Mark Papermaster - CTO at AMD

(Joel Beasley at 00:00:00) Hello, my friends. Today we are talking to Mark Papermaster, the CTO at AMD, and we discuss what it takes to stay ahead of the competition in a hyper-competitive space, a preview of what's to come with their Zen 3 architecture, and how open and candid conversations are the foundation of a healthy culture. All of this right here, right now on the Modern CTO podcast.

(Mark Papermaster at 00:00:24) Here we go.

(Joel Beasley at 00:00:25) This is the Modern CTO podcast. Hello. Hello.

(Mark Papermaster at 00:00:38) Hi, Joel. How are you?

(Joel Beasley at 00:00:39) Fantastic, Mark. How are you doing?

(Mark Papermaster at 00:00:41) I'm doing great.

(Joel Beasley at 00:00:42) This is going to be excellent. We're going to have a lot of fun today.

(Mark Papermaster at 00:00:45) Yeah. Thanks for inviting me.

(Joel Beasley at 00:00:47) Look at that background. Is that the AMD campus right there?

(Mark Papermaster at 00:00:51) It is. I chose that as my virtual background. And honestly, I'm in Austin today, which is where that campus is. We're in Santa Clara. We're in Austin, and we're multinational. But that is what the weather looks like in Austin today. It's a very nice day. You're Bay Area based, Joel?

(Joel Beasley at 00:01:11) I am. Yeah, I'm in Tampa area, but we're actually, yeah, we're actually, well, Tampa Bay. Yeah. But we're actually looking to move out to Dallas Fort Worth, so we're going to be out there in about two weeks.

(Mark Papermaster at 00:01:24) Oh, great. Welcome to Texas. Yeah. Almost.

(Joel Beasley at 00:01:27) Have you been there long?

(Mark Papermaster at 00:01:29) Oh, yeah. I've been away. I've been in the Northeast for a period of years. We just moved back from California. I still split time a bit, but for the last almost three years, we've been mostly back in Texas, and I'm a University of Texas grad. I love the state. I hope you love it.

(Joel Beasley at 00:01:47) Yeah. I have family from Lubbock, and so they talk about it a lot. But I haven't been out there too much. I've traveled through it, but I'm excited. So we're going to be out there for two weeks. We're going to explore, see what's going on. We have two little ones. So it's going to be an exciting adventure for the family. So how long have you been at AMD?

(Mark Papermaster at 00:02:08) I've been at AMD nine years. I've been in the industry over 35 years, but nine years as CTO and running our engineering efforts here.

(Joel Beasley at 00:02:16) What got you involved with technology to begin with?

(Mark Papermaster at 00:02:20) Well, that's a great question. I've always loved technology. I was just always a gadget guy as a kid growing up, and I did a summer job working with actually a cancer researcher who's an electrical engineer and was just showing me all the ways that actually being an engineer, working with the data, you know, helping solve disease. I thought that was fascinating. And that's what I ended up majoring in at school at University of Texas. I was an electrical engineering major, and I've just been so fortunate to work in and around technology my whole career.

(Joel Beasley at 00:02:58) Oh, yeah. So did you ever do anything else with, I mean, I guess you're at AMD now. So you make processors that will then be training algorithms to then help find cures to diseases. So it's kind of coming full circle.

(Mark Papermaster at 00:03:12) It is. I wish I could be solving those types of tough problems, you know, creating new medicines, vaccines for the pandemic and so on. But at least we do play, I think, a very significant role in providing the kind of technology that can speed the efforts of the scientists.

(Joel Beasley at 00:03:30) Yeah. So that gives a nice sense of purpose. Does that come out corporately? Do you guys talk about that to your employees?

(Mark Papermaster at 00:03:38) We do. In fact, we're very active in actually assisting directly scientists, academic institutions. We just donated over 12 petaflops of supercomputing capability to 21 universities and research institutions to speed the research around the pandemic, whether it be vaccines themselves, molecular dynamic studies, even social activity that could affect the spread of the disease. So the computing is going to, I think, be put to excellent use and our employees know that we're very active in supporting that and they know how important our technology is in this regard.

(Joel Beasley at 00:04:20) That's pretty interesting. So I want to make sure I understand that right. So you, these, a lot of computing. I won't remember the exact number. A lot of computing you donated. But are you guys in the business of being able to do that, or is that something that you had to spin up and create for this specific scenario?

(Mark Papermaster at 00:04:41) We're definitely in the business of supercomputers, and that's the engines underneath. So, you know, you think about the high performance products that we do, CPU, GPUs, specialized semi-custom implementations. They span everything from game consoles to new game consoles coming out this fall from Sony and from Microsoft, PCs, servers, all the way, you know, gaming devices that people use, all the way to supercomputers like the most recent announcements from the US Department of Energy where an award was given to AMD for our CPU and GPU and working with Hewlett Packard Enterprise to develop two computers. It'll be the largest in the world starting next year.

(Joel Beasley at 00:05:28) That is exciting. So then will people, today, and correct me if I'm misunderstanding this. So today, can people go like a cloud service and use your supercomputers?

