing 60,000 "Users" in a 10,000-Person Company
Nakagawa cites reports from AI labs suggesting agents could automatically take over connected devices and run models on them, and asks what cyber defense will mean in that world. Patel describes two layers. The first is on-device agent security. The second is HP's Workforce Experience Platform, a manageability platform HP is building. It monitors how many devices and agents exist in an organization's infrastructure. It also lets IT administrators define which agents may or may not be deployed on machines and what governance applies to them. Each agent might even have something like an Active Directory identity, the way human employees do.
Patel describes the scale with a number. Today a large customer might manage 10,000 devices with 10,000 users. Tomorrow, he says, it will manage 10,000 devices and 50,000 agents, which means effectively monitoring 60,000 users for a 10,000-employee organization. That, he argues, is why the manageability layer is critical.
Will Every Home Have a Data Center?
Nakagawa asks how workstations, not just laptops, could change how people interact with AI, including at home. Patel says a workstation like the Fury can host large models locally, which makes it a strong fit for enterprises and small businesses. People with that requirement could extend it to their homes. But a home could also have an OmniDesk Mini hosting agents, while a regular PC or phone handles other applications. He describes a range from basic "agentic PCs" HP is developing up to the Fury level. He again insists the world will not be "either local or cloud." Even at home, a local machine would handle most of the load but should be able to reach the cloud when needed. For a home user, he says, what matters most is that agents manage tasks such as the personal calendar and travel booking. Hardware should be sized to each customer's needs.
Nakagawa explains why he asked. A previous guest on his show, Silicon Valley angel investor and All-In podcast host Jason Calacanis, said that people running AI on powerful home workstations won't worry about token costs, which changes their behavior because they no longer watch a usage meter. Patel says, "I love what he said," and calls it exactly HP's thesis. He adds that homes will have different needs, so HP plans "good better best" options rather than one size fits all.
Nakagawa also mentions a Japanese startup he covered that is developing manipulation robots for homes and says each home would need a workstation. Patel says that for robotics, yes, workstation-class power would be needed. For basic use cases like the ones he described, a less powerful machine would be enough.
Beyond the Laptop: Form Factors, Robots, and Expanding Workstation Markets
Patel describes Personal Systems as more than laptops. It covers laptops, desktops, displays, peripherals such as headsets, keyboards, and mice, HyperX gaming peripherals, workstations, and related services and solutions. On what comes next, he returns to his opening theme. The PC will become "a digital twin of yourself." It will capture the user's personal knowledge graph and understand their context better than any other device, as an always-on, context-aware device for personalized experiences.
He says this will "absolutely" lead to new compute form factors beyond the classic laptop, but he declines to give concrete examples because he cannot share HP's roadmap. When Nakagawa asks whether HP is thinking about robots, Patel says HP's AI focus is currently on compute, print, workstations, and "AI stations." HP looks at both horizontal solutions and specific industry verticals. If a full solution required something like a robot, even one not made by HP, HP would consider partnering to deliver it. "But that's not something as of now we have thought of but you never know," he says.
On workstations, Patel says they have mainly served high-performance niches: CAD/CAM, architecture, engineering and construction, heavy graphics, and media and entertainment rendering. Local AI and agents, he argues, will expand their addressable market beyond those classic uses, because they combine compute power with the ability to host high-capability models on device. He places workstations at the center of HP's "edge AI" positioning.
On how personal AI should work across many devices, Patel says HP designs around a user's whole workflow, not single-device solutions. His examples are hardware-to-hardware: an HP PC working smoothly with another HP PC, with HP printers without the usual printing steps, or with HP headsets. He says the same principle will apply to full workflows in the agentic world.
Looking Five to Ten Years Ahead
Asked how they imagine HP five or ten years from now, Patel says the AI-at-the-edge opportunity will open up beyond where it is today. In conversations with customers, he says, HP sees that token cost, security, and data sovereignty are "all real issues" in real AI implementations. HP's aim over the next few years is to provide "outcome-based solutions" to these pain points by investing in its portfolio, solution stack, and services. He adds that technology is moving so fast that HP will keep innovating across its portfolio toward that vision. He offers this as the only "sneak peek" he can give, without a specific roadmap.
