Rene Haas in Conversation with Top Leaders

Arm CEO, Rene Haas, takes you behind the boardroom door with technology’s most inspiring leaders. Tech Unheard is a lively podcast series that lets you listen in on one-on-one conversations with industry leaders as they discuss everything from the potential of artificial general intelligence (AGI) to keynote nerves.
Rene and his guests explore the drivers behind each leader’s path and analyze the most pressing trends in their space — all while sharing a few entertaining anecdotes of success and failure along the way.

Chris Miller: Explaining Why Some Countries Fail to Capitalize on Technological Progress
Join Chris Miller, author of the book, Chip War: The Fight for the World’s Most Critical Technology, and Arm CEO Rene Haas as they discuss the role of the semiconductor industry in the story of economic globalization and the challenges facing governments around the world.
Here, Chris shares the genesis for the book with Rene, and gives us fascinating insights into the past, present, and future of the increasingly pervasive computer chip.
[00:00:00] Rene Haas: Welcome to Tech Unheard, the podcast that takes you behind the scenes of the most exciting developments in technology. I’m Rene Haas, your host and CEO of Arm. Today, I’m joined by Chris Miller, Economic Historian at the Fletcher School at Tufts University and Fellow at the American Enterprise Institute in Washington.
Chris is the author of several books, including most recently Chip War: The Fight for the World’s Most Critical Technology. He sat down with me to talk about his work as a leading academic in the semiconductor industry.
Chris, thanks so much for joining me.
[00:00:39] Chris Miller: Good to see you, and thanks for having me.
[00:00:41] Rene Haas: Oh, my pleasure. We are in incredibly interesting times here, January of 2025, as we get into the new year and everything associated with new administration. There’s a lot of things that we can cover that’s going to impact our industry, but you and I connected a few years ago when you were writing this book, Chip War, which is a magnificent piece of work, and we’ll talk more about that. But I’d like to maybe start, Chris, and just maybe have you give your background, how you got into the field that you did. However, you want to tell the story.
[00:01:13] Chris Miller: Fantastic. Well, as you know, I have no background in semiconductors or anything related to them. I’m a historian by training, studied history in college, decided that I wanted to pursue a career as a historian and so did a PhD in economic history and in particular in Russian economic history, which surprises a lot of people because there aren’t that many connections between Russian economic history and the semiconductor industry. But one of the questions that I got really interested in was what explains why some countries, despite having brilliant technologists and scientists and Nobel Prize winners, seem to have a lot of the key ingredients for technological progress but nevertheless failed to actually capitalize on them. That was the fate of Russia over the last couple of decades. And so I got interested in, well, why was Russia, despite having a lot of the preconditions, not nearly as good at computing as you might think. And that’s how I first got interested in why the US has an extraordinary semiconductor industry and Russia failed to develop one despite its extraordinary scientific and technological talent.
[00:02:19] Rene Haas: And why history? Why Russia? Were you, when you were growing up, enamored with going to museums and understanding history? And obviously I don’t have a degree in history, but I find that if you look back in history, it teaches you almost everything about the present and the future. But what was the catalyst for you? And what kind of got your curiosity into all that?
[00:02:37] Chris Miller: You know, I think it was, it was similar, desire to understand how the world came to be the way it is. And, you know, I’d studied a bit of economics and a bit of social sciences and found all those interesting, but the best lens for really making sense of the world actually seemed to be to understand how it came to be. And so to me, history was both fascinating in its own right, but also I think really illuminating for making sense of the world as it is today.
[00:03:03] Rene Haas: And the angle in Russia, was it USSR or was it Peter the Great? Or how far back into Russian history and economics?
[00:03:14] Chris Miller: You know, I started by getting very interested in the history of the Soviet Union. I was born in 1987, so I don’t really have any memories of when the Soviet Union actually existed. And when I started studying it, it seemed like this extraordinary alternative universe, with a totally different economic and social structure, which made no sense, but nevertheless existed for a full 70 years, an entire human lifespan in a country. And so that’s what first got me curious as to how did this country and this system work. But then eventually I ended up writing both one book on the Soviet Union and one book that stretches all the way back to Peter the Great and spent a number of years living in Russia as well and still have great affection for the entire history and culture of the country.
[00:03:58] Rene Haas: Oh my gosh. So do you speak Russian?
[00:03:59] Chris Miller: I do. Yes. Yeah. I spent a lot of time there before the political dynamics changed for the worse the last couple of years.
[00:04:06] Rene Haas: And I remember, you know, you and I were chatting in fact when you wrote the book, and I remember at the time a story internally, that I got an email, a blind email from a gentleman who was a history professor who wanted to talk to me. And I was thinking to myself, why does a history professor want to talk to the CEO of Arm. But I found the story of how you, maybe stumbled is not the right word but I’ll use it anyway, how you stumbled into telling the story that you did, because my understanding was that was not originally the story that you were intending to write, i.e. the history and story of the semiconductor industry.
[00:04:37] Chris Miller: That’s right. I really did start with the origins of the Russian semiconductor industry, which I thought was curious insofar as it wasn’t developed at all. Despite that, you know, just a couple of years after the first integrated circuits were invented in the US the Russians, or the Soviets at the time, were trying to build their own. And so the first question was, well, why did the US succeed and Russia fail? But as I was doing that research, I came to realize that although I sort of knew chips were everywhere in my phone and in my PC, I, like I think most people, hadn’t really realized just how pervasive they’d come across our entire lives. And I also hadn’t realized the extent to which the story of economic globalization, which everyone sort of knew, was in large part driven by the chip industry and how semiconductors were probably the best case study and how a product can be designed in one company and manufactured in a second and assembled in the third using chemicals from a fourth. There was really no more complex production process than semiconductors. And so you combine the sort of question of the rise of the modern tech sector with the rise of globalization. You couldn’t understand any of that without really putting semiconductors at the center of your analysis. And like most people, I knew what chips were vaguely, but I’d never really thought about them as being important. And yet, as I looked at the world, I realized I couldn’t really make sense of anything unless I understood better how this industry functions.
[00:06:02] Rene Haas: What were some of the things as you got into your research? And as I said, the research you did on this book was just magnificent. I’ve talked to a number of colleagues and I started in the semiconductor industry in the middle eighties, so before you were born. And colleagues of mine reading the book, we just felt like, my gosh, we’re reading the story of our careers in terms of how you just cover the entire landscape. But what were some of the surprises as you started to do your research and talk to the key influencers in the industry, what were some of the things that were most surprising to you?
[00:06:32] Chris Miller: Well, I think the first surprise really was the pervasiveness of semiconductors. You know, I, as I mentioned, I knew there were chips in computers and phones, but I think like most people, I hadn’t really come to terms with the fact that there were hundreds of chips in a typical car, for example, or in basically every electronic around my home from my fridge to my microwave, there were chips. Or that as you connected more devices together, that was all driven by improvements in semiconductors. I’d sort of thought of all those as discrete parts of the economy, whereas in reality, it’s chips that unite them all together. So that was one big surprise, just the centrality of semiconductors and enabling every sector of the economy to grow. The second surprise was looking at the way in which no one, not a single country in the world was anywhere close to self-sufficient and producing their own semiconductors. And you had these extraordinary value chains stretching across every major economy. And it was this collective effort that made possible the innovations that semiconductors required and that they in turn enabled. That was a surprise to me. And the third surprise was something that I’d been aware of through kind of discussions of Moore’s law, but had never really thought about, and I think most people haven’t really thought about what it means to have exponential growth over half a century. And that dynamic, when you compare it to any other segment of the economy, is totally extraordinary. And thinking through what that has enabled, I think was the third surprise. I sort of knew about it, but I never really thought about it. And when you do start to think about it, it really is a mind boggling in terms of the rate of progress.
[00:08:10] Rene Haas: Now there was a point in time where the US was quite vertically integrated as far as semiconductors go. And I, again, I started my career back at TI and back in the day TI had fabs all over Texas. There were packaging facilities in Texas. The test equipment that we used at TI to test chips were based in the United States. Was it just economics, do you think that pushed them offshore, and them not being TI, but the whole industry? And, or, if you look back in terms of the decision that drove that, could the US have done anything different relative to keeping that capability inside the United States?
[00:08:49] Chris Miller: You know, I think one of the most fun parts of my research was interviewing folks from companies like TI who were working there even before you in the fifties and sixties. I had a chance to speak to some people who were in the really early stages of the industry. And as you allude to, they were producing not just their own test equipment, but their own materials, their own wafers, their own chemicals. Everything was done in house and you had to do it like that at the start because that was, there were no suppliers. But one of the things that I think I realized was that as technology got more complex, it just became impossible for one company to specialize in ultra purified chemicals and ultra complex software tools and ultra capable lithography systems. You had to have the value chain split out into different companies so that you could have the specialization that made technological progress possible. So, I think that dynamic was certainly inevitable. The fact you have not just integrated but separate suppliers for different types of components and processes. I think it was also to some degree inevitable and in a lot of ways a good thing that you had a integrated supply chain that stretched across all the world’s advanced economies because it meant that you could defray capital costs of investments, not just in the US, but also globally. And so it was a good thing you were able to access the Japanese market in European markets and Asian markets, etcetera as well. And I think the third thing was that as the technology became more specialized and you know, I think if you look at lithography, which I spent a lot of time studying there’s great examples of this, you really needed the combined expertise of the best optical experts in Germany, plus the best chemical experts in Japan. You needed to tap into that global talent base. And so that was also a key learning that I took away. It’s easy to, I think, have a simple view that we wish it hadn’t become so globalized because it becomes harder to control. And I understand that impulse, but I think you’d also have lost a lot of technological progress had it just been stuck within national boundaries.