(Mark Papermaster at 00:05:40) They do. Clouds are just now starting to do that. Joel, it used to be just the massive national labs at the various countries that could afford to put up these supercomputers, and it was forward in national interest, but they typically open it up for scientists in those countries or even internationally to get to those supercomputers. But it is very interesting that you say, well, can people just rent time? And that's what hyperscale is, the big companies where there'd be AWS and Microsoft and Google and Oracle, IBM, and so on. And that's just US. Same thing internationally. They're now standing up supercomputers so people can rent time. And so it just makes it easier for scientists, but also students and others to get access to that kind of computing that used to just be at these massive national labs.

(Joel Beasley at 00:06:36) So is it a partnership with these large companies like Amazon Web Services and all of these, or can you actually go to AMD.com and spin up a supercomputer?

(Mark Papermaster at 00:06:47) Yeah. So we're the technology provider. So whether it be those big hyperscalers or whatever we partner with, systems companies to build up those supercomputers. So it's not AMD.com where you can get that supercomputing, but it is as simple as logging into one of those cloud services and just searching for the kind of capability you need and getting an account and away you go.

(Joel Beasley at 00:07:13) That's exciting.

(Mark Papermaster at 00:07:14) Yes. It's truly an exciting development.

(Joel Beasley at 00:07:18) So I read a little bit about you leading the redesign of the engineering process, and I was just curious, you know, open broad question, how do you come into an organization of that size and change it?

(Mark Papermaster at 00:07:31) Well, I've always had a great relationship with AMD over my years. So it's 35 years in the business. I've been at IBM. I've been at Apple and Cisco before nine years ago taking this role as CTO and running our technology and engineering. And AMD always had such a storied history of innovation. But what we found ourselves as I joined, as Lisa Su joined, our CEO, what we found was that what we had to do is marry with that incredibly innovative spirit that was and is stronger than ever at AMD, a different kind of R&D process that would allow AMD to never miss a beat, to be out there every cycle. So every technology has a refresh cycle. It might be annual. It might be every 18 months, where, you know, you've got to come out with that next generation. And so we did have to retool engineering to make sure that it could, you know, that AMD would in fact be on that regular cadence and not have a slip. So it is a big challenge, but a great challenge because what it is, my sense, and certainly what I tried to do with the team was get goal alignment. Got to really get everyone to where they bought into not only the vision of what we need to be, but the key levers that if we could implement these key levers, then it could change the way in which we ran development going forward. So we spent a lot of time upfront on the goal alignment. What are the challenges? Engineering is about problem solving. And what people sometimes miss is sometimes the biggest challenge is to frame the problem correctly so that the thought leaders can get their head around it and get those innovative juices of solving the right thing. And then you have to give the team space. You have to create an environment where they can innovate, an environment where they can collaborate, an environment where they can really communicate without fear. Right? You don't want the messenger being shot. So it was really all of those elements. It was setting the right goals, setting the right processes, even how we organize the teams, Joel. We set up to where we wouldn't just be working on the very next processor, for instance, that we shipped. We'd be working on one generation and the next and then could leapfrog. When one team finishes, you leapfrog. And that way, you ensure a continuity of that innovation, development machine.

(Joel Beasley at 00:10:08) It sounds brilliant. So I was talking with Paul over at Fox. He says he feels that processors are a space that if you stand still, you go backwards very quickly. How do you stay ahead?

(Mark Papermaster at 00:10:22) Yeah. No question. You look at what's going on in processors, the capability is doubling about every two years, about every, I'll call it two and a quarter years. So you can't stand still. You have to have an innovation pipeline, and it's a great question because it used to be that the bulk of where you drove that processor development came from the semiconductor technology. You know, you'd look at that new semiconductor technology, and that alone would allow you to shrink the chip so you could add more transistors. You could be at the same price range. Right? So whatever you know, you could keep your product at about the same price range, and you could really increase the speed at which it ran, the frequency. So everything kind of rose up generation to generation. But it's not that same equation anymore. Transistors are actually becoming more expensive. We're getting down to where the very transistor that makes up the base, the foundation of a microprocessor or a graphics processor is only becoming, you know, several molecules of thickness of what's called the gate, the switch of the transistor. And so as you start getting these molecular limits, it becomes more expensive. Innovation will keep semiconductor technology continuing to go forward. So we'll still get more transistors packed in every generation, but you don't get the same frequency, the same speed increase, and the costs are going up. So it doesn't matter. The customer's demand, as you said, that you get more and more performance every generation. So for those of us in the industry, that just means more innovation. We're innovating with how we design more features to give you more performance on top of that. We're innovating around power management. We just had a huge initiative in AMD to increase our energy efficiency 25 times from 2014 to 2020, and we achieved that goal. I'm very proud of what the team did. So whether it be designed for performance, designed for energy efficiency, you absolutely can't miss a generation. You cannot stand still, and we're incredibly focused on that at AMD.

(Joel Beasley at 00:12:41) I love the way you talk. So much passion.

(Mark Papermaster at 00:12:44) Yeah. Actually, in our business, you have to have a passion. It's very hard what we do, but you'll find, as you talk to those of us in chip design, it is all about passion.

(Joel Beasley at 00:12:57) And then that desire and drive to keep moving forward, you know, what you were talking about, discussing how they get smaller and smaller and smaller, and there's got to be some sort of limit. I think you call it the gate or the switch. There's some sort of physics limit. What happens next? Are there going to be new methods of creating these processors?

(Mark Papermaster at 00:13:18) There are indeed. You know, because I've been around a while, I've already hit about two or three times where someone declared the end of semiconductor advances because we hit a hard wall and, you know, it's just engineers, it's their very makeup. They're psychological makeup. Anytime there's a wall put in front of them, it's like a challenge. It's a, you know, figure out how you climb over the wall. You go around the wall. You find a way to blast right through the wall, but that's what engineers do. And that's what's happening on the semiconductor front. So you're right. We are just getting down to a few molecules thick. But now there's innovations on different types of materials, you know, that you can actually build up that semiconductor. It's moving three-dimensional. It's even, you know, the very structures that are being used are changing. So there just is not a lack of innovation. And so, you know, we see breakthroughs on the horizon.

(Joel Beasley at 00:14:18) That is exciting. Yeah. My mind works like that, too. I see the wall. I'm like, how do I go over it, under it, through it? And then when I interact with my wife, I remember that sometimes they just want to talk about the wall. We're not trying to problem solve the wall. We're just talking about how the wall exists. That's what we need to do right now. So any technologies that you're specifically excited about? If you had to guess, what would come next or what's the most realistic moonshot type deal, what technology do you think that would be?

(Mark Papermaster at 00:14:52) Well, before the moonshot, you're seeing a change afoot right now, and that is something that we pioneered at AMD really ten years ago, and that is starting to put CPUs and GPUs together. So, you know, the GPU, think about, you know, if you're a gamer. You buy your CPU and then you buy your GPU and you pick and choose. What we started doing ten years ago is integrating very tightly the CPU and GPU. And you've seen that in, for instance, you know, the PCs that come out now, you can put in an additional graphics chip. But the performance has gotten so high with the integrated GPU that we have with our CPU that you don't have to. You can get triple-A game titles that play beautifully just in a single chip because it's integrated so well. The functions, you know, that want to be in the CPU or GPU and how you manage and get the best performance wherever the software application or game that you're running wants to be, and that's growing. And so game consoles, of course, deploy that same approach. And now we're moving that same type of what we call heterogeneous computing. You know, it's different.

(Mark Papermaster at 00:16:08) It's not just one type of engine. It's a central processor, a CPU, a graphics processor, GPU. It's other dedicated accelerators and coming together in heterogeneous computing. So you're seeing that in those supercomputers that I just described to you. Now those are built with a CPU and GPU that are designed to work very, very closely together.

(Mark Papermaster at 00:16:28) You know, when you're a programmer, you don't have to think, am I programming for the CPU or GPU and manage where I'm storing the data? It's what's called coherent, meaning it's all sharing the same address where you store the data. So this is programmer speak for making it easier for them to code, leveraging this new type of approach of putting engines together. So that's not even, you know, far future. That's what we're doing now, and it's very, very exciting.

(Mark Papermaster at 00:16:55) And that's going to take us to the next decade of enhancements of these types of heterogeneous computing approaches. And then beyond that, it's going to be more and more what I call accelerators. So what's an accelerator? It works with that general easy-to-use programming device, but things that need incredible computation. Can you send it off to that device?

(Mark Papermaster at 00:17:18) Of course, we're using, you know, GPUs and programmable gate arrays and specialized applications specific ICs for that today. But you'll eventually see quantum computing where, you know, you operate at extremely low temperatures. You have just breakthrough kind of computing that can be an accelerator. So you could be doing your general tasks in the kind of computing we're developing day to day and then send off, you know, the most difficult problems to quantum computing where it can calculate a result and send it back. So, again, Joel, there's just so much innovation coming.

(Joel Beasley at 00:17:57) And I'm such a geek. We're so excited. To be honest with you, the entire company over here, we've got about, you know, ten, fifteen employees, has just been buzzing on our Slack for your arrival. Jake, our producer, is like a huge fan. He builds his own computers, and he always uses the AMD stuff.

(Joel Beasley at 00:18:17) And I told him, I said, look. I know you have a hundred technical questions, like details about product releases. Let's just put that aside and just have a nice conversation with Mark, and then you can send him an email after if you have some really highly detailed questions. But I was kind of curious. Are you guys building any quantum computing, or how do you look at quantum computing?

(Mark Papermaster at 00:18:39) Well, right now, we are building out what I call that heterogeneous computing infrastructure. So the way that we're working with quantum is we want to be what I call the best hub for a quantum accelerator. So we're designing, you know, our computing such that it's very easy to attach. We create many connection lanes so you could connect up to an accelerator like quantum compute. Our whole software approach, Joel, is open source.

(Mark Papermaster at 00:19:09) We want, you know, whether it be academia, whether it's industry, we want to be able to partner with everyone out there to develop a software approach that everyone can innovate on. And we want that to extend, you know, all the way through, as quantum becomes more prevalent in its deployment. So that's how we see ourselves. We'd like to be a glue for collaboration in the industry and have a platform that makes it very, very easy to collaborate.