One data center.
Autist.
PC HP.
Okay, we have Ketan Patel, president, Personal Systems at HP, on our show. Thank you so much for joining us today.
Thank you for inviting me.
So, you lead all of the personal computer business at HP, and here we are with NVIDIA's RTX Spark-powered PC, and now a laptop becomes sort of a miniature data center. So what changed with PCs with RTX Spark? And how is PC changing in this era, AI era?
Yeah, it's an exciting time in the personal computing world. PCs have been tools which you've been operating for the last many years. You generally use this as a device which you give in some input and it gives you certain output. It has been predominantly like that for a while.
But what we believe is a completely new opportunity of how personal computing is moving to a more personal intelligence kind of a device. And essentially this really happens because of a couple of things. One, in the world of agents, humans and agents are going to work together on a compute, and a lot of agents will be deployed on the PCs tomorrow. Today one PC is getting used by a single user, but tomorrow a PC will be used by one user and 10 agents onto it.
So it fundamentally changes the way compute will be consumed and how humans and agents will work together, which essentially means that compute has to become much more powerful on the ground in terms of how it can allow multiple users to operate in an environment and intelligently manage tasks. Because agents will, of course, navigate through different applications, through different commands which they need to drive for driving a specific user outcome, and that's why the systems have to get really capable.
And this RTX product, which got launched last week, and this is our OmniBook Ultra, which is our flagship product around this, is an amazing launch which we are very excited about. Because it is having now unique capabilities of bringing silicon architecture where a CPU, GPU and a memory is unified into a unified architecture. And this machine has 128 GB unified memory, which essentially means now when a workload happens locally, the efficiency of the workload which operates on the machine is like of a class of data center. And that's where I think the architecture change, how the silicon architecture is changing, is going to help with this launch.
Secondly, how these entire agents are actually driving meaningful outcomes. It's not just agents operate, but different kinds of user profiles. Either you are a freelancer, you are a developer, you are a creator, you're running a small business or you are in an enterprise environment. How this is essentially changing your workflow and your productivity and experience is what we are very excited about.
So we used to call the PC sort of an Intel Inside, but are we going to call the PC like NVIDIA Inside sort of thing? So GPU is going to be in every PC.
The silicon architecture is moving dramatically across different silicon architectures. It could be Intel, AMD, NVIDIA, MediaTek, Qualcomm. There are a lot of silicon providers. The great thing for us is we work across all the silicon partners and ensure that how we are bringing in solutions for the right customer segment using the right silicon architecture. Because different silicon architectures are going to be useful and more efficient for specific use cases at specific price points, and that's why that range which we will have is going to be super important for the users. So I would not say that there is one architecture only. There are multiple architectures which will coexist based on the use cases.
I mean, I wonder if we're going to see every PC has GPUs to run AI locally in the future.
I would say that every PC will be more capable, including an integrated GPU or the way new architecture of silicon is going to look like. It will have intensive graphics capability. Either it is UMA or it is discrete graphics, that is secondary, because eventually there'll be a graphics capability irrespective in the machine, and as they get more and more powerful.
NVIDIA launching their products like RTX Spark, what does it mean to a PC manufacturer like HP?
I think it's a great addition to the category, as I said, because this now brings a very different level of capability on the device. And this will essentially allow us now, for the RTX Spark product, developer is a classic user for these devices. Now developers in the past had to obviously go to cloud to access a lot of tools to develop their solutions. Today now, with almost a 120 billion parameter model which you can locally host on this machine, you can practically now start doing your work more efficiently, with practically no token cost for whatever task can be done locally, more securely.
Depending on what you are, if you are a freelancer, if you are a gig worker or if you are an enterprise, as I said earlier, there are different needs of security layers. It can be securely executed. And more importantly, you don't even need internet. So if you are traveling and if you don't have a great internet connection, you can still work on AI workloads through local models. Makes your development cycle much faster, efficient.
And this is going to be a breakthrough for the developer segment, given it just changes the way they look at some of the products. And with NVIDIA's CUDA stack, which is going to be super important for the developer community, this makes it like a fully integrated solution and just not hardware alone.
So how much of the computing previously done in the data center can be processed on the personal device like this? You said that a 120 billion parameter model can run on this device. So how much workloads can we get done by this device?
Yeah. So now we're talking about one device, but essentially this is not about just one device. You will have various different kinds of compute products across what we will bring in, can host from a 4 billion parameter model to a 20 billion parameter model to a 50 billion to 120 billion. There are different kinds of models which can be hosted on different kinds of devices.
Not just laptops.
No, not even on a desktop.
All of the...
Yeah. You know, on a mini desktop you can actually have, let's say, a four billion parameter model which can run a specific task. So when you combine all this together, now the power of hosting local AI goes from a 4 billion parameter open-source model to different kinds of models depending on the use which you have.
Our assessment is that today the majority of the workloads are managed on the cloud. Assume 95/5, if I can say that number. Tomorrow, over a couple of years, we believe this will be like 50/50 in terms of how these workloads will get divided between local and cloud. You can do a lot of work through these local models, which will save a lot of token cost, which is a big topic today. As you know, token economics is becoming more and more a topic, largely when you have to deploy in a large organization which is having 10,000 employees. Imagine if you're giving access of AI to everyone and the cost associated with it. So economics is an important topic.
Some company spends a billion dollars.
Exactly. Yes. So it's a big topic, right? And it's a right topic because on one side you want to give access of AI to every employee of your organization, which I think is important in today's world. But at the same time, organizations are cognizant of the cost associated and the ROI of that. So if you really balance all this, the world is going to be essentially hybrid as inferencing happens, and a combination of local, as I said, from small model to 120 billion parameter model to a cloud model.
There is also another thing which we are very excited about, is that it's not about just model on the PC, individual personal computing. As you think about on-prem AI, because a lot of customers are also concerned about security and data sovereignty. Because a lot of intelligence of your own company, you don't want that information to go in a public domain, which becomes an intelligence for somebody else.
But you still want the power of a good model. So if you want a bigger model, we have a product like a Z Fury, which is part of our workstation-class product, which can host even up to a one trillion parameter model. And you can put it on the premises. You can actually stack them. Let's say if you assume you are stacking 10 of them, you can have a 10 trillion parameter model sitting on that, and you can run everything locally. So now you have on-device, then you got on-prem, and then you got cloud. So that's how going to be the future.
Most of our audience is general business professionals. Some of them are engineers, some of them are sales and marketing. So to make it more understandable for them, how are we going to work with each one of them, like on-device AI, on-prem AI, cloud AI?
Like as you called out, let's talk about a sales professional. What's a typical life of a sales professional? Your main job is to interact with clients. That's your ultimate aim, that my interaction with clients should be the most productive and I come across as somebody who's adding value to my client. But what does a salesperson today do? Spend time on probably 100-plus emails, spend time on preparing meetings, maybe at least 30 minutes per meeting for prep time.
When the client meetings are happening, most of the time they capture notes, because that's what happens in that use case when you're interacting with a customer. So your focus is having a conversation, but also capturing my next action items, right? And then the third task is, after the client meeting, then look at your notes, pull out everything, do the follow-up, either send email or send a task to someone, etc.
Now imagine today, tomorrow, in the world of local AI for a salesperson, a local model is already on your machine, which fundamentally accesses now critical information for your client meeting preparation. So as you come today morning and you have a meeting at 11:00 a.m., that local model has already accessed your application across either you are using, let's say, any customer CRM or whatever information you need to fetch. The client briefing document is readily available, all free, all local. The data has not gone out because there's a lot of confidential client information which you don't want to be published outside of your network. But then you have got the briefing without all the task of going through emails, and meeting preparation is a huge productivity gain.
And as you go to the meeting, when the meeting is happening, the local model which is on the machine and the meeting agent can capture the meeting as you discuss in the conversation. So the person is focusing on client interaction rather than worrying about capturing actions or capturing other discussion points. And then the agent, which is powered through the local model, is taking respective actions based on the meeting and sending emails or tasks to different users. These examples can be extended to different kinds of personas in an enterprise environment or a small business environment or a consumer environment.
But RTX Spark-powered machines are really expensive right now, yet. The higher-spec one is, I think, over 1 million yen, which is really costly. But it costs also to use the AI models on cloud, getting more expensive right now. So do you think this could sort of widen the economic gap between people who can afford and cannot?
Yeah, and that's why I gave you an example that this is not the only product which can have some of these local model capabilities. This is like the highest-performing kind of device at a certain price equation, which could be useful for a certain class of users like developers or game creators. They can afford this kind of cost because the associated ROI, both in terms of their productivity and the cost which they would otherwise incur through a regular cloud access, can make their economics work.
But imagine another kind of a user who can still buy a $1,000 product and still have access to local AI and can do their basic task. Because if you are a sales professional, you don't need that kind of a device which is having so much of capability. And we have a starting point of a product called OmniDesk Mini, which is a mini desktop, which can hold an entry-level price point and still ensure all these local AI capabilities which I spoke about. Take it throughout different kinds of range, go up to RTX Spark, and then from there on go to a workstation-class product, which are even on-prem devices, as I spoke. So we have solutions across different price bands for a different kind of user set.
I've seen many AI coders, AI developers using MacBook or Mac machines, and everyone doesn't want to close the laptop up so that they don't stop the AI workflows, and they are using the Mac machines. So how would you make the case for choosing Windows machines like HP?
I think there are different ecosystems, and every user has a choice to pick up an ecosystem which works for their use case. I think the good thing about what we are trying to do from an HP perspective is, number one, we are trying to make this a more open ecosystem for anyone. So there's no restrictions, right? So you can pick any model, you can pick any application, you can seamlessly work across different types of operating systems in the market. So for a user, they don't have to worry and get locked into a single ecosystem.
That's very important because you may have a phone of a specific operating system, or you have a PC on a different operating system. In the world of agentic, you want to give commands through your phone, but your PC should be really accessing the command and still work. So how the interoperability and continuum happens across the ecosystem is what we are focused on. And as I said, different kinds of users will see advantage through the ecosystem of what we are providing, and then they make their choice.
Well, when you think about interoperability, Apple has a smartphone, the laptop and PC, but HP just has a PC, workstation, mini desktops and sort of thing, doesn't have a smartphone. So how does it work?
Yeah. And that's why I'm saying that we are an open ecosystem player. So you can choose any kind of operating system, from Android on one side and Windows on the other side, and you get that freedom rather than getting locked into a single ecosystem.
What do you think about pricing of the future machines? Like, machines like those RTX Spark-powered machines are getting cheaper in the future and more affordable?
In the world of technology, as you know, scale on any technology over a period of time helps on the cost. Now I can't predict that this is going to be a cheaper cost after a few years. But given how unified memory architecture-based products will be accepted in the world of agentic AI, and as the scale improves through different silicon architectures, which I spoke earlier, it is going to be helping users over a period of time in terms of how they'll be able to access a better technology at the right kind of price equation, so that there is a good value out of your every single yen or dollar which you spend.
Yeah. But how does the recent price cycle of RAM play out in that pricing?
There has been a huge demand as hyperscalers really started building capability. Memory is the biggest commodity which is constrained, and memory is the most important commodity in the world of AI. So when you think about demand-supply equation, there's always cost in any category. It's not about just memory here, which is what the world is facing in the last 12 months, as you see cost of memory going up significantly year on
Year. No different in our category as it is in any other category. And our aim is to make sure that we make memory-efficient systems so that in a lower memory footprint also you get the maximum output. That's our aim. We continue to work on our own architecture. We continue to work with Microsoft in terms of how do we optimize this better. That's what our continuous effort is on one side.
On the other side is how do we bring in value to this machine locally versus what the total cost which a customer is paying. Our own internal analysis right now says that by using local AI on a PC, you are getting almost $650 per user advantage for basic tasks. And as you go to advanced tasks, it goes even up to $1,200, $1,300 per user.
That's a huge saving against the kind of additional price which you are paying for the machine, which is giving you an additional input cost right now. But imagine the kind of savings you get just because of the hybrid AI. So when you look at the total cost equation, this still works out even with the cost increase which has happened because of the recent commodity pricing.
I think for younger people, they're more used to smartphones, like typing on the smartphone. I think I heard they just start to use PCs when you start working in a company after graduating college and something. But does AI provide a new reason to own a PC for them?
We believe so. We are in this business and we are very excited about it. I think phones are going to be still an integral part of anybody's life. It's your primary device. I don't think there is going to be any difference of opinion there. But in the world of AI, as I said, as you deploy more and more agents onto the machine, you need a little more powerful surface area to host it, and that's why PC will become a very critical component.
Second, PC was always an extremely good creation device of anything which you wanted to create. And with now more power sitting on the machine, this will even make it better for your user. But more importantly, the way we were thinking, or the way we are thinking, what I just said earlier to you is we want to make sure that what we do on the PC complements what they can do from a phone perspective, because phone will also become a command center in some way of giving command. But the PC will execute command.
Yeah.
So I think both will coexist in that sense, and that's our aim, to make sure how we are driving that experience across phone and PC together. But essentially the time spent on the PC is going to go up versus what historically it has been. And when I say time spent, it's not a user's own time, because another aim we have is, in the world of agents, the agent is working for the user on the PC on behalf of a user. So the overall usage will go up.
So time spent on PC doesn't mean the time you're in front of the PC.
That's right, because in the background your defined agents are working on your behalf individually for you, and that makes it productive. Imagine today how many hours you work on a PC, maybe between 2 to 8 to 10 to 12 hours, depending on who you are. But tomorrow when your PC is sleeping, the agents are still working. So you are now using the full core capability of the architecture available 24/7, which I think is going to be a huge advantage for people to use the capability the machine is having and you are not using 24/7 today.
So is it going to be a common thing for us to use both phone and PC, like a command on phone and run AI locally on PC? Is it going to be common?
I think device continuum is going to be important, that I give command from here, the command gets executed here. I do command from here, I access it on the phone. That's part of the normal life which we'll see in a normal user scenario. And more importantly, a lot of work HP has done on building device-level security.
So we have a stack called Wolf Security, which we have been working on, a BIOS-level secure root of trust. That's going to be super important in the world of agentic, because agents can operate in a way which can be unpredictable. And how we are using our security stack to contain the agent to operate in a certain way, so that the right level of governance is applied to that, is also going to be important.
So when we think about what we are providing, it's not just about a product at a certain price. A product at a certain price which can deliver a certain user experience. The example of a sales professional I gave, the kind of ROI and economics associated with it, the security which is going to be an important element associated.
As you brought up the security, we've seen hikes in the number of cyberattacks in Japan for the past few weeks. I think as devices are more connected, there are more risks being exposed to those attacks. And AI, as you said, AI can be disruptive in those situations. But is it the reason why we should run AI locally?
One of the reasons. I would not say the only reason, but definitely local AI and some of the container technology which I'm talking about, that's why you should also choose HP. And that's the differentiation which we are trying to focus on, because this is a core problem in the world of agentic which every customer is going to deal with. Security and trust are going to be important, and HP gets associated with that in terms of solving it over the years, even in a non-agentic world, in a cybersecurity world. That's the reason why we started endpoint security controller, first in the industry, because we realized hacking can also happen through the endpoint.
I've seen some reports from AI labs that say that AI agents can automatically take over all those connected devices and run the models on those devices. So I think that in the AI era, there's going to be more risks being exposed to those AI agents running everywhere. So how do you defend the devices? What's cyber defense going to mean?
So as I said, the one defense which we are obviously focusing on is on-device agent security. The other thing is this Workforce Experience Platform which I spoke about. This is a manageability platform which we are creating, which essentially monitors the number of devices and number of agents in your infrastructure. It also gives IT or administrator a right to define what agents can be deployed on the machine, what cannot be defined, what kind of governance it has to have. Each one of them may have an Active Directory like a human, each one of us have in our enterprises.
That is a super important layer of security from a manageability perspective we are also working on. So you get two kinds of security layers associated, one on-device, the other at a manageability level. And if you look at it from a customer who has large deployment today, you are managing 10,000 devices with 10,000 users. Tomorrow you are managing 10,000 devices and 50,000 agents. So you now have to monitor like 60,000 users for a 10,000-employee organization. So that entire manageability layer is super critical, which is where HP's Workforce Experience Platform plays a significant role.
And I also want to touch on the workstations that you talked about, which is the Fury machines that can run Kimi K3 on-device.
Yes.
So that's a massive machine.
Yes.
And that runs the GB300.
GB300, yes.
Yeah. So that says that we can put data center-scale AI on the side of the desk. But I think these workstations are going to be implemented in the workplace first. But I can imagine this is eventually coming to our home as well. How could those workstations, not just laptops, change how we interact with AI?
Yeah, because these are various kinds of use cases you can think of for an enterprise or a classic small business. This is an amazing solution because it can host this kind of a model locally. You can extend this to your home if you have that kind of a requirement. But you can also have an OmniDesk mini kind of product at your home which is fundamentally hosting agents for you. And while you're on a regular PC, a phone works on the rest of the applications.
So you can have a scenario where you have a local model through basic-level agentic PCs which we are developing and go up to ZGX Fury level depending on your need. And then we believe that the world is not like either local or cloud. The fact of the matter is that you will have a hybrid environment all the time. So even in a home situation, to your point, you can have this PC or Fury kind of a product which will do most of your loads. But when it requires to go to the cloud, it should have flexibility to go so that you have those operations going on your behalf.
Eventually your core need is, as a home user, if my agents are ensuring that my personal calendar is managed, my travel booking is getting done, how each of these agents are doing specific tasks for me and make it meaningful for me. I think that to me is most critical. And the sizing of the hardware and which is the right hardware can be done depending on the use of a specific customer.
The reason why I asked this question is that I had my guest, the angel investor Jason Calacanis, that he's from Silicon Valley, and he runs a big podcast, All-In. And he said the people who run AI on those powerful computers like workstations at home, there's not going to be a lot of token cost using those workstations locally. So that would change their behavior, users' behavior, because they don't have to care about the usage meter anymore. So what do you—
I love what he said. It's great, because this is exactly the thesis we too have, because it will solve a lot of your issues without really going into the cost of tokens, as we spoke about, and it does your job easily. So yes, the only thing I am stating is that every home may have a different kind of a need. And the good thing is, what we are thinking of is depending on the needs, we have good, better, best kind of solutions so that it's not like one size fits all.
Yeah. I think I just covered the Japanese startup that's working on the robots, the manipulating robots at home, and they said they need to have a sort of workstation at every home. So is that what we're going to see in the future?
Yeah, if you are trying to do that kind of application, maybe yes. But if you have a very basic use case like what I was citing before, maybe you don't need that kind of powerful machine. So as I said, if you're doing robotic kind of work, of course you will need workstation kind of product, because you need the right level of power equation associated with it.
Okay.
Yeah.
You lead the personal systems business at HP. The personal system means a variety of devices, not just PCs. So I was wondering what you see as the next big thing beyond PC in the AI era.
Yes, you're right. Personal systems includes not only PCs, because sometimes PCs gets referred to only laptops, but PCs, which is a combination of laptops, desktops, all the kinds of displays, associated peripherals including headsets, keyboards, mice, also peripherals for our gaming devices like HyperX, which is an amazing gaming lineup we have. And then workstations, as you called out a couple of times, and associated services and solutions along with that. So that's the full stack of personal systems, and the innovation can happen in various ways.
Agentic AI can actually change the way personal computing experience is going to be. As I started the conversation, a PC is going to be a digital twin of yourself. It needs to capture your personal knowledge graph. It needs to know your context better than any other device, because this will be a context-aware device all the time for you. And that's how our aim is to deliver personalized experiences.
Now, can this lead to different kinds of device form factors in the future? Can it be beyond the classic form factor of a laptop, what you see here? Absolutely yes. You will see different kinds of compute form factor evolving over a period of time. I can't give you a concrete example today because I can't share my future roadmaps in terms of what I'm going to have. But essentially this is absolutely the direction where you will see the form factor evolution will happen, because this is going to be a very critical context of a device for driving the right level of agentic experiences.
Can I possibly take a peek at your future roadmap?
You can try.
Are you thinking about robots?
Edge AI, as I said, we are an edge AI company, and AI is fundamentally focused around compute, print, workstations, as well as AI stations kind of devices right now. And we have to think about both horizontal solutions for different use cases and then specific vertical use cases also. We are focused on some specific industry verticals. And in that context, if we need to partner with someone to provide a full solution stack which includes certain solutions like a robot, even if it's not an HP robot, if we should provide a full stack solution, we will consider partnering with someone and say, hey, this is how we do an end-to-end solution stack. But that's not something as of now we have thought of, but you never know.
I think in terms of workstations, there are more use cases than before, right? There are more potential customers than before. So how are you seeing the use cases expanding in the future?
Yeah. Earlier workstations were used mainly as high-performance compute products. So if you have CAD, CAM applications and architecture, engineering, construction, some heavy graphics usage—
Sort of a niche.
That kind of a segment is buying workstations. You have rendering, like the studio kind of requirements, so media and entertainment, basically. So there were segments which used to prefer workstations largely because of the compute load.
But in a world of agentic AI and local AI, the workstations, because of the capability it will have, both in terms of compute power as well as ability to host higher-capability local models on-device, it will drive different kinds of additional addressable market, I would say, and create more opportunities for those categories to expand beyond classic workstation use cases. Which is where I think this entire edge AI positioning which we have as a company, workstations are going to play a critical role in that entire opportunity.
Yeah, we're going to have, I think, a variety of devices in the future. But how does personal AI work across all the devices?
As you think about solutions, it should not be a pointed solution on one device. And so when we think about it from a design or a workflow, we think about the full workflow of a user rather than just only on our device, and make sure the solution which we are crafting is thought through across devices, including our own device. How a PC works better with another HP PC seamlessly, or how a PC works with our printers in a much more effective way, that you don't have to go through a normal route of going and printing. So how a PC and print will work together, or how a PC with our headsets works seamlessly, and so on. I'm giving examples of hardware-to-hardware seamless connectivity. But the same thing is applicable when you talk about a full workflow in the world of agentic tomorrow.
On the last note, you said you can't disclose the roadmap, but I was wondering how, as a leader of the personal systems business, you imagine the future of HP, like five years or ten years from now?
Yeah, I think as you think about the next few years of horizon, clearly the AI at the edge opportunity will unlock versus where it is today, because in our discussions with various
Customers, we see the AI implementation on ground, and some of the issues which we spoke about related to token cost, related to security, related to data sovereignty are all real issues. And our aim over the next few years is to provide outcome-based solutions to solve these fundamental pain points of the customer.
Now, what does it essentially mean? That we want to make sure that we are investing as a company on our portfolio, on our solution stack, on our services in order to deliver this. That's our aim.
Now, technology is moving at such fast pace that we will continue to focus on this outcome of vision of ours and then continue to work in innovating our entire portfolio to deliver that promise. So that's how you can do a little bit of a sneak peek in terms of what we will continue to focus on.
It's great to hear all of that. Thank you very much for coming.
Thank you for inviting.
Yeah.
[music]
Article published · Updated