[00:10:44] Rene Haas: No, I think that’s completely right and you stated very well, the sheer complexity of the shrinking of the transistors and the pace at which we were shrinking them required global innovation in terms of contributing to all that. And whether it was ASML or applied materials or Tokyo Electron, you needed companies across the planet and the globe. In your research, what I remember going back down history lane here, and I was again, the TI in the eighties and at that time, Japan Inc. was really starting to become hugely influential. When I was at TI, TI was the number one semiconductor company in the world by revenue. By the time I left, I think it was Toshiba, NEC, and Hitachi that were the number one, two, and three. This is before Intel started to catch momentum. Why did Japan lose the magic? There was a time where the US was looking at all kinds of tariffs and restrictions and anti-dumping with the Japanese. Why do you think the Japanese lost the recipe?
[00:11:42] Chris Miller: You know, I think there were two challenges that Japan faced in the 1980s that didn’t really become visible until the end of the decade or even the early 1990s. And the first was that many of the key Japanese firms, especially the big conglomerates, were making investment decisions based less on profitability and more on market share. And so if you measure by revenue, certainly it’s true, they were among the largest in terms of revenue. If you measure by profitability, that wasn’t always the case. And the fact that they were these big conglomerates that were closely tied with the banks that were funding their capital expansion plans meant that they didn’t face the same market pressures as US firms, which seemed like an asset and was an asset in terms of market share, but wasn’t an asset in terms of profitability. And so the moment Japanese banks started pulling back and companies had to look towards capital markets to fund themselves, they realized that they weren’t actually profitable enough to raise money in capital markets. So that was something that was a surprising vulnerability relative to what everyone thought it was in the 1980s. So that was something that was a surprising vulnerability relative to what everyone thought it was in the 1980s. So that was one reason. The second reason was that Japanese semiconductor firms were generally, especially the big conglomerates, because they were vertically integrated to a much greater degree, they were more inward looking. And again, I think that seemed like an asset at times, but it ended up being a vulnerability because they were less in tune to the ways that technology was changing. And you mentioned Intel, which really was able to capitalize on the emergence of the PC years ahead of any Japanese competitors, because they were looking at a broader ecosystem than the Japanese firms were looking at.
[00:13:12] Rene Haas: Yeah, I know. I feel like my tombstone is full of experiences of companies that were going through. Then I was with NEC Semiconductor during the middle of 1990s. And to your point, I think that’s exactly what hurt these companies was the vertical integration and insular viewpoint. Because when, when the internet in quote happened and suddenly the world was flat and everything was open. Suddenly it was very, very difficult for the Japanese to move quickly. And, obviously then as your point to both the combination of the PC and the internet, suddenly the world had completely changed. And then we kind of bridging onto that, should nations be very, very prescriptive in terms of taking care of local manufacturing? There’s an Intel question there, obviously relative to the US, but it’s even more broadly, whether it’s automobiles or other critical industries. Do you think semiconductors, when you look at their importance and criticality, that policy needs to be shaped, there’s CHIPS Act, but maybe even more broadly, something even more comprehensive?
[00:14:07] Chris Miller: So I think if you go back to where we started the conversation about the benefits of having internationally integrated supply chains, that illustrates the risks of being solely focused on a national manufacturing base. Now, you know, if you’re the United States, you’re the world’s largest economy. And so you’ve got the most scope, I think, to try to focus on a manufacturing base that’s domestic relative to smaller countries. But even still, I think there are real, certainly costs and also risks involved of focusing solely on domestic manufacturing. I think for, you know, certain government applications, it’s understandable governments want that, but that’s a small share of the market, obviously. And, and that’s, you know, why I think, I do give some credit to some of the policymakers in places like the US and Japan and Europe, who have been trying on the one hand to build up manufacturing in general in Western countries, but also not solely focused on their own country and still recognizing the importance of components and materials, et cetera, being traded between countries. I don’t think any of them are solely focused on their own manufacturing base and not recognizing kind of the importance of these broader international linkages. But I do think it’s a constant challenge to balance on the one hand, the political impulse to have more at home, and on the other hand, the industry impulse to let’s be as efficient as possible. And there are complex balances to be struck there.
[00:15:31] Rene Haas: Do you think governments understand it well? I mean, obviously we’re in the midst of a transition here in our own government, but I think you raise a really important point because on one level, people can look at CHIPS Act, for example, and say, I’m just going to give money to people or grants to who are end quote manufacturing inside the United States. But to your point, it’s just one small piece of a gigantic value chain, whether it’s equipment, whether it’s packaging, the wafers are just a small piece. Do you think that’s well enough understood, the broadness of the problem?
[00:16:02] Chris Miller: Well, that’s a good question. Yeah. Yeah. It’s a challenge, I think. You know, it is a challenge with regard to any industry, the government’s trying to understand, industry will always understand industry better. I think in the case of the semiconductor industry, especially in the United States, but I think it’s true across advanced economies, there had been much less interaction between industry, the semiconductor industry and government over the prior 10 or 15 years than over the last five years. And so there’d been a whole generation of folks in government who hadn’t really engaged with the semiconductor industry. And I think there’s been a lot of learning that’s been done over the last five or so years in the US and Japan and Europe about what the supply chain looks like, who the major players are, what their needs and requirements are. Certainly there’s still a major gap between what the typical person in government and what the typically person in industry knows that’s inevitable. But yes, there’s been lots of learning and studying that’s been undertaken the past half decade or so.
[00:17:00] Rene Haas: Have you got it, I know that when I met with folks in Washington, when I would just do our rounds with our team, they would ask questions about our industry. And I would quote your book. I would say, “Look, if you want, if you want, a primer on learning how this industry works, go pick up Chris Miller’s book.” Have you been inundated with requests for, “hey, teach me more about how this industry works” from government, both in the US and abroad, I would say, because the book is a global, it’s a global industry in the book obviously has global impacts.
[00:17:26] Chris Miller: Yeah, I think there’s been a lot of interest in both US and other governments and also I think the other key player here is the media, which plays a big role in educating the public and also educating government about how the industry works. And I think in the media as well, outside of specific tech-focused publications, there’s been a fair number of journalists who find themselves having to write a lot more about this semiconductor industry and how we had to learn a lot over the past couple of years about how it actually works.
[00:17:52] Rene Haas: There was a period where the semiconductor industry was not considered the most alluring sector to be talking about, let alone writing about. Why do you think that was?
[00:18:00] Chris Miller: You know, I think there were a couple of dynamics there. One was the rise of big software firms in the 90s, 2000s, 2010s, which were, of course, using a whole lot of semiconductors, but the public perception was that it was all about software or all about the consumer internet driving technological progress. And there’s, you know, a degree of truth to that. I do think that the fact that if you think of the big tech companies that emerged in the 2000s and 2010s, they were largely software focused. That shifted how the public in general viewed the tech sector. And so Silicon Valley, which of course was named after silicon, I came to be associated with social media as a result of that. I think that that has changed the last couple of years partly due to the supply chain dynamics, partly due to the politics, but also I think due to the fact that there has been a recognition that actually over the last couple of years, and I think looking forward into the future as well, it’s going to be advances in the hardware that will be just as important as advances in the software, especially when it comes to artificial intelligence. And I think the public perception of that has begun to shift in the last couple of years.
[00:19:09] Rene Haas: Yeah, I think so. It’s, it’s nice to see one of the largest market cap companies in the world being a company that’s in our sector and now you have Broadcom in the trillion dollar club. I know that when we were having to explain, you know, post/during the NVIDIA transaction, and then as we were going public, what does Arm do exactly? It’s to your point. It’s all that software has to run on something. I want to talk a little bit about China. Obviously there’s a lot we could cover on that topic, but you know, in your book and subsequently afterwards, you were referring to just end quote, how far behind China might be in terms of catching up with Western technologies, whether that’s around their internal capability to build EUV machines and, or having the fabs being able to catch up to where the leading edge nodes are with a TSMC or Samsung or Intel, whatnot. If you think about when you wrote the book and where things are now, is China catching up? Are they about where they were when you wrote the book? Or have they fallen behind?
[00:20:07] Chris Miller: Well, I think it’s a, it’s a complex question to answer because it depends what exactly you want to measure. And of course if you measure progress in design capabilities versus fabrication capabilities versus the tools themselves, you get different answers. There’s two ways that I think I look at this that we’ve got pretty good data on. You know, the first would be what’s the most advanced fabrication capability in Taiwan versus in China. And you know, there, I think we saw last year, SMIC rolled out relatively high volume manufacturing of their 7-9 meter capability, which we saw TSMC do about five years previously. So that’s a ballpark five year gap and you can debate around the margin, but I think that’s pretty good data. And you can actually track that gap historically, and what you find is that over the last 15 or so years, it’s been around steady, steady around five years that every year there’s advances at SMIC, TSMC and the gap is pretty standardized over time. So there, I think there hasn’t been a major change in one direction or the other. I think the other interesting place to look is when it comes to semiconductor manufacturing equipment. And there, you know, again, you can debate how best to interpret the data. But if you zoom into lithography tools, which has been one of the key areas of focus. You know, what you find is that China’s imports of foreign lithography equipment are actually at record highs right now, which, you know, doesn’t suggest a rapid degree of domestication. I think for some of the older, less capable lithography tools, there is some domestication happening, but at a pretty slow rate. And so there too, there’s a gap that remains. I think if you look at other spheres, you get different answers. Design, for example, there’s obviously very capable chip designers in China. There’s been some domestication of some of the chemicals and materials, but I guess the places where we’ve got the most visibility into you know, full like unlike comparisons between China and Taiwan, for example that gap remains broadly unchanged.