(Joel Beasley at 00:19:36) That sounds really smart. It reminds me of, like, Elon Musk open sourcing his patents on the batteries so that everyone could move them forward or buy them. It seems like your approach to open source will allow you to remain competitive. And, I mean, also, since it's open source, you could attract talent that way, right, through seeing who contributes and who's doing things. I really like that model.

(Mark Papermaster at 00:19:59) It creates a whole innovation community. It's a proven model. It just works.

(Joel Beasley at 00:20:04) Yeah. Life is better together. Right?

(Mark Papermaster at 00:20:06) Yes. Exactly.

(Joel Beasley at 00:20:08) When you got to do—I happen to have an Xbox. I play maybe thirty minutes a week of video games just to go blow off some steam sometimes. And I was curious when you got to do those partnerships, did you personally get to be involved with, like, meeting with the Xbox teams, or was that just someone else on the team?

(Mark Papermaster at 00:20:28) No. We, you know, as we got the whole dialogue going, I certainly was involved with the leadership with both Microsoft and with Sony. And it was, again, I'll go back to the thing we were just on. It was about, you know, co-innovation and collaboration. You know, what did we need to do? What were the goals? We then created our own group, a separate group that worked with Sony, a separate group that worked with Microsoft, and they, of course, did the engineering. I don't claim anything to do with the amazing engineering that was done across that collaboration because that's done at a very, very detailed level. But the results are phenomenal. When you look at that generation that, you know, was out there with Xbox Series X and the PlayStation 4.

(Mark Papermaster at 00:21:22) And think about gaming when the physics is being calculated real time. It used to be that when you played games and anyone had hair that, you know, the character in your game was going, they typically used to have skull caps. Well, in that generation of gaming, the math, the physics is running. It's calculating what does the hair look like when it's blowing in the wind. And in that generation of games, you see it.

(Mark Papermaster at 00:21:47) Now with this next generation coming out, it's even beyond that. It will include—this is with our Radeon DNA too. It's our latest in graphics processing architecture, and it includes, in addition to even more rapid processing of all that math to give you beautiful imaging, it also has ray tracing right in the chip. So as the light is bouncing around, you know, from the sun and what is it reflecting off of—water or a car in your car chase—it's mimicking the actual light rays and how they behave in real life.

(Mark Papermaster at 00:22:22) It's calculating it, and it's rendering it right back to you. And so it's incredibly lifelike.

(Joel Beasley at 00:22:28) I saw a demo of Unreal Engine, like their next version coming out on a YouTube video, and they were talking about all those points and polygons. And I'm not an engineer over there. My experience is more in the B2B business logic space, but I still can geek out over there. And I was blown away by all the points. They were talking about like billions of points and they were doing the dynamic sun and the hair and the wind and showing all of the environmental changes in real time. And that stuff's only possible with, you know, amazing chips, right, behind them.

(Mark Papermaster at 00:23:01) Absolutely. No. It's tremendous processing power underneath.

(Joel Beasley at 00:23:06) Can you give me—this might be a tough one. I don't know. But can you give me, like, explain to me as if I'm a small child, like, exactly what a processor is or what it does?

(Mark Papermaster at 00:23:18) Sure. So what you're doing, all of us think about processing from the standpoint of the application we run. You're on a smartphone. You're just running an app. You're on your PC, and you're doing a video conference like we're doing today.

(Mark Papermaster at 00:23:37) But, you know, you're playing a game. You're creating a document. All of that is a program that you're running. So someone created that application. They wrote the code.

(Mark Papermaster at 00:23:48) They created that application. They probably did it in a very high level programming language like C or C++, or maybe they use, you know, Python. That's becoming, you know, very popular. A number of new languages coming up, but they wrote at the high level because that's your most productive way to create an application. What we do at AMD in our processors is we make it easy to translate from that program that you wrote into the best possible performance and the best possible experience.

(Mark Papermaster at 00:24:20) So we think about how do you take that program, break it down into low level instructions we run on our processor. That's called a compiler. So it takes the program. It's compiled. It runs on our processor, and then we do the lifting from there.

(Mark Papermaster at 00:24:36) So we take those instructions. We optimize. We run it as fast as possible. Now does it run in a CPU? Does it run in a GPU?

(Mark Papermaster at 00:24:43) Make sure it's optimized for both. Get the answer back as quickly as possible. It has to be 100% correct. We spend an amazing amount of money and focus and engineering innovation to make sure that every chip we go out—every chip that goes out—is exactly right and performs those calculations as expected and that it delivers the best experience. That experience would be, you know, we have the whole set of coding to send it to your display.

(Mark Papermaster at 00:25:14) We have the things we've done like FreeSync such that as you, if you're watching a game, you don't get any tearing of the image, any stuttering going on. If it's, you know, running on the computer that we've—that's just a pure calculation, that's Excel—typically the best experience there is to get you that result with the highest possible performance. So, you know, we think about our role at AMD and other processor vendors. We're that engine under the covers so that when you're running that application, you just love it. And you're really pleased if you bought a new product from a previous generation, that it's just faster.

(Mark Papermaster at 00:25:58) If it has graphics, it's just, you know, much more lifelike going forward. And so that's our role in the food chain is to provide better and better computing and give people a better experience.

(Joel Beasley at 00:26:13) And then you guys have a new architecture coming out this month. Right?

(Mark Papermaster at 00:26:17) We do. We've got two products we're launching, one later this week, a new CPU architecture. We have a family of high performance CPUs called Zen. Part of the whole turnaround of AMD is rearchitecting our development process, rearchitecting our CPUs to be the highest of performance. And this is the third generation of Zen coming out, and the designers—I couldn't be more proud of them. They just made changes around almost every aspect of the computing to make it, from my previous comment, run better, run faster, give a better experience, be a better gaming processor.

(Mark Papermaster at 00:26:54) So we've got more details on that coming out this week, and that architecture could be coming out in leadership Ryzen desktop processors this fall. And then on the graphics side, it is our second generation of Radeon DNA. That's the graphics architecture that will be coming out in the new PlayStation, the new Microsoft Xbox Series X, this fall. And they'll be coming out in our next Radeon graphics product that we'll be announcing on October 28. So two big announcements coming up this month.

(Joel Beasley at 00:27:34) Who picks the names for these products like Radeon?

(Mark Papermaster at 00:27:39) You know, that's a great point. I love watching that name process. It's really—I personally just love the process. I'm never the one that's trying to pick the name. I like it best when the team picks it.

(Mark Papermaster at 00:27:54) So you think about Zen, our whole family of processors. That was our code name. Our internal code name before, you know, not a single marketing person was involved, and the team named it Zen. Our lead architect actually named it Zen, and it stuck. Everyone loved it.

(Mark Papermaster at 00:28:11) And so by the time the marketing team came along, they said, we're going with Zen, and then we played off of that to create Ryzen, which is our brand name there. And likewise, when you look at EPYC for servers, when you look at Radeon and our products there, you know, it's generally a mix of typically the engineers being engaged. Look at our Ryzen Threadripper. They call it Threadripper because it's just such a beast. That was, you know, just 100% named by the engineering team.

(Mark Papermaster at 00:28:40) So it's always very fun watching the naming process play out.

(Joel Beasley at 00:28:46) I love the strong engineering culture over there. Sounds like you guys have a great group of people together.

(Mark Papermaster at 00:28:53) We're an engineering company at heart, and everyone just, again, they love what they do. And I think it's actually one of our secrets at AMD. It is about our culture. It is that people really love the innovation process here and work very, very well together.

(Joel Beasley at 00:29:11) Yeah. It's like a competitive advantage. Right? Definitely. So in this market, I have another guest question, and this is from a financial person. I don't know why I said it like that. It's like, oh, it's from an investor person. I'm not going to put their name. Sorry, Nigel. Yeah.

(Joel Beasley at 00:29:29) So Nigel from T. Rowe Price, he says that the industry looks like a classic race to zero and a commoditized space. But at the same time, there are a lot of niche uses where the margins for the processors and innovation will remain high. Like, how do you think about that?

(Mark Papermaster at 00:29:48) Well, there's always a race to commoditization. So, you just look at any industry. There's a desire for commoditization because it makes whatever that product is more available to people. So, you know, we are about trying to get computing to the masses. We want to get more efficiency of our computing.

(Mark Papermaster at 00:30:11) But put your financial cap on, and, of course, you need to return to shareholders. You need to have a sound financial basis. We spend hundreds and hundreds of millions of dollars each year in terms of our R&D expense. And so what we do to make sure that we bring the financial return to our shareholders is make sure that as we're providing value, that it is differentiated. Rather than pure commoditization, although we've done everything we can to make it affordable, we need our product better. We want to have the highest performance, the best ease of use, the best ability to integrate our technology in tailored solutions like those game consoles I described.

(Mark Papermaster at 00:31:04) Those are all things that allow us to leverage a base commodity like an x86 processor, but the way we implement it, have a command value. The same thing with our graphics processing. It's got to bring a value that people are willing to pay for. So that's how you differentiate yourself when you compete and what certainly is a race to commoditization.

(Joel Beasley at 00:31:31) I'm glad I asked that question. I debated in the prep call for this, like, whether I was going to ask that or not. But then I was like, you know what? This guy, so experienced and so professional. I bet you he's just going to knock it out of the park. You did. Thank you.

(Mark Papermaster at 00:31:47) All right. I love talking about, you know, this whole area. So you're going to be hard pressed, Joel, to find a question that I'm not going to enjoy chatting with you about.

(Joel Beasley at 00:31:58) I don't want to. I just want to have a great conversation. So what is, like, the biggest problem slash opportunity that exists today in the processor space?

(Mark Papermaster at 00:32:12) Well, you know, I think one of the big challenges goes back to what I said earlier. The fact that it's getting harder to just rely on a process technology and have some design innovation. The degree of design innovation to keep processing going forward at the historical rate that I described, doubling about every two years, two-plus years, it does mean that you're having to specialize more on what is that application that's being run. So there's more specialization if it is gaming. There's more specialization if it's supercomputing and artificial intelligence.

(Mark Papermaster at 00:32:55) There's more specialization if it's not necessarily in a data center, but what's called the edge of the data center, right, you know, at the base of a cell tower with 5G emerging and about to create this massive pipeline of data flow across the world. And onwards into embedded devices, you already see that with all these Internet of Things. You know, that scenario, we've chose not to play, is, let's say, a processor on a thermostat that's on the wall or embedded in your refrigerator. That we believe, you know, is in fact already highly commoditized.

(Mark Papermaster at 00:33:33) But the challenge is that what you're seeing is a development where more and more specialized computing is needed.

(Joel Beasley at 00:33:42) I like that. Right? And then that way because it's all about how, how you mentioned earlier, market and value and being useful to others is the way you grow a business. Right? And so your expertise and being able to do things like dedicated team to help make a processor specialized for, like, a health care specific use.

(Joel Beasley at 00:34:02) Right? That's going to be very, very useful.

(Mark Papermaster at 00:34:06) Yeah. We did that. The other thing I probably didn't mention that I should have is we also designed in a modularity with what we do. So as you need to tailor, as you need to specialize, we've, you know, changed our engineering approach to be highly modular. So it allows us to customize for a given compute application without having to start at ground zero and construct it, you know, transistor by transistor as a new and different product.

(Mark Papermaster at 00:34:32) We can leverage our work that we've done.

(Joel Beasley at 00:34:35) I love it. Can we talk a little bit about, like, professional development?

(Mark Papermaster at 00:34:40) You bet.

(Joel Beasley at 00:34:41) Yeah. So you were at IBM for, like, twenty-six years. Is that correct?

(Mark Papermaster at 00:34:46) That's right.

(Joel Beasley at 00:34:47) All right. So I was trying to come up with, like, the most valuable thing to ask because twenty-six years is a long time. And so I want to know, what is the one takeaway you have from that entire experience?

(Mark Papermaster at 00:34:58) Well, you know, IBM at that time was, I think, one of the storied companies in terms of driving a very rigorous employee development and training. And so, you know, I'm very appreciative of the kind of training I had. Early on, I had a graduate degree I did while I was working. They were very supportive to get a graduate degree. I chose to go into technology management and was able to get, you know, not an MBA, but MBA training, many courses taught by MBA professors, management training along the way.

(Mark Papermaster at 00:35:33) So IBM is an integrated developer from chips all the way to end products. I think, you know, IBM for me was just a great way. I learned a lot. I think I was, I feel proud I was able to contribute a lot back. The products did very, very well during the tenure I was there.

(Mark Papermaster at 00:35:52) So it's, you know, IBM is clearly a storied company when you look in the annals of computer development.

(Joel Beasley at 00:35:59) Yeah. When you were there, so many things were happening. You know, two of them that I remember pretty well were, like, the browser wars and then the processors just, I remember driving over this bridge and there's this billboard, and it felt like every time we drove over that bridge, it would be like a new version of the processor being up on the screen. I was like, oh, what number is it today? And so it's very competitive.

(Joel Beasley at 00:36:25) Do you have any good memories or stories from either the browser wars or the competition in the processor space from being at IBM?

(Mark Papermaster at 00:36:34) Well, back up before the browser wars, let's just talk about the advent of the PC because that really shook up the industry. And IBM had been known for proprietary approaches and, looking at the PC space, to really open it up, actually published the specifications of, you know, how to build a PC, and it just spawned an entire industry. Many of us working in the development of computers today wouldn't just, would not have such a robust field to be able to work in had that PC not been opened up to create a standard. And then, again, the idea was let performance, let differentiation win out, very similar to, you know, to our thinking that I described to you at AMD. And you look at, there's always been, I'll say, wars.

(Mark Papermaster at 00:37:26) You know, there was VHS versus Betamax, which many of the listeners here won't even know what I'm talking about, but, you know, the old videotape standards. There was operating system standards. I mean, you know, Windows and macOS, Linux are so prevalent today. But, you know, think about the whole myriad of other operating systems that are still legacy operating systems or very active operating systems today but, you know, have had various battles along the way. And browsers, just another example of that.

(Mark Papermaster at 00:37:58) I mean, such development of, you know, with the very start with Mozilla and early browsers, and now it's a very robust competition. So I think it's just symptomatic of the industry. Competition is good. And the best way to have competition is to have an open framework, to have a set of standards that we can compete with each other, but yet still work together. And when you have that, you create a very vibrant and robust technology ecosystem. The end users, the customers, they benefit.

(Mark Papermaster at 00:38:38) So we're big believers of that at AMD. I just think it's just proven to be the path to the most impactful innovation at the end of the day.

(Joel Beasley at 00:38:49) And that's something that exists in the industry right now between you and the competing chip manufacturers? They all use a similar standard?

(Mark Papermaster at 00:38:58) In many areas, yes. And in many areas, there's a lot of hard work required for us to be able to land on those standards. You look at how, you know, you just look at a PC. You plug a connector and you use a USB. That's a standard, and that's very actively worked on.

(Mark Papermaster at 00:39:18) Or the other big interconnect standard is PCIe. And so, you know, those type of interconnect standards are evolving, and I think the industry is working very, very well. The next big challenge I see is in software, particularly around machine learning, and landing on standards there that can create an open ecosystem. This is something we're very committed to. We've opened up our software stack around machine learning.

(Mark Papermaster at 00:39:42) Others are proprietary that are out there. But you're starting to see more collaboration on this front, and we're certainly very, very active in promoting that. And then another area, Joel, that I would say is security. Everybody has to trust that whatever computer you're working on, that your data is safeguarded. And so I'm starting to see a lot more collaboration in the industry on security and making sure that all of us can have full trust in the computing.

(Mark Papermaster at 00:40:11) We just, you know, rolled out a set of features with our servers that can allow encryption where the user controls. They have the keys. They control unencrypting the data. And, in fact, Google Cloud just rolled out just a few weeks ago what they call confidential computing that uses those keys and protects that. VMware just announced support for encryption capability last week, and others are coming and partnering with us as well.

(Mark Papermaster at 00:40:41) So, yeah, there's a number of areas for collaboration that we can build on success stories we have and do even more.

(Joel Beasley at 00:40:50) Are there big companies like this in other parts of the world that we don't know the brand of, or is this just what is available in the world? Like, when we talk about, like, the AMDs or, like, the Intels and, like, the different, is that, like, a global thing? Or if I went over to, like, a country that maybe doesn't have English as a primary language, they would have an entirely different similarly sized processor manufacturer?

(Mark Papermaster at 00:41:17) Processing development, processor development is international. So we are a multinational company. Our competitors are multinational. We have engineers across the world, sales and distribution across the world.

(Mark Papermaster at 00:41:31) So, you know, computing, commodity computing like we do, x86 GPU, you know, the world needs computing and that's who we service.

(Joel Beasley at 00:41:41) Nice. So back to your time at IBM. When you were there twenty-six years, then you moved over. You could have done anything you wanted in the world. Like, anybody would have hired you. They'd be like, come lead us, Mark.

(Joel Beasley at 00:41:53) Right? So why did you choose AMD, and what was the story like about, like, how you met them and how that all went down?

(Mark Papermaster at 00:42:02) Well, I had known AMD for years because I had, at IBM, I'd used their chip technology and used it very successfully. And I always really admired the innovation of the engineers. They're just fantastic. And the culture. Culture was one of engineering collaboration.

(Mark Papermaster at 00:42:22) And so I got a call from a gentleman who had been the head of R&D during my time at IBM, and he was on the board at AMD. And he said, Mark, this is an opportunity of a lifetime to have an impact in the industry. And I said, I don't know. I'm busy. I've got these other things going on.

(Mark Papermaster at 00:42:45) He said, this was Nick Donofrio. Nick said, Mark, if you don't take this role, CTO and run the technology and engineering, I might need to come out of retirement because this is that kind of opportunity. And I have such immense respect for Nick. I said, Nick, you don't have to come out of retirement. I'm going to take this role, and it has been phenomenal.

(Mark Papermaster at 00:43:06) I've probably never worked harder and never had more fun in my life.

(Joel Beasley at 00:43:11) Oh, that's awesome. And I think that's useful when you share stories like that because other people are, you know, at companies for a long time and they're curious about, like, what the next move is for them, and I find that from the feedback I get from the audience that just stories of how things go down tend to be some of the best way to help because that's what we're here to do, like, help the next generation of leaders not make the same mistakes. Right? Do better than we did.

(Joel Beasley at 00:43:36) Yeah. So—

(Mark Papermaster at 00:43:37) It's a responsibility all of us have to, you know, to mentor others and pass it on.

(Joel Beasley at 00:43:43) What are some of the lessons that you've learned multiple times over?

(Mark Papermaster at 00:43:50) Well, probably the biggest lesson is just communication and transparency. I mean, we stress it amongst our teams. I stress it. My leaders and my team stress it. When you don't have just open and candid communications, you just aren't working on the right things.

(Mark Papermaster at 00:44:10) You know, people, anyone that they think that they can sweep a problem under the mat and it goes away, it's just wrong. Problem festers. I mean, so you really, you know, I think one of the things that I always stress is, and I touched on a little bit earlier, create the right culture. Create a culture for openness. That in turn creates a culture for better problem solving.

(Mark Papermaster at 00:44:32) That in turn creates a culture for better innovation and risk taking and risk management. Make sure you take the right amount of risk, not too much risk. It just all builds on itself, but it starts with that foundation of phenomenal communication and collaboration.

(Joel Beasley at 00:44:52) I like that. Communication is something that you work at. I mean, it's a hard muscle to keep really strong. Like, it seems like you're always working at it. Right?

(Mark Papermaster at 00:45:02) Well, Joel, you're pretty darn good at it, given, you know, given when I look at what you do. Congrats.

(Joel Beasley at 00:45:09) Thank you so much for the kind words. I really appreciate it. We have actually about ten minutes left here, and I was just curious. One of the things that people have been asking us, and we have not done this yet, is they've been asking us to put together, like, small groups, right, of people so they can talk because they said, like, fifty is too many when they go to a conference or whatever. They kind of want a group of, like, you know, eight or so, maybe seven, eight people to get together and, like, talk about whatever the topic of the group is, whether it's, you know, hypergrowth or just being a CTO.

(Joel Beasley at 00:45:43) And so now I'm kind of soliciting feedback from people because I've never been a part of one of those groups. Like, I speak a lot, so I'll go speak at groups and things like that, but I've never participated in one of these small groups or mastermind type things. Have you ever done so, and do you have any experience there?

(Mark Papermaster at 00:46:00) Yeah. Let me give you a couple comments. First of all, I do think that's extremely important. So first, I'll say, what do we do internally at AMD, particularly with COVID-19 and many working from home, and most all working from home. We have quite a bit of our locations are still in a work from home mode.

(Mark Papermaster at 00:46:18) And, you know, what we found is it caused us to even expand what we used to do before. We used to have small meetings where, you know, a cohort where a small group of people could get together, talk about what's going on. Hey. What challenge are you facing? Well, how did you solve it?

(Mark Papermaster at 00:46:35) And you hear how another person may have an idea about solving that problem. So just in any company in the industry, I'm sure most are doing what we are. They're really supporting that and getting virtual, like the mobile meeting today, getting many small cohorts where it's easy to get that kind of dialogue going on. But we used to all be able to network and go to an industry event and have those kind of side discussions, you know, in a hallway at an industry get-together. And so, more than ever, I do think there's a need.

(Mark Papermaster at 00:47:08) I have participated in several where we've, you know, it's obviously pre-COVID, but now post-COVID, it's been a format like here where we've had, you know, a number of CTOs. We've done this, you know, CTOs, CIOs, or, you know, technology folks, and just create a forum for discussion. I personally found it very beneficial. So I recommend, I do recommend it.

(Joel Beasley at 00:47:31) So, yeah. Because we're all about what's useful to people because that's our mission. So when you said CTOs together, is that, like, within AMD, like, different business unit CTOs, or have you been a part of groups that are, like, another type of group?

(Mark Papermaster at 00:47:45) Yeah. No. I've done both. So I did give you that example internally, but what I was referring to in the last comment was externally. So I have participated in some CTO roundtables, some CIO roundtables.

(Mark Papermaster at 00:47:59) And, again, it was good to hear how other people were dealing with the challenges with COVID-19 environment and just other industry challenges.

(Joel Beasley at 00:48:09) Do you have, and we can cut this part out if you'd like. But do you have any favorite ones or ones I should look into? And do you remember which ones you were a part of that I could go check out?

(Mark Papermaster at 00:48:19) Not anyone specifically. I mean, there's different CTO forms that are out there. I think they're all very good. But yeah, offline I'll shoot you—let me get you the connection points, and I'll shoot it over to you.

(Joel Beasley at 00:48:34) Yeah, I don't want to give you too much work or to-do list. You're a guest here. But if it comes up and it's easy, that would be great because—

(Mark Papermaster at 00:48:41) Sure.

(Joel Beasley at 00:48:41) Just I'm curious. I would like to participate in one because if people are asking me to do one, I'm like, well, let me go participate in one first and see what it's like.

(Mark Papermaster at 00:48:51) Right? That's a great idea.

(Joel Beasley at 00:48:54) Now is there anything—we've got about six minutes left. Is there anything that we didn't get to that you want to get out there into the world?

(Mark Papermaster at 00:49:01) Well, I think the thing I would just challenge all the listeners to do is, you know, just think about the dialogue that's occurring. I mean, it's a turbulent time in our society right now. And, you know, we sometimes—your audience is a technology audience, and so we can say, well, that's not us. We're not having to get in the middle of that. But I think all of us, you know, it's a time where our technology is more important than ever.

(Mark Papermaster at 00:49:32) We are enabling people to take on challenges like COVID-19. We're enabling people, whether it be in their work or their play, to be more productive and more effective. And I would—you know, I did spend some time with you today, Joel, just talking about, I think, aspects that often go untouched, and that is the very culture that each of us develop in our workspace. And, you know, what are each of us doing to create a workplace that's fair, that's diverse, and that just values everyone's opinions? I think those are really key elements of a successful and thriving technology and engineering culture.

(Mark Papermaster at 00:50:16) And so that would be, you know, if I was to leave one challenge to your listeners, that would be it.

(Joel Beasley at 00:50:22) Oh, awesome. So well spoken. What a professional. I was so excited to have you on, and I'm even more excited to have you on again next year. We can catch up.

(Joel Beasley at 00:50:32) So we did it, my friend. We made a podcast. How do you feel?

(Mark Papermaster at 00:50:35) I love it, Joel. It was just a fun conversation with you.

(Joel Beasley at 00:50:38) Thank you so much for coming on, Mark. We'll talk soon. Thanks. Take care. Bye.

(Joel Beasley at 00:50:45) Thank you so much for listening. And if you found this episode useful, please share it with a friend or colleague who you think would get value from it. And if you have topics that you'd like to hear discussed on the podcast, either add me on LinkedIn or send me an email, [email protected]. Every time I get an email or LinkedIn message, it absolutely makes my day and inspires me to keep going.