The Deep View: Conversations

What happens when powerful AI no longer has to live in the cloud?

In this episode of The Deep View Conversations, we talked with Mark Papermaster, CTO of AMD, about why the next major shift in AI could happen on the device sitting on your desk.

Papermaster explains how computers could soon run sophisticated models and teams of private AI agents locally, offering greater speed, security and control without recurring token costs. He also makes the case that AI will be even more transformative than the smartphone because it will be embedded across nearly every device, industry and aspect of daily life.

The conversation also covers:
  • Why open ecosystems matter in the AI era
  • How AI is accelerating science, agriculture and industry
  • The growing energy demands of AI
  • How leaders can reinvent workflows with agents
  • Why local AI could reduce cloud dependence and vendor lock-in
  • Papermaster’s lessons from four decades in technology
If you’re interested in less lock-in, open ecosystems, and how enterprises can run AI more efficiently and privately, this conversation offers a look at what a more distributed and secure AI future could look like.

Subscribe to Deep View Conversations for interviews with the leaders shaping the future of AI, business, and technology: tdv.transitor.fm 

And don't forget to sign up for The Deep View daily newsletter. We don’t just cover AI, we decode it. In a world flooded with hype, we deliver sharp, no-nonsense insights to keep you ahead of the curve and help you put AI to work every day: subscribe.thedeepview.com

Creators and Guests

Host
Jason Hiner
Editor-in-Chief of The Deep View

What is The Deep View: Conversations?

From frontier labs and enterprise platforms to emerging startups reshaping entire industries, The Deep View: Conversations podcast interviews the brightest minds and the most influential leaders in AI.

Mark Papermaster: AI is more than that. And the reason I think it's more than that is AI is going in every device. Siri and the other agents on the other the other phone vendors are all AI-ified now. You look at how the PC is where we do most of our content creation. And it's all amplified because our productivity is going way up because everything that we've done from whether I'm photoshopping, whether I'm creating artwork, it's all productivity amplified with AI, our ability to take entire end-to-end workflows and put them together. So it's across every domain, it's being embedded in every device around us.

Jason Hiner: Mark, welcome to the Deep View Conversations. You're pretty well known in the industry, but for those who aren't familiar, tell us a little bit about your role here at AMD.

Mark Papermaster: You bet. Jason, I joined AMD in an exciting time. Lisa and I were brought in a number of years ago, almost 15 years as AMD is a storied technology company, but really needed a reset. We were brought in to bring focus on high performance and really bringing a culture of just being completely a bankable partner. And Lisa became a CEO very, very quickly after we arrived and it's just been an incredible journey that we've been on that arrives at a great culmination today with some incredibly exciting products that we rolled out at Advancing AI.

Jason Hiner: What a ride over the last 15 years. For sure. And exciting stuff to come. How about for you, what gets you most excited about coming to work every day?

Mark Papermaster: Jason, I have a dream job. I was brought in as CTO and I am providing a direction oversight to not only our engineering, but our direction, our strategy, our technology strategy. And you think about over this period of time, it's been the dawn of AI. And AI stretches across our entire portfolio. So we have a breadth of portfolio from supercomputing to the biggest clouds to on-prem data centers to PCs to embedded devices. And so in my role as CTO, I get the opportunity to work with the incredibly smart people of AMD and guiding and making sure that we're meeting the customer needs. And to do that, I spend a lot of time with customers. And being the industry veteran I am, I've been around over four decades. I listen very, very well to customers. And I'm able to really connect their needs to our roadmap.

Jason Hiner: Sir, you have had this amazing career, working at IBM, Apple, Cisco, now 15 years at AMD. How did you get into tech? Tell us about the journey, what the journey has been like for you. And then lastly, what's your advice to people who are building their career in tech now?

Mark Papermaster: What a great question. Well, I always tell people, follow your passion. And I say that with heart because that's what I did. I started with a conception that I knew exactly what my passion was because I was so excited by the space program. So I became an electrical and computer engineer and space, you know, being the great frontier it was. I started with IBM, developing the computers that went on space shuttles. And it was great. But it turns out, I realized it wasn't my passion. I had taken some courses in computer architecture and chip design. And so after interning with IBM on the space shuttle program, I said, hey, do you mind if I take my remaining internships with the chip design work you're doing?

Jason Hiner: Wow.

Mark Papermaster: And it was indeed my passion. So I've been so fortunate. And it was the very first CMOS devices and it's grown now to such complexity where we, as we announced today, over 300 billion transistors on our new AI high-performance compute. But that's, you know, the recommendation I have to people is particularly in the AI era, what we are all doing is adding our value, our critical thinking. And if you bring a passion with that, so you're going to be using AI, AI is going to amplify what you do, your passion. I think you're going to have an incredibly successful career path.

Jason Hiner: What a great, very human set of advice, you know, that AI even more can really amplify sort of human ingenuity, you know, human passion, human intention. And so I love that. Well, you've also seen the PC revolution, the Internet revolution, the mobile revolution, and now the AI revolution. What's different now and also exciting and what's also sort of the scariest part.

Mark Papermaster: Yeah. Well, again, I have been very fortunate to have been through all of those reflections as you call out. And, you know, I thought I went through the biggest revolution because I joined Apple on running iPhone and iPod for Steve Jobs in 2008 and, you know, here was the phone, you know, becoming the computer in your pocket, the phone in your pocket, the camera in your pocket. So it was all of these things that, you know, brought us the pieces of our lives together in one device. And it was all of that. It was truly transformational. I think for all of us, as you know, your phone's sitting right there, it's like an extension of our body. AI is more than that. And the reason I think it's more than that is it's actually riding on the steps of every transformation before. And encompasses all of that. AI is going in every device. Our phones now, I mean, you know, you see, you know, Siri and the other agents on the other, the other phone vendors are all AI-ified now. You look at how the PC is where we do most of our content creation and it's all amplified. I use that word yet again, very intentionally, because our productivity is going, you know, way up because everything that we've done from whether I'm photoshopping, whether I'm creating, you know, presentations at work or I'm creating artwork, it's all productivity amplified with AI. And, you know, and as you go into much higher performance computing, agentic AI is able to just put on steroids our ability to take entire end-to-end workflows and put them together. So it's across every domain, it's being embedded in every device around us. And it's not just our work lives, it is also our personal lives and the example I gave. So it's just, it's really becoming a part of every aspect of our lives. And I didn't even mention physical AI. Physical AI is robotics. And so robotics means that AI is going to becoming a companion. That's going to be, not only it'll start off in factory floors, but when we have trust in these devices, it's going to be in our homes and helping us in every daily task.

Jason Hiner: When you think about, you know, your family, your friends, your community, what's the part, there's a lot of fear out there around AI. What are the things that you worry about? What are the things you think we want to make sure we get right? So that it benefits everybody and so that we can avoid some of the things that we've learned where, you know, deploying AI, deploying not just AI, but technology in general has caused some challenges in terms of society, in terms of, you know, again, community, all of that.

Mark Papermaster: It's a great question, Jason. And I think we play a special role at AMD addressing some of those fears.

Jason Hiner: Okay.

Mark Papermaster: You know, first of all, we're open source and open ecosystem with everything we do. We think there needs to be competition. And so, you know, we, with the breadth of our portfolio, we're up against giants and we're gaining a lot of share. Part of the reason we're gaining share is the fact that we're open. Open means you can trust it because you have millions of people that can see exactly the code source that you're running. If it's an open, open weight model. And in fact, even when there's constitutions and some of the models are becoming open of how those models were created, that's how we're developing trust with our open ecosystem. Or allowing there to be many partners that can work with us, listening to customers, tailoring it to their needs. And, you know, one of the other fears is rightfully so about power and energy. And we take that head on at AMD where we strive to bring every generation of product forward, you know, multiple fold in energy efficiency from the previous generation. It's our designers wake up every day driving efficiency. And the way that we do that is you think it's all in the transistor design, how we put it together. Turns out you have to optimize across the whole stack, across all the way through the end applications to optimize how you get the best performance at the absolute minuscule amount of power that you expend. And so, you know, these are things that we can control. We try and focus on how we can help build that trust. And these are areas that we're incredibly focused on at AMD. And we take it very seriously. We're part of the food chain of AI. And we take our role very seriously.

Jason Hiner: Mark, you talked to a lot of customers. You talked to a lot of people using AI. You talked to partners who are working on things, you know, that won't be released for months or years. So you know the direction that this is heading. What gives you the most hope right now in terms of where all of it's going and what you think sort of capabilities are going to be unlocked?

Mark Papermaster: Yeah, you know, honestly, Jason, it's the flip side of the previous question you asked. There is a lot of fears out there. And I believe, as I said, we're going to conquer those fears by transparency, openness, and making sure people are comfortable that we're being responsible how we develop AI capabilities. This question is the flip side. What gives me incredible hope is I see how AI is being applied for the good right now that's having already incredible impact. I get the opportunity to work with CTOs of pharmaceutical companies, drug discovery companies. We power the top two U.S. supercomputers, which are accelerating time to discovery. We're working with scientists working to unlock the secrets to enable fusion energy. And AI is accelerating. It's really called AI for Science. And this is what we dive into and announced today, the MI430X. It's a variant of the big MI455X, which excels at the most demanding AI and training, AI training and inference tests. The MI430X does that. It runs AI applications extremely efficiently, but it adds a super high precision. It's called double precision math that enables scientists who are working on ensuring that we're modeling the most difficult weather situations. Again, molecular discovery. It works hand in hand with quantum computing, which is coming around the corner. You're going to see the AI for Science being paired with quantum computers. And we work with the quantum computer companies enabled by our adaptive compute. So to me, I think it's easy to focus on the fears. And of course we have to address the fears. But I could not be more excited and enthused with what I'm already seeing on AI being applied to substantially improve all of our lives.

Jason Hiner: A lot of that's the HPC, the high performance computing parts of the journey, right? That so many of these companies are on. So you see it in the sciences. You see it in industry. Agriculture. Are there other places that we don't talk about a lot that are there other places that we see it? You just mentioned a few.

Mark Papermaster: Yeah. You go look at John Deere, Caterpillar. I could name five companies that are focused on agriculture. You would not believe it. They are some of the most technology advanced companies in the world. I mean, and they are driving technology that's dramatically improving the yield of crops. And they combine it with equipment that can have drone oversight over crops, analyzing minute areas that might be suffering some small area of blight that can be cut off before it takes off and damages a broad area of produce production. So it's, again, I could give a myriad of examples. But this goes back to the question of why is AI, why is this transition different?

Jason Hiner: Yeah.

Mark Papermaster: It truly is every aspect of our lives that it can bring really game-changing improvements in not only productivity, but just innovative ways to go about what it is you're trying to accomplish.

Jason Hiner: Yeah. All right. I haven't asked any questions about other things you've announced here at Advancing AI. The event that AMD put on here this week, when you talk about all of these things, Helios, this amazing set of technologies that's going to power the most advanced AI, sort of the next breakthroughs in AI. Right? It is going to be sort of the building blocks of a lot of those things. And then you have your CPU, which now is increasingly important for things like AI agents, which you talked about. What's the story that you tell when you talk to people about all of these new things that have been years in the making, as you also mentioned, and they're really just part of, they're a step on a roadmap that goes into the future as well. When you tell the story about all these things that AMD is doing today and putting its energy in, and customers are buying, how do you tell the story? What do these things sort of mean to you and your team and the things that you're building and the way that sort of you all find purpose in what you're doing?

Mark Papermaster: Well, when we talk to customers, we talk to them about how we help them solve their problem. Right? How do we help your business? And, you know, typically for businesses, it's, look, my compute needs are drawing dramatically. And my bill is basically unaffordable. You know, they plot for it and it's not affordable. And that's where AMD is, of course, focused on providing leadership technology, but it's leadership technology that gives you a cost advantage that's energy efficient and lets you run your workload on the best tailored kind of computing, whether that be in the cloud, whether that be on your premise, whether that be your personal AI or an embedded device. And AI compute demands are growing, Jason, so dramatically that that's actually a different story than they've heard before, like, oh, I can optimize what I run where, and, oh, AMD, you're doing it where you don't lock us in. You're giving us a choice of how I can run AI. I don't have to run with one vendor. I've got choice. You don't lock me in. It's an open stack. It's an open ecosystem. And I can run where it makes most sense for the task I have at hand. And that question, you know, that discussion, I should say, resonates very well because we don't force a solution. Like, here's our solution. What's your problem? It goes the opposite way.

Jason Hiner: Yeah. The lock-in thing comes up a lot when we talk to enterprises.

Mark Papermaster: Yeah.

Jason Hiner: You know, that comes up all the time, for sure. So the open ecosystem part of what you do, yes, clearly has some, some resonance there as well. Okay. I want to ask you the same couple of questions I tend to ask everybody in this podcast, especially most of the time I'm talking to leaders, right? Which is that what is your best leadership tip in the age of AI? Because, and this often comes down to even yourself, like, how do you maximize your own time for maximum leverage?

Mark Papermaster: Yes.

Jason Hiner: Because everybody, you know, I think the promise with AI was like, oh, it's going to do a lot of the things we don't want to do for us. And yet at the same time, it makes our time even more sort of critical.

Mark Papermaster: Yeah.

Jason Hiner: And so everybody, every leader I know is grappling with this.

Mark Papermaster: Yes.

Jason Hiner: How do I optimize my time for maximum leverage? So what's your tip for other leaders?

Mark Papermaster: Well, there's the practical day-to-day tip, and that is, you know, the agents available to you today, just, you know, derivatives of OpenClaw and the things that you can just have truly helping your day-to-day, the things that you didn't love that took a bunch of your time. You have to really spend the time to tailor the models of your choice to you. You have to tell it who you are, what you like, how you have to really invest to tailor that model. There, you know, all the model support files that record and keep persistently your style, what you want. So one, I'd say, invest. If you want to use AI to be productive, you've got to actually take a little time to understand it, invest, and really train it to be most productive to you. So that's one, and I've gotten a lot of success with that, and I get a lot of help every day from my agent. I have a personal agent.

Jason Hiner: Amazing.

Mark Papermaster: But I'd say beyond that, when you step back, and when I'm talking to CIOs and heads of technology and engineering, the customers that, many customers I talked to on a weekly basis, I urge them to really step back now in this agentic AI era and reimagine their workflows. And it's hard, because we've all been working years on them, well, here's what I do, and I can tweak this part, and I can get 5% productivity gain. I can get here, I can get another 4% over there, and in the end, I might get 15%, 20% productivity gain.

Jason Hiner: You know which knobs to turn.

Mark Papermaster: You're turning knobs.

Jason Hiner: Yeah.

Mark Papermaster: That remains very, very important. But with agentic AI, you can chain together those point productivity gains. That's great. But you can do much more than that. You can actually reinvent workflows. You can create agents that have subagents, and the sub agents might be ones trained with Jason's kind of knowledge of how to be very, very media-astute with all the demands that go with that. It might have another agent that really addresses, let's say, a very technical piece of what it is you're trying to do. And you then tap those subagent experts, and you can reimagine how it is you're trying to, so there's a major proposal that a business leader needs to get out. And it can tap these combinations of subagent experts and finely tuned models that have your company's knowledge base, and you can reimagine how you can get your output, your product out to market. And it can be faster, but it can also be better because of the reasoning capability. Again, the ability to tap a database and a number of variables that go with that, larger than you and I and other humans can handle. And it reimagines what you can do.

Jason Hiner: Very cool. All right, last question. What's the AI tool that you're using right now that maybe a lot of people maybe don't know that you would recommend people try because it can make a difference for them?

Mark Papermaster: Well, no surprise to any of your listeners. I use our technology. So I'm tapping AMD CPUs, which are, you know, we just announced our new 2-nanometer CPU. So when I run on the cloud, I'm running, you know, on the fastest CPUs, our new Instinct. But for me, as an executive role, I do ship some tasks off to that big computing. But to me, what's really amazing is what you can now do in local AI. What you can do on a PC, we announced today that on a PC with our new Gorgon Point, you can handle up to 300-billion-parameter models. It's amazing.

Jason Hiner: On a PC.

Mark Papermaster: Couldn't even imagine that a year ago.

Jason Hiner: No.

Mark Papermaster: So what you used to have to go pay a lot of money to to run on the cloud, you're now running in your PC. I'm doing that. So I'm, you know, as I think about our strategy, our roadmap, how am I going to take that to the board? I'm in a dialogue at a PC having, I'll say, a partner brainstorm with me. And how do I want to think critically about that? And why is that important? Because you have a better security. It's private, sovereign, fast, you know, no token costs.

Jason Hiner: Exactly.

Mark Papermaster: So I'm using a Ryzen AI or, you know, Ryzen AI Halo and I can literally have this private dialogue and get great insights.

Jason Hiner: Unbelievable. So that little Ryzen box that essentially you could have what is like a team of agents just sitting on your desk. You've already paid for the tokens. You're not, you know, running.

Mark Papermaster: If it's an open weight model. Yes.

Jason Hiner: Yeah.

Mark Papermaster: Yeah. It's your team right in front of you.

Jason Hiner: Very cool. Mark, thank you for the time. It's been a lot of fun. Great conversation.

Mark Papermaster: Thank you, Jason. Really enjoyed it.