[00:22:08] Rene Haas: Is it EUV and the mirrors associated with it? Do you think that is, kind of, if you look at the one critical component of the semiconductor value chain and say, gosh, China’s just not making head roads in terms of closing the gap. Is it EUV primarily, or do you think there’s other pieces of the value chain?
[00:22:23] Chris Miller: You know, I think EUV is certainly the hardest single thing to replicate. The fact that it took ASML three decades and many billions of dollars to first develop illustrates just the scale of the challenge there. But I also think that that’s just one of many tools you need and there’s many precise chemicals you need to manufacture semiconductors, and so there’s a lot of different units that make it so hard. And if you were to solve the EUV problem, you’d still have many other challenges you’d have to work out, even though EUV is the biggest of the challenges in my view. And so I think, you know, it’s worth tracking really closely, what evidence can we see of competitors to ASML coming online. I think right now, not much evidence that that’s making progress, but of course, you know, that could change in the future. And that will be worth tracking closely.
[00:23:12] Rene Haas: Now, I know you said you’re a historian, not a technologist, but you know a lot about our industry. You’ve talked to a lot of people and you’ve done an amazing amount of research. Are there new areas of technology that you find that could be very interesting, whether it’s around photonics, different materials, quantum. With the advent of AI, are there areas that you have bumped into from a technology space that you’ve looked at and say, “Hey, now this, in the next 5 to 10 years could be incredibly interesting and compelling in terms of changing how products are designed and developed”?
[00:23:41] Chris Miller: You know, I think the AI boom is simultaneously very good for the existing structure of the industry because big customers are spending huge sums buying as many high end chips as possible. But the fact that they’re spending huge sums also creates a very big incentive to create new technologies that can do the same for cheaper. I think there’s lots of work being done on how within the existing paradigm can you economize on different facets, certainly photonic interconnects being an area of great focus right now. I don’t know about photonic integrated circuits themselves and their maturation. I think quantum computing is a place where, you know, there’s been great optimism about quantum computing for a long time. And it’s, I think, had cycles of optimism and pessimism. I guess I’m struck by the number of people working on quantum computing capabilities that they envision will integrate seamlessly into classical computing. And so rather than sort of a trade off, you’re going to have quantum and not use as much classical, you’re actually going to use both, I think is a way a lot of people now see them developing. And so, that’s actually not a bad thing for the semiconductor industry. It could be a good thing if you open up entire new capabilities and as a result demand more classical chips, as well as whatever quantum capabilities that we’re able to develop. And so when I kind of look forward, it seems to me that we’re still going to need very large volumes of as many high end processors we can make, even if quantum proves as a capable as the most bullish estimates suggest.
[00:25:04] Rene Haas: Yeah, it feels that way at our company. We view it as more likely a hybrid type of solution where you’ve got kind of classic compute working with a quantum machine that could do acceleration of such, But it is an incredibly fascinating time as far as where that all is headed. Are you surprised again, I think your book came out and was it ‘22?
[00:25:20] Chris Miller: That’s right.
[00:25:21] Rene Haas: Yeah. So Chat GPT had maybe not had its lightning rod moment. But two, three years later, are you surprised at the velocity of which we’re seeing all of this innovation taking place with AI workloads, whether it’s again around training or inference or reasoning models, et cetera, et cetera.
[00:25:36] Chris Miller: You know, I think, I think everyone’s surprised. I think it’s been an extraordinary two and a half years since the Chat GPT moment now. You know, I look back to when I wrote the book and I had a chapter on NVIDIA looking in particular at NVIDIA’s AI efforts. I had not a full chapter, but a fair bit on Google and its TPUs, but I could never have predicted just the speed at which both innovation has happened, but also a recognition that you can pretty quickly turn this innovation into real products. I think that’s been the surprise. And so the fact that all of the world’s big tech companies from Microsoft to Meta and across the board are now rapidly increasing their spending on data centers, which means on semiconductors. That’s a surprise to them. It was surprised to most of the chip industry. It was a surprise to me too.
[00:26:25] Rene Haas: Are you an AI optimist or, or pessimist or somewhere, somewhere in the middle?
[00:26:29] Chris Miller: Well, I think I’m broadly an optimist. I mean, I think we collectively have a challenge in making sure all of the investment that’s being undertaken leads to products that are monetizable and economically value additive. Whenever you’ve got this big surge of investment, there’s always some uncertainty. Will we have the business models to justify it in the long run? And so that’s, you know, a task that I think technology companies will have to prove out over the next couple of years, but it seems to me just given the rate of improvement over the last two and a half years since Chat GPT, there’s going to be a whole range of applications that we’re just beginning to get our heads around. I mean, I like the analogy of going back to the internet. You know, if we’re in like 1995, Google had yet to be founded, social media as a concept didn’t exist. And so I think we’ve got a long runway of turning new technologies into new products.
[00:27:21] Rene Haas: I think, yeah, I use that a lot myself, which I think is a very good parallel. And the internet in many ways created a lot of new jobs and opportunities and eliminated some to some extent, as all major technological innovations do. Do you have a perspective on that as a story and watching this closely in terms of, “hey, this time it’s different,” or “no, it’s going to look just like it’s always look for these type of things”?
[00:27:47] Chris Miller: You know, one of the very early computer researchers in the 1950s, a gentleman named Licklider or Lick as he was known to his friends, he worked for DARPA and played a big role in funding a lot of the early computer projects. And he wrote a paper in the mid fifties that laid out how he spent most of his day. And he estimated that he spent, very smart guy, but he spent 85 to 90 percent of his day, he estimated, on intellectual drudgery, things that ought to have been automated, but we’re not automatable. And I reread that paper recently. And I asked myself what share of my day is spent on things that I wish was easily automated. And I think it’s not 85 percent, but it’s still shockingly high. And so, I think most knowledge workers are actually going to embrace automating more of the stuff that now is impossible to automate, but they would actually greatly benefit from automating. And I think, you know, we easily forget that in the early days of the PC revolution programs like VisiCalc, the early version of Excel, were, when they were first brought to market in the late seventies and early eighties were seen to be catastrophic for the accounting industry because you no longer have people running the books manually. And in fact, the accounting profession has survived and thrived despite automation that was brought to them. And I think that’s the right analogy for AI. We’re gonna have lots of applications and lots of different professions, but on net, it’s going to make most of these professions much more pleasant because the not fun part of the job will be automated and the interesting and the fun part of the jobs will largely be maintained.
[00:29:11] Rene Haas: Yeah, that’s a good way to look at it. It’s a bit of the way I tend to think about it too at times where if I look about our industry, semiconductors, and you made a great point in terms of as the chips became more and more complex, they had to be outsourced to other parts of the world for manufacturing. As the designs become more and more complex, you spend far less time on invention, and more and more time just figuring out does this thing work and how do I verify it and how do I validate it? And bug fixes and verification is actually the lion’s share of what it takes to develop intellectual property and or chips. And if AI can help with the drudgery of that I think it’s going to be a good thing. I promise not to get too much into the politics of the new administration and such. But as you go into thinking about the new administration and, and how the world’s going to look like for semiconductors, do you think it will be a better time in terms of, the friction will get less in terms of the globalization of semiconductors, including China, or do you think we might be in for a bumpy four years?
[00:30:07] Chris Miller: You know, I think there’ll be a lot of continuity actually between the incoming administration and the Biden administration in large part, because what the Biden administration did had a lot of continuity with the first Trump administration. When you look at efforts to promote domestic manufacturing to the CHIPS Act, that was legislation that first emerged in Congress under the Trump administration. And I think there’ll be continuity broadly there. Maybe some changes are around the margin, but I think, you know, that’s a bipartisan issue in Congress for sure. I think also on the restrictions on technology transfer to China and to other countries, I think those are going to largely persist as well. Maybe some change around the margin, but you know, there are two, if you look in Congress, you’ve got Republicans and Democrats broadly supportive. So, you know, certainly is there going to be scope for zigzagging and scope for tweaks? No doubt. Does every president want to rebrand policies as their own policy rather than repetition of their predecessors? No doubt. But I think when you zoom out, I think there will be a lot of continuity. I think the one place where maybe we’re going to see a bit of change is on the question of trade policy and tariffs, where, you know, on the one hand, we did see the Biden administration impose new tariffs on semiconductors imported from China to the US, but I think President-Elect Trump has talked a lot more about making tariffs a central pillar of his trade policy. And so we’ll see, you know, over what time horizon at what scale, but I do think I expect more tariffs, which will certainly impact any industry that’s as globally integrated as the semiconductor industry.
[00:31:36] Rene Haas: It’s going to be exciting for sure. Is there another book in the works?
[00:31:39] Chris Miller: You know what, I’m spending so much time just tracking everything that’s changing in the semiconductor industry, I’ve got no time for a new book. I hope to dive into one soon, but nothing to announce yet.
[00:31:50] Rene Haas: Got it. Cool. Chris, thank you very much for spending time with us.
[00:31:53] Chris Miller: Hey, thank you. This is a lot of fun.
[00:31:59] Rene Haas: Thanks for listening to this month’s episode of Tech Unheard. We’ll be back next month for another look behind the boardroom door. To be sure you don’t miss new episodes, follow Tech Unheard wherever you get your podcasts. Until then, thanks for listening to Tech Unheard.
[00:32:13] Credits: Arm Tech Unheard is a custom podcast series from Arm and National Public Media. Executive Producers Erica Osher and Shannon Boerner. Project Manager Colin Harden. Creative Lead Producer Isabel Robertson. Editors Andrew Meriwether and Kelly Drake. Composer Aaron Levison.
Arm production contributors include Ami Badani, Claudia Brandon, Simon Jared, Jonathan Armstrong, Ben Webdell, Sofia McKenzie, Kristen Ray and Saumil Shah. Tech Unheard is hosted by Arm CEO Rene Haas.
Catch Up on the Previous Episodes

Mike Gallagher: “Make Original Mistakes”
Having first served in the Marine Corps and later in Congress, Mike Gallagher now manages Palantir’s defense business in the private sector. Now, he shares his story with Arm CEO. Rene Haas, explaining the role of AI in national security, both present and potential, and revealing his best “life hack.” Join Mike and Rene to learn whether leaders are born or made, and why great leaders aren’t afraid to take risks or make mistakes.
<Rene>[0:06]
Welcome to Tech Unheard, the podcast that takes you behind the scenes of the most exciting developments in technology. I’m Rene Haas, CEO of ARM. In this podcast, I’m sitting down with some of the best and brightest in the industry to share insights, stories and visions for the future. Today, I’m joined by Mike Gallagher, a former congressman and Marine Corps military intelligence officer. Since leaving Congress in April, Mike has joined Palantir, the software platform for big data analytics. Mike was brought on board to lead their growing defense tech business, marking his first shift from the public to private sector. Mike, it’s a pleasure to have you on the podcast.
<Mike>[0:41]
It’s an honor to be with you, Rene.
<Rene>[0:42]
Great to be together. We met when you were in Congress heading up a bipartisan China select committee. So we have a little bit of history. So thanks so much. So your background is fascinating. Former Marine Corps, served in the Middle East, if I’m correct. Served in Congress. Now, you’re head of defense at Palantir. How does that happen?
<Mike>[1:04]
Contingency, happenstance, luck? No, I’m from Green Bay, Wisconsin, originally, but I think I always had a fascination with the world outside of Wisconsin and outside of the United States. So when I went to college, I knew I wanted to study international relations, then was working at a think tank the summer after my sophomore year in the U.K. and I got assigned to this project studying terrorist targeting methods. And we had just invaded Iraq the year prior – this was 2004 – and I became fascinated by the Middle East. I became fascinated by our response to 9/11. So I went back to Princeton. I changed my major so I could learn Arabic, which was a bad decision as a junior because you had to go to class every morning at 9 a.m. But I loved it. And that led me to think about, okay, what would I do with these language and regional skills and the military – I don’t come from a military family – jumped out at me and I saw it as an opportunity not only to scratch that intellectual itch, but serve my country, pay back a debt I felt I owed. And I felt like I didn’t want to be sitting on the sidelines while the country was at war and also test myself physically, mentally in terms of my leadership. And it was just kind of that – I didn’t know anywhere else where I could have that intense crucible and, as a 21, 22 year old, be in charge of, you know, 50-ish Marines, I wanted that leadership challenge. And so the Marine Corps was just a great fit for me. The Marine Corps is a great program for people who aren’t at ROTC or don’t come from a, you know – I didn’t come from a military family, as I said before, so it was just, it really jumped out to me. So, a circuitous path that led me to join the Marine Corps, which is one of the better decisions I’ve made in my life.
<Rene>[2:47]
So I want to find out also about Congress and Palantir, but let’s – let me talk about the Marine Corps for a little bit. Did you do Parris Island?
<Mike>[2:54]
No, so if you’re an officer, you go through Quantico. And I went into a specialty called Counterintelligence, Human Intelligence, and did another specialty school after that deployed right out of school to Iraq. And they did two back to back deployments. But the usual journey or that I mean, the journey for an officer in the Marine Corps is to go through Quantico.
<Rene>[3:17]
So Quantico, how many folks – I’m sorry. I’m now, I’m just so fascinated with this piece here. How many folks get through it? Do some people drop out?
<Mike>[3:26]
Yeah. Yeah. In officer candidate school is usually where the most attrition happens. I forget the attrition rates, but it’s tough that the initial assessment, a lot of people get injured. And then there’s less attrition once you get to the basic school, but some people do drop out and they’re really just kind of screening you to see if you have what it takes to be a Marine Corps officer intellectually, physically, the physical standards are very high. I think that is one thing that distinguishes the Marine Corps. We tend to set a pretty high bar physically in terms of the physical fitness test. But I loved it.
<Rene>[3:58]
And you were what, your mid-twenties at this time or how old were you?
<Mike>[4:02]
Yeah, probably 21 – yeah, 20. I was 20 or 21 when I went to officer candidate school the summer after my junior year of college and then 22 because I went right after – I graduated from college, I got commissioned as an officer the day I graduated. I then had about four months before I had to report to Quantico for the basic school. So I went up to Middlebury and did their Arabic immersion program. Middlebury has a phenomenal language program. You take a language pledge where you’re only allowed to speak that language, and really you cram two years worth of coursework into a few months. And so I was able to really hone my Arabic prior to reporting to Quantico, which then served me well when I deployed, because as a human intelligence officer, I was doing a lot of interrogations, I was doing a lot of source operations. So being able to communicate directly in the language, even though I kind of spoke more of a stilted, formal version of Arabic, what’s called the Fuṣḥā. I didn’t speak the local dialect or the Amiyah. It still served me well and allowed me to go places and do things that some other people weren’t able to do.
<Rene>[5:02]
I feel incredibly inferior based upon – at 21, I was just hoping to get out of electrical engineering university and get a job for that. That’s unbelievable. How did you get into politics?
<Mike>[5:12]
So, a long story short, Marine Corps for seven years. Couple of deployments. Got to work for H.R. McMaster, General Petraeus, a little bit of a stint of tours in the intelligence community. I was a Senate staffer on the Senate Foreign Relations Committee, the Middle East guy, for two years. I then moved back to Wisconsin to be the national security adviser on a presidential campaign when our governor, Scott Walker, ran for president, which was a phenomenal experience, even though he didn’t win that campaign. That was in 2016. And so I was back home in Wisconsin, and I knew – I was actually trying to pivot to a private sector career. I’d use my G.I. Bill to get my Ph.D. in international relations. And so I really – the conception of the career I had in mind was I would do private sector and I would teach as kind of a side hustle and a way to scratch my academic itch. Then my congressman unexpectedly retired, a great guy named Reid Ribble. And because I had just been on a presidential campaign going around the state and the country, talking to people that were engaged in politics about the future of foreign policy, some people in northeast Wisconsin asked me if I’d be interested in running, and it was very intimidating. It was not, I mean, I was not – I don’t, I don’t have any political lineage. I’d never really been in front of the camera before. I didn’t know anything about political fundraising. I was a national security policy guy. And so at first I thought, no way, I can’t do this. But then I thought, you know, here I am criticizing the direction of U.S. foreign policy, criticizing Congress. I felt like – I was 31 or 32, why not step up. So I got into the race. I beat a long time state senator and some others in a primary and then won what was a very contested general. My district used to be a very competitive district and then it became very Republican. And so it was kind of just right place, right time. And maybe to connect it to my Marine Corps experience, I did view it as an extension, a different way to serve the country and continue to serve focusing on the issues that I had focused on in the Marine Corps defense, Middle East, national security, but also just felt like this was another opportunity to throw myself into a crucible that was very uncomfortable for me. And I tend to think like every once in a while, every three or five years, you should really get outside your comfort zone. And that’s the only way you grow as a leader and as a thinker.
<Rene>[7:35]
One of the things that folks have talked a lot about is leaders, are they born? Are they made? And I think of anyone who’s kind of gone through what the Marine Corps embodies to me is the quintessence of leadership, not to mention being on the battlefield. Your viewpoint on leaders: born, made? Can you teach it?
<Mike>[8:00]
I think they’re made, you know, recognizing everybody has different strengths and weaknesses and different gifts that are, in some sense, innate. By and large, I still think leaders are made, and I think the primary way in which you become a better leader is by failing a lot. I mean, it’s not like I emerged from Quantico as an impeccable Marine Corps leader. I made a ton of mistakes while I was deployed. I continue to make mistakes as a leader. I remember vividly one time I had a Marine who had what’s called a negligent discharge, which is when you fire your weapon accidentally, it’s not a good thing to do, particularly in the Marine Corps. And it caused a lot of drama with our local unit. And my initial response was to try and like make sure there was as little fallout as possible, i.e. nothing that could jeopardize our unit and my own career, quite frankly. And then I had a boss who was an amazing guy, and he basically took responsibility for the whole thing. He said it was my responsibility to get these Marines ready for deployment. Don’t punish the Marine, punish me. And that was like a jaw dropping moment when I realized that – what it means to be a leader, like you have to take ownership of those under your command and you have to put their welfare ahead of your own. And that was a lesson I had to learn the hard way. And I was quite ashamed of myself, quite frankly. And so it’s a continuous journey. But I do think leaders are made. Final thing I’d say, Rene, is I think sometimes people have like a Hollywood conception of leadership in their mind, particularly when they go into the Marine Corps. I think the challenge is adapting timeless principles of leadership to your own unique personality. Like I would never be the like, you know, hardcore drill sergeant type. That’s just not my personality. I’m a bit more collaborative, professorial. And so I had to adapt kind of the Marine Corps vision of leadership to my own unique personality and innate traits, if that makes sense.
<Rene>[9:54]
Totally. And I agree with you. I think leaders are made. There are qualities we’re all born with. But we also have the ability to adapt and be made and learn, etc., etc.. And one of the things that I talk about a lot with our company is around resiliency and making mistakes and the fact that you just can’t learn and develop and get to that next level without making mistakes. Now, going into your commentary about making mistakes and learning: you served in Iraq. How do you recover from mistakes when you’re actually in combat? That seems to be a whole different level.
<Mike>[10:29]
Well, listen, I mean, I was an intelligence officer, so I’m not trying to pretend like I was some, you know, big combat hero or anything like that. And by and large, I was the beneficiary of the sacrifices that were made far earlier in the surge, you know, and we lost a lot of good Marines in western Al Anbar province. But by the end of my second deployment, I mean, we were walking around, you know, without our body armor. It had largely shifted to more of a like, a humanitarian in some sense, a humanitarian and civil society building mission. But yeah, yeah, there’s certainly, it’s still a life or death enterprise, and that can be hard for certain people to deal with. But that’s where I think, you know, having men and women to your left and your right that you trust really building a shared culture and ethos. I think the Marine Corps does that better than maybe any other organization on earth, right, everyone, you know, rich, poor, black, white, men, women, if you make it through the basic Marine Corps crucible, you’re Marine. Everybody bleeds green, as the saying goes. And so it’s only by having a strong team around you that you can cope with failure. And to the point about mistakes. My sort of personal mantra in life is just to make original mistakes. My only goal is to not be making the same mistakes over and over again. And I try to encourage my subordinates to make original mistakes and give them enough of a leash where they don’t feel like they’re going to get punished for taking intelligent risks.
<Rene>[11:45]
Absolutely. I think about my career and the most growth I’ve ever had has been making mistakes and the learnings. So now you’re at Palantir. Amazing company. Alex Karp, fantastic CEO. Palantir has been on fire. Before I ask you about the transition. I’m not sure how many of the folks who are listening to this know exactly what Palantir does. So maybe tell us, what does Palantir do and what was the inspiration to kind of go from public service to joining them?
<Mike>[12:13]
I mean, the inspiration was I have been aware of Palantir’s work for a while. It had started to be deployed into the Marine Corps as I was getting out of the fleet and I had some friends that had used it very early on and were blown away. I mean, If you think about like what an intelligence officer in the Marine Corps does, right? In my case, you collect information from humans. You sort of write that up into a report. You can plot certain information on a map. You brief things on PowerPoint slides. All of this is very time intensive and involves a lot of humans in the loop. It was very like analog, even back then. So along comes a piece of software that allows you to automate that process and speed up your decision making, make sense of disparate sources of data, and now use satellite imagery and take a process that used to involve hundreds of human beings and thousands of hours and distill it down into, you know, one human being and make it a matter of minutes. I mean, you can see the benefits in terms of creating decision advantage. And so today on the battlefield Palantir software operationalizes AI and allows our fighters to make sense of the battlefield, see bad guys, target more precisely. And then in the boardroom, on the commercial side of our business, Palantir’s software – basically the way I like to think about it is anywhere in a business where information is instantiated on a whiteboard or in an Excel spreadsheet or in the mind of a human being, you can now embed that into our core product, which is called Foundry, and then use AI to automate various processes and just move faster than your competitors and save money in the process.
<Rene>[13:48]
Two questions. So first off, Palantir, I think externally has a brand that people are associating with defense. And as you said, you guys use it in the military. But do you think the commercial opportunity for Palantir is larger? It feels like it should be just given the problems that you described and the scale of data that exists in a commercial sector.
<Mike>[14:10]
Well, Palantir, you know, it’s unique. It certainly got its start in the wake of 9/11 and looking at the failures that led up to 9/11 and believing strongly that they could have been prevented with a better use of technology and allowing decision makers to better connect the dots in terms of what our enemies were doing. But then it evolved and in the last five years, the commercial business has really taken off. And now we’re in this unique position where we have a thriving commercial business and a thriving government business. But really what intrigues me the most is the opportunity to take the lessons learned from both sides of the business and cross-pollinate them. Right, because there’s certain insights we have because we’re working with warfighters in the Pentagon, helping them make sense of the battlespace or helping them, like, digitize a process that used to be a matter of PowerPoint slides that we can apply to the commercial business and vice versa. The Marine Corps is launching a big barracks modernization initiative right now. I see an opportunity to take some of our insights in the construction field, in the commercial world and apply it directly to the defense business and help our military automate the so-called tail – that’s all the supporting establishment – and use those saved resources to sharpen the tooth – that’s the things at the tip of the spear, that’s the war fighters, that’s the weapons that directly enhance our lethality and increase near-term deterrence. So that’s the opportunity I see from both a business and a national security perspective. And increasingly, if you look at the defense industrial base, it used to be the case that most of our biggest defense companies had thriving commercial businesses and would actually subsidize their own R&D and not ask the U.S. taxpayer to do it. Because the defense industrial base has grown smaller and ossified, it is increasingly rare to see a company like that that has both thriving commercial and defense work. And so I think this is the model we need to return to. And I think that’s sort of the unique position Palantir is in right now. And hopefully we can harvest those commercial insights in order to better serve the warfighter going forward.
<Rene>[16:08]
Do we need more Palantirs from the perspective of an interesting comment you just made, because I started my career 100 years ago at TI, and this is 1980s. Ronald Reagan’s our president, Star Wars. We had a huge semiconductor effort that was all about doing work for Patriot missiles and everything that was really during the Cold War. And to your point, TI, at that time largest semiconductor company in the world, had a gigantic defense semiconductor business. And as you said, that’s kind of gone by the wayside. Do we need more companies doing the kind of work Palantir does – not directly in your competition, but to your point of a thriving business that helps them on the defense side? Because I wonder, how can government keep up without it?
<Mike>[16:52]
That’s right. Well, we should obviously have no competitors. There should be no – I’m just joking. Yes, every business’s dream.
<Rene>
Now that you’re out of politics, you can say that. Yeah.
<Mike>[17:00]
Yeah, exactly. Exactly. I would put it this way, I think we need more non-traditionals to be, to get into the government space to cross the so-called valley of death that Palantir has spent the last two decades and spent billions of dollars doing. It is far too hard for a company who wants to help the government and help the Defense Department or the intelligence community to do so. Because the Defense Department is not an easy customer. It’s part of the reason why we launched our FedStart program to allow the younger and non-traditional tech startups, which still face these incredible barrier entries to get into the defense ecosystem and not have to do that journey that Palantir did. And so I think if we do that, we can have a defense ecosystem that isn’t dominated by a small number of defense primes, but it’s far healthier where the primes are working with non-traditionals. But where the rubber meets the road is just how the Pentagon buys things and how it spends money. And we’re still struggling with this problem of – we tend to sprinkle out, you know, money for non-traditionals in innovation and dilute in small grants, as opposed to DOD making big bets on non-traditional companies. And I think fixing that, combining that with a greater tolerance for what’s called multi-year appropriation so you can provide some predictability to companies that are trying to build things for the Defense Department, is the path forward. all of this redounds to making sure we have the best and brightest human beings in these companies willing to work with the Defense Department. That’s another way where we can build a bridge between Silicon Valley and the defense and the technology industry and the core national security community.
<Rene>[18:44]
Yeah, makes sense. And it’s a good segue into this question, which I think you are probably in a very unique position to have a perspective on now that you are in the private sector, but obviously served our country in Congress. And this is around AI, and should there be a national policy around AI? And when I say national policy, it’s – obviously that’s an umbrella term for the government having a much higher involvement in terms of security, safety policies, procedure guidelines, etc., etc. What’s your view on that?
<Mike>[19:18]
Well, in one sense, I’m not sure it makes sense to talk about a national AI policy per se, because AI is not a single piece of technology. It’s a cluster of technologies. And the risks and opportunities are different depending on what specific aspect or what specific cluster you’re talking about. So I think the good part of the latest executive order, which is the closest we have to a national policy, is that it does recognize that sector-based approach and defers to the various sectors in terms of drawing on their sectoral proficiencies to cultivate AI development and manage the risks. But the problem is that we haven’t put any real investment behind that policy. So put differently, I think the most important national policy we can have where the federal government has a clear and unique role with private sector input, is for the Defense Department to spend more money on responsible AI use. Right now I think the numbers that the Defense Department spends about 0.2% of its budget on i.e. increasing that amount to just 1% or 8.42 billion. To support our troops with the most advanced form of software available from commercial providers would have an outsize impact on our defense and our deterrence capabilities. And so that to me is one area where we do need more national involvement. Another is having a policy that forces the Defense Department and other agencies to adhere to existing law. There’s a commercial preference embedded into existing law so that the government doesn’t try to build software and AI tools itself, and yet this law goes violated on an almost daily basis and we waste a lot of money on government off the shelf solutions when we should be seeking to buy commercial off the shelf solutions. So there’s a series of things I think we can do in the pure defense space that amount to a national policy and balance the sort of need to go fast and innovate with the legitimate concerns about the safe and responsible deployment of AI, if that makes sense.
<Rene>[21:24]
Makes sense. But should it be regulated? Should large language models be regulated and tested before they’re available for the consumer, for example?
<Mike>[21:34]
Well, I think the best way to ascertain what the right form of regulation is, is a field to learn or a test fix test approach. Right, you field AI with the end users and operators with workflows that are relevant to their missions. The models are then improved through iteration with operators in the field, and then you refine the systems as you extend it to larger groups over time. And in that journey you figure out what the right guardrails are. I think it would be the height of hubris to think that a few bureaucrats could sit in a room, even if next to them were the leading minds and private sector leaders when it comes to artificial intelligence and construct a perfect regulatory framework that still allows room for going fast, you need to approach it with that field to learn and test fix test approach.
<Rene>[22:25]
In the private sector, when people talk about things they hate most about their jobs, they’ll say, I love the technology, I love the company, I love the people, but God, I hate the politics inside my job. You worked in a domain called politics. What did you like most about that role inside politics and what was something you look back and say, gosh, that was not fun to do.
<Mike>[22:46]
Well, the latter question is easier. The thing I hated most was fundraising. And I do think the fundamental dilemma, if you’re a member of Congress, is that you are having to do your job while also raise money to run for your job simultaneously. And in the house where you’re on a two year time horizon, that can be very difficult. Just carving out enough hours in your day to do responsible oversight, attend all your committee hearings, and then have to go across the street and raise money. I never enjoyed fundraising. I always found it weird, maybe because I have Catholic guilt, to ask people for money, and so not having to fundraise anymore has been a huge blessing in my life. So that was definitely the bad, the good. However, and I think I had the benefit of, I knew I never was going to make it a career. I believe in the model of a citizen legislator. I think that’s why I’m a proponent of term limits. I think members tend to stay too long and it should be a season of service. So that was very liberating for me. And so that gave me the freedom to really focus on the committee work. I was most passionate about the Armed Services Committee work, chairing the Cyberspace Solarium Commission and then ultimately chairing the Select Committee on China was profoundly rewarding. I love that. And then what do you do when you’re back in your district? Well, one, you go round to all the different businesses in your district and you just learn what people are making, what their daily lives are like. It has a way of helping you fall back in love with where you’re from. I didn’t fully appreciate it when I first ran and that was really rewarding. And you learn a ton or you help people solve problems, helping people get their VA benefits, helping them deal with the thorny immigration issue, helping them deal with weird government regulations. I mean, it’s sad that sometimes you have to call your congressman to get those issues fixed. But if you’re the congressman and you can really help someone, that’s a big deal and that’s super rewarding. So I love that. The final thing I’d say, Rene, which I think you’ll appreciate, is, you know, it’s really hard to build a team in business in the public sector. So it took me a while to really get my office team and culture exactly how I wanted it. But in those final two years, I really felt like we had an incredible team. Everyone was kind of operating by commander’s intent. And our decision making was really fast. And that was really rewarding for me because you’re working with a lot of young people on the Hill, too, which is, it was just energizing. So I miss that team.
<Rene>[25:05]
You’re an amazing speaker. When you went up and did a campaign speech in your district, prepared remarks or you just wing it?
<Mike>[25:14]]
You know, the first four years, very prepared. I had never been in front of a camera. I was incredibly nervous the first time I did a local TV show, let alone going on a national news show. And so I would overprepare. I would often write my speeches out and I would memorize them. But then I would say I got comfortable doing everything extemporaneously and I went too far in the complacent direction of not preparing. And now I find myself having to remind myself that I need to prepare and that I’m a bit out of shape. I will say, however, I do think the one life hack or maybe two life hacks that have melded into one that have given me a competitive edge are that in the Marine Corps, I started up I started waking up really early and I would devote, you know, after I do my sort of Catholic thing in the morning, I would devote an hour just to writing. And I do think forcing myself to write out my thoughts, even if it didn’t result in a publishable Op-Ed or article or a talking points that I would use on TV, it’s still a discipline that I practice every single day and I think helps me communicate, because my view is that if you can’t write clearly, then you can’t think clearly and you certainly can’t speak clearly.
<Rene>[26:28]
Yeah, that’s kind of how I do it. What I do is an outline of things and a flow that I want to go. And when I’m doing a presentation, for example, I always drill on my team, don’t show me the slide I’m presenting, show me the next slide. So what the next slide does for me, it just gives me a context of the story I’m trying to tell and that there’s a continuum to it. Brett Favre, Aaron Rodgers, who do you take?
<Mike>[26:52]
Bart Starr, number 15. Every Sunday, you see me wearing number 15, Bart Starr’s jersey.
<Rene>[26:59]
It’s a great answer. A truly gifted politician. I give you two choices and you come up with a third. Would you ever run for president someday?
<Mike>[27:08]
Genuinely, I do not look in the mirror and think ‘one day I’m going to be president’. Like I am at core still that kind of, I think of myself as a national security professional, so I’d love to serve if the moment were right in a national security job, but there are no plans in my head for running for president one day. My wife is pregnant with twins. So we’re going to focus on having children and building out our family for a while before I get any political ideas. I’m trying to be you, Renee. I’m trying to learn how to be a leader in the private sector. So I’m on month two of this, so I’ve got a long journey ahead of me.
<Rene>[27:45]
Oh, my gosh. I don’t know where to start there. Mike, thank you so much for giving us the time here. This was terrific.
<Mike>[27:52]
Thank you, sir.
<Rene>[28:00]
We’ll be back next month with more exclusive conversations and insights from the world of technology. Make sure you follow tech unheard wherever you listen to your podcast.
Tech Unheard is a custom podcast series from Arm and National Public Media. Executive Producers Erica Osher and Shannon Boerner. Project Manager Colin Harden. Creative Lead Producer Isabel Robertson. Editors Andrew Meriwether and Kelly Drake. Composer Aaron Levison. Arm production contributors include Ami Badani, Claudia Brandon, Simon Jared, Jonathan Armstrong, Ben Webdell, Sofia McKenzie, Kristen Ray and Saumil Shah. Tech Unheard is hosted by Arm CEO Rene Haas.

Jensen Huang: On Leadership and AI’s Industrial Revolution
In this first episode Arm CEO Rene Haas speaks with Jensen Huang, the CEO of NVIDIA, a true visionary, his former boss, and his personal mentor. They dive into Jensen's journey, the future of AI, and how NVIDIA's unique culture of relentless innovation and ambition continues to push the boundaries of technology.
[music comes in]
Rene:[0:00]
Welcome to Tech Unheard, the podcast that takes you behind the scenes of the most exciting developments in technology. I’m Rene Haas, CEO of Arm. At Arm we’re shaping the future of computing with the industry’s most powerful and energy-efficient compute platform designed to unlock the full potential of AI. Our technology is at the core of innovation
Rene:[0:24]
for leading companies across the globe. In this podcast, I’ll be sitting down with some of the brightest minds in the industry to share insights, stories and vision for what lies ahead.
Rene:[0:34]
Today, I have the privilege to speak with Jensen Huang, the CEO of NVIDIA. A true visionary, my former boss and a personal mentor of mine. We’re going to dive into his journey, the future of AI and how NVIDIA’s unique culture of relentless innovation and ambition continues to push the boundaries of technology. We sat down and met at NVIDIA’s headquarters in Santa Clara to talk.
[music crescendos and then fades out]
Rene:[1:02]
Ready to go?
Jensen:[1:03]
I was ready the moment I walked in.
Rene:[1:05]
It’s great to be back.
Jensen:[1:06]
Well, thank you. Yeah, it’s great. It’s great to see you.
Rene:[1:08]
It’s great to be back here at NVIDIA. This building did not exist when I worked here many, many years ago.
Jensen:[1:14]
How many? How many years ago now? 20? (When you first started?)
Rene:[1:17]
I started in 2006. I left in 2013.
Jensen:[1:21]
Yea, see? 20 years.
Rene:[1:22]
Yeah. 20 years ago. These buildings did not exist. It’s a, it’s a nice feeling to be back, though.
Familiar. Thanks for, thanks for spending the time.
Jensen:[1:27]
Thanks for having me.
Rene:[1:28]
So now that you’ve grown so large, one of things I’ve always been curious about, Jensen, with NVIDIA is hiring. The culture is one of a kind. The company does things in a one-of-a-kind way. How do you identify folks who are going to be successful inside NVIDIA?
Jensen:[1:42]
We’re not always successful in doing that. Look how you turned out. [laughter] That’s [laughter] it’s always a shot in the dark. I think that the interview process is not an excellent way to judge whether somebody is a good fit. I mean, obviously, everybody could pretend
Jensen:[2:02]
to have a very constructive conversation. You could learn a lot from just watching YouTube on how to interview. And so, you know, the technical questions, of course, people
Jensen:[2:12]
even share what NVIDIA technical questions are. And we try to be as rigorous and difficult as possible. But – but it’s hard. I think that my method is always I go back to reference checks, you know, and I ask them the questions that I was going to ask the candidate. And the reason for that is you could always make make for a great moment, but it’s hard
Jensen:[2:33]
for you to run away from your past. And so I think those are good. I like asking one in-depth question and just thinking about how they reason through it.
Jensen:[2:43]
But I think in the final analysis, NVIDIA has been successful for a lot of people. Our attrition rate’s very low, as you know. And and so it’s a really diverse environment with a lot of really interesting people in the background. And we have people from from just about every great company in the world and somehow, we’ve made them successful here. And so I
Jensen:[3:02]
think that that on the one hand, building a great company is about getting great people. On the other hand, building a great company is really about creating the conditions by which those people could do even better than they thought they
Jensen:[3:13]
could. And, you know, a lot of that has to do with being transparent about explaining what NVIDIA’s vision and strategy and what makes us work. As you know, I spent a lot of time doing that. And our company has always been known for its transparency about explaining what what challenges we have, what opportunities we have, what strategies
Jensen:[3:33]
we’re executing. And information is flowing fairly readily inside the company with respect to, you know, what is it the company’s strategies are. I always find that it’s strange
Jensen:[3:43]
when companies have too many silos and, you know, need to know basis. I think obviously, you know, people don’t need to know what they don’t need to know. But the more that they know, the more they’re empowered to be able to make good decisions on our behalf. And so I try to err on the side of transparency. I try to err on the side of empowering people.
Jensen:[4:03]
And as a result, you know, the company is one of the, I think we’re the smallest large company in the world.
Rene: For sure.
Jensen: You know, I think that’s just that comes with the the incredible productivity
Jensen:[4:13]
of the people. And we have 30, 30,000 people or so maybe a little bit more than that now. And they’re making hundreds of decisions a day. And if all 30,000 of them are, you know, statistically moving in the direction, making decisions that are ambiguous decisions oftentimes, but they’re making it in the direction of of what is in the
Jensen:[4:33]
company’s best interest long term. It adds up really fast.
Rene:[4:37]
One of the things that always amazed me is that– you know, back to that point–and again, I don’t know whether it was hiring the right people or self-selection, but by having senior leaders who are extremely comfortable with ambiguity and the fact that you would reach down into different layers, the organization, i.e. the project, is what’s most important.
Rene:[4:54]
I just wondered how did that happen? Is it just something that as you grew the company and you had senior leaders who were aligned with your vision, that it grew up that way? Because it was just amazing that
Rene:[5:04]
so many of the senior leaders here when I worked with NVIDIA, worked at NVIDIA, they were completely fine with the fact that you would just reach around and get the right people in the room to solve a problem.
Jensen:[5:14]
Well, first I didn’t ask them, as you recall.
<Rene>[5:17]
I do recall.
Jensen:[5:18]
And the reason for that is because you shouldn’t have to ask permission for something that is that obvious, you know. And so the reason why we said it that way is that NVIDIA was designed to be a full-stack computing company, we were designed to be a company that would build GPUs and CPUs and networking chips
Jensen:[5:38]
and switches, and we would do architecture and design of chips and develop system software and create algorithms and even, you know, create solvers.
Jensen:[5:49]
And so how would you organize such a thing where everything has to work together on the one hand, but you have to build it in parts on the other hand. [Rene: Yeah.] And so the way we solved the problem was, instead of having organizational silos, we thought of the organizations as a place where the leaders can groom people,
Jensen:[6:09]
create conditions for them to succeed, be of service to them, to help them remove obstacles and such. But the mission; the mission is the boss, and it cuts across the whole company.
Jensen:[6:19]
So it can cut across systems and chips and networking chips and software and algorithms, and it can cut across all kinds of domains. And by organizing that way, we also created transparency, you know, all these silos became porous. And when organizations are porous, it tends to be better, you
Jensen:[6:39]
know, because you have a lot more people who are able to help you criticize it. You have a lot more people to help you prove it. And so I love this, the porosity of our company, if you
Jensen:[6:49]
will. I just love that everything is transparent and everybody’s helping me make it better. And,
you know, [Rene: Yep] it’s not like everything is in some kind of, you know, dark silo.
Rene:[6:59]
You almost acquired us, which would have been would’ve been fun. But you acquired Mellanox.
Jensen:[7:02]
I know you’re still sad about it.
Rene:[7:04]
I’m still sad. [laughter] Every day I cry a little bit, but. But I’m here.
Jensen:[7:08]
Thank you. [laughter] But you guys have done so well. You have done so well.
Rene:[7:11]
But you did acquire Mellanox, which has been not only an amazing acquisition in terms of your strategy, but it also just seems like seamlessly, to your point of a porous organization where the mission trumps everything… From the outside in it looks incredibly seamless in terms of execution. How did how did that happen? I mean, how did you make that so? I mean, M&A is so tough.
Jensen: It is tough
Rene: It is culturally very tough.
Jensen:[7:32]
Yeah, it is tough. Well, first of all, there are ten people, I think maybe more ten, 12 people on the Mellanox management team, the NVIDIA Israel management team that sits on E-staff. We have
Rene:[7:46]
That’s great. That makes a difference.
Jensen:[7:46]
architecture, we have research, we have software systems, the chips, we have nets and switches, we have NVLink switches now. We used to have just InfiniBand product line, but now we have a whole Ethernet product line. In the short time that we’ve been together, the product portfolio
Jensen:[8:07]
of Mellanox, well, quadrupled and they’re integrated into every aspect of everything we do. If you look at the transformation and you recall
Jensen:[8:17]
the acquisition, our vision was that the unit of computing was no longer going to be, for example, a GPU, which is really a peripheral. Arm, helped us, in fact, quite importantly, to transition into a company that was building an SoC. And now remember what an SoC is, an SoC is basically
Jensen:[8:37]
a whole computer, whereas a discrete GPU is the last thing that comes up in the computer. The CPUs com up, the boot ROMs come up, the operating systems come
Jensen:[8:46]
up, and eventually the GPU comes up. In the case of an SoC, you have to bring the whole thing up yourself. And so it caused NVIDIA to to evolve from being a algorithm company, which is really what a GPU company is to a computing company. That was our first entry and the SoC wasn’t easy for us in the beginning. We built some amazing
Jensen:[9:07]
ones now. And then the next evolution for us was building systems and DG X1 was our first. In fact, I’m still quite fond of of SHIELD,
Jensen:[9:16]
which is our Android TV computer and I’m very fond of it because it was really NVIDIA’s first full system that we created and –
Rene:[9:26]
The learnings on SHIELD must have been amazing. Now looking back, because I remember when we started.
Jensen:[9:30]
That, yeah, it is still the most popular Android TV box that people –
Rene:[9:34]
Back in the day. It was a PlayStation Xbox controller with a display and we were just think to ourselves, how do we do this?
Jensen:[9:40]
Yeah, it is still my favorite thing that NVIDIA has ever made.
Rene:[9:42]
I completely forgot about that. Yeah.
Jensen:[9:44]
That’s really good.
Rene:[9:67]
That was a system.
Jensen:[9:48]
Yeah, I learned a lot.
Rene:[9:47]
Yeah.
Jensen:[9:49]
I learned a lot. And to this day, we’re still maintaining the software.
Rene:[9:52]
It was utterly unobvious that there was a fit in the marketplace for this. And I remember folks inside the group suddenly having to source a whole set of components. [cross-talk]
Jensen:[10:04]
Exactly. It was my excuse to turn NVIDIA into a systems company and people will ask me, you know, the DGX1, which is the the computer that changed everything. Or, you know, how did that come about? Well, DGX1 one is just a very large SHIELD.
Rene:[10:18]
Very large SHIELD. Yeah. Yeah.
Jensen:[10:20]
And and so to to me, the fact that SHIELD was made out of plastic, and DGX1 weighs 600 pounds, you know, that transition wasn’t a big deal. The big deal was that we were now able to build systems and and then when we bought Mellanox, the big idea was that the computer was no longer going to be that node, but the computer
Jensen:[10:39]
is going to be the entire data center, that the data center is going to be the unit of computing. And if you don’t, if you don’t design GPU, the CPU, the NIC, the
Jensen:[10:50]
switches, all of the transceivers, and connect everything together and be able to boot that system up, you know, from nothing and get everything all wired up. Get everything all running and distribute workloads across it. If you don’t do that, you’re really not going to understand what it means to build these AI superclusters and that transition,
Jensen:[11:09]
that vision was so clear that it was necessary for galvanizing the two teams. You know, in order to galvanize teams, you have to have a very clear
Jensen:[11:19]
vision. And we had a very clear vision. And that vision was also very tangible because you could see it sitting right in front of you there, super cluster and got all the gear from both companies. And, and so the vision was clear and inspiring. It’s tangible or we have to make it tangible. As CEOs you have to make abstract things tangible.
Jensen:[11:41]
And we went off and built it. And so anyways, I also think that their culture is great.
Rene:[11:46]
Yeah. And that clarity really helps. But going back to kind of the vision thing for a second, and there’s another thing that I do when I tell stories about the company. SHIELD a good example. CUDA in the early days, chasing oil and gas, is a good example where it’s completely unobvious.
Jensen:[11:57]
People didn’t realize that. In fact, that was our first. Was first. Yeah that was.
Rene:[12:00]
Completely unobvious what the the real end quote killer app or end state is yet you have an incredible resiliency to experiment with ideas early and test them even though the market doesn’t either appear ready and or have the definition for it. What do you chalk that up to? Is that incredible intuition? Is that seeing around – [cross talk]
Jensen:[12:23]
We’ve had good intuition, you know, ten times in the company, as you know, and the benefit that that NVIDIA has is we are surrounded by extraordinary people. [Rene: Unquestionably. Yeah Unquestionably.] I mean, yeah. These are the finest computer scientists, the finest strategists and business people in the world. And they’re egoless and they want to do great things.
Jensen:[12:43]
And and so I think that that one, we start with that. I think the second part is we’re good at intuition. I think we have a good intuition about
Jensen:[12:53]
what problems need to be solved and how to get us from where we are today to becoming the company we want to be. And so I think our intuition’s good about what the what the various stepping stones are. And, you know, each one of the things that we did a lot of, I was asked, you know, why are we building SHIELD? I mean, what a waste of time. And I said, we’re going to be a systems company someday
Jensen:[13:13]
and all these systems are going to be connected to cloud services. Why go break our pick on the largest systems? Why don’t we go do this one first? And if we can’t do this one, we’re not going to do
Jensen:[13:23]
the large one. And so to create the conditions where the company could go go learn some new skill, fail, but not not damage yourself, you know, and so.
Rene:[13:33]
Can that only happen in companies where the leader is or was a founder? Because again, very, very few companies do what you just described, both in terms of being having clarity of vision, but also resiliency to continue to understand where to go. Is that – there’s been a lot written recently about founder mode versus manager mode, and obviously you’re a founder
Rene:[13:53]
leading a company 30 years later. It goes without saying the amount of success you’ve seen. But can this only be done, what you described, by the founder leading the company?
Jensen:[14:03]
I don’t think so. I think you’re doing great at Arm. You know, when I watch you do your work, I’m very proud of it.
Rene:[14:08]
Well, I learned from you, which is not [Jensen: Appreciate that.]– which is being truthful.
Jensen:[14:12]
Yeah. And I love watching you do your work and it makes me happy. It brings me great joy and pride. I don’t think so. I think that it is true: you have to have great resilience and you have to have perseverance. And I describe it as pain and suffering, as, you know, [laughter] and
Rene:
Teaching moments.
Jensen:[14:32]
You know, yeah, pain and pain and suffering is how you feel.
Rene: I felt it.
Jensen: And yeah, and in a lot of ways you have to get used to it. You have to get
Jensen:[14:42]
used to the idea that there’s pain and suffering involved. And, you know, that the journey to success is not about one achievement led by another achievement and another achievement. It’s not like that. You know, there are big setbacks, sometimes there’s embarrassing moments, you know, when you’re a CEO and and you haven’t enjoyed any of that yet. But
Rene:[15:02]
It’ll happen.
Jensen:[15:03]
Well I hope it happens, because it will be good for you. But, you know, all those moments are, I don’t know what I learned from it, but it made me stronger, you know, and I know I could survive it. I know I didn’t like it at the time. But when I look back on it, those are the moments where you
Rene: That’s where you grow.
Jensen:[15:23]
That’s right. You are most proud of yourself. You’re most proud of your company and that you
survived it. And so, so I think the company- our company is strong because we have
Jensen:[15:35]
lots of stories like that. You know, in the halls of this company are just are just filled with extraordinary stories of one set back after another, set back after another set back.
Rene:[15:47]
And with many leaders who– who went through it.
Jensen:[15:50]
Yeah, most of them are kind of like, oh, this isn’t nearly as bad as when that happened. [laughter] You know, every time something happened, it’s like, oh, it’s not. This is nothing. So the ability to to be able to go directly to remember when that happened, this is nothing. And yet this is incredibly painful. It’s it helps the company move through
Jensen:[16:10]
these challenging times.
Rene:[16:11]
So you and I have been around this industry about the same amount of time and some of the stuff that’s going on with AI, I know I feel this way, were things that I just thought I would never see. That the future generation would be able to experience the kind of transformation that seems to be taking place. It feels like to me not to sound Star Trekkie, but this
Rene:[16:31]
is the final frontier in terms of I can’t imagine what is next beyond what we’re seeing with artificial intelligence broadly. How do you feel about it? Are we accelerating so greatly
Rene:[16:41]
the transformation of industry that we’ve never seen before? Is there anything next after this? It’s just unbelievable what we’re seeing.
Jensen:[16:46]
I guess I’ve always expected that computers would demonstrate intelligent behavior. That we would be able to write software so well. And I thought we would write it, that algorithms would eventually solve problems in a way that seemingly the computer is intelligent. I never thought
Jensen:[17:07]
that it would result in an industrial revolution. And what I mean by that is, and you’ve heard me say this, that for the very first time, the computer industry has now transcended
Jensen:[17:17]
beyond the traditional computer industry, that for the very first time we’re now no longer a tool, an instrument. But we’re now a manufacturing industry. And so what I mean by that is, you know, right now, while we’re talking, our phones are in our pocket. It’s not being used. And when I’m not using it, you know, I’m not using this tool. It’s not
Jensen:[17:37]
doing anything for me. And most computers are that way. My laptop’s in my office, it’s doing that. Most people’s computers are that way. If you need that tool, you go use that tool. However,
Jensen:[17:47]
this new industry of AI factories, which is what we’re building now, they’re running all the time. Whether you’re using it or not, they’re producing tokens, they’re ingesting data, they’re producing tokens, they’re generating intelligence. Intelligence is being manufactured at a very large scale. And the idea that this computer used to be
Jensen:[18:08]
an instrument, a tool, is now a factory, a manufacturing thing, and that is producing incredibly valuable things at very large volumes. And so this is a new time
Jensen:[18:18]
for our industry. This has never happened before. And the idea that computers are now the manufacturing instruments, the machinery behind this incredible thing called tokens, intelligence tokens, is just an extraordinary idea. And so we’re at the beginning of a new industrial revolution.
Rene:[18:33]
Is it racing faster than you thought it would? And you have been closer to it than anyone with AlexNet and DGX1 and have seen the pace of innovation. From where I sit and we’ve been looking at it inside Arm quite deeply since I took over, it has gone far faster than I would have imagined two and a half years ago. Far faster than imagined, even a year ago. You’re involved in everything around it. Is it moving even
Rene:[18:53]
faster than you imagined?
Jensen:[18:55]
No, we’re trying to make it go faster. We’ve gone to a one year cycle, and the reason for that is because the technology has the opportunity to move fast. And because we are now not just building chips and we know that the rate of progress of chips anymore, if you’re lucky, with a new process node, you get a few percent.
Jensen:[19:15]
That’s incredible. And so how do we get X factors of performance with each generation, well the way we solve it is we designed six or seven new chips
Jensen:[19:25]
per system and then we use co-design to reinvent the entire system and invent new things like NVLink switches and new system racks that allow us to drive copper across the entire back spine of a system to connect all of the GPUs together in very large packages and 3D packages and such. We’re using all
Jensen:[19:45]
kinds of techniques to do that. As a result, we could deliver 2 to 3 times more performance at the same amount of energy and cost every year. And
Jensen:[19:55]
that’s another way of essentially reducing the cost of AI by two or three times per year. And that is way faster than Moore’s Law. And so you compound that over, right, five, six, ten years, we’re able to drive incredible cost reduction for intelligence. And the reason why we’re doing that is because we think that this is at a time
Jensen:[20:15]
when we all realize the value of this. If we can drive down the cost tremendously, one, we could do things that inference time like reasoning.
Jensen:[20:25]
You know, today when you use ChatGPT, which is an amazing service, I use it every day. I used that this morning you hit enter and your prompt is loaded and it generates the output. But in the future it’s going to iteratively reason about the answer and come up with a tree search maybe, and maybe it does some kind of iteration and reflect on its own answers,
Jensen:[20:46]
and eventually it produces an output. It might have gone through a hundred, a thousand inferences, but the quality of answer is so much better. We want to drive the
Jensen:[20:55]
cost down so that we could deliver this new type of reasoning inference with the same level of cost and responsiveness as the past.
Rene:[21:03]
I have seen a demo of the OpenAI model that does reasoning and it was shocking to your point. It went through a logic tree. You could see the tradeoffs it was making simply the way a human would, yet at a pace completely unlike the way a human would. But then as you fast forward and this is what’s so fascinating to me about this, what’s going on now, is that exactly to your point, you’re introducing systems
Rene:[21:24]
full data set and infrastructure at a pace the industry has never ingested at before. CPU’s bought every two or three years, they ultimately depreciated. Now you’re building systems
Rene:[21:34]
on an annual beat. People want to pay for those systems and deploy them as fast as possible.
Jensen:[21:39]
Right now we’re talking it’s so easy to say, but, you know, we’re delivering new computers that are this room size thing each year. It’s all the cabling, all the networking, all the switching, all the software. Yeah, it’s really quite insane.
Rene:[21:52]
Do you see it? And I’m not asking you to forward forecast. But this is more just a technology ingestion question. Can it continue at the current pace?
Jensen:[22:02]
Yeah, I think so. But it has to be done in a systematic way in the sense that everything that we do, we do in an architectural way. And what that means is that the software that you develop for yesterday’s clusters like Hoppers and that software is going to run on Blackwell and that software will run on Rubin. And the software that you create
Jensen:[22:21]
for Rubin is going to run on Hoppers. Well, this architectural compatibility is really quite vital because the investment of the industry on software
Jensen:[22:31]
is a thousand times larger than the hardware. Not to mention no software ever dies. And so if you develop software, or you release software, you’ve got to maintain the software as long as you shall live. And so the architecture compatibility that the idea of CUDA is not that, you know, there are millions of people programming to it. The idea of CUDA
Jensen:[22:52]
is that there are millions of GPUs, several hundred million GPUs that are compatible with it.
Rene:[22:58]
Software doesn’t die.
Jensen:[22:59]
Yeah. And so whatever investments that you make on one GPU, you can carry forward to all the other GPUs and all the software you write today will get better tomorrow. All the software we write in the future will run in the install base. And so, number one, we have to be architectural and really disciplined about that. Second, even at the system level, we’re super
Jensen:[23:19]
architectural now. We’ll change pieces of the technology to advance system design without you having to leave everything that you did
Jensen:[23:29]
yesterday behind. And so, for example, you know, when we first came into the data center business, a hyperscale data center had power distribution that was like 12 kilowatts per rack, while Blackwell’s 120 kilowatts per rack. It’s ten times, ten times the density. Now, of course, it’s ten times the density,
Jensen:[23:50]
and it reduces millions and millions of dollars of servers and compressed it into one rack. And so the amount of savings, energy savings
Jensen:[23:59]
and, you know, space savings, it’s just incredible.
Rene:[24:03]
That is very similar to our story. You know, the Arm architecture has been around for 30 years and we have software that’s been written for it for decades. And that is what people sometimes don’t always understand.
Jensen:[24:11]
Everything we do out on every single Arm chip, we care for it. We just showed something the other day. Somebody did some benchmarking and Grace was four times the performance per energy per watt than the best CPU in world. [Rene: Bravo.] And energy efficiency is vital. That everything.
Rene:[24:29]
Do you see anything architecturally starting to break when you go from 500-megawatt data centers to five gigawatt data centers? Just relative to the network latency, things of that nature, without getting into proprietary stuff, do you kind of a high level physics standpoint start to see some things that start to break?
Jensen:[24:48]
Everything breaks, physics is obeyed, which is the problem, but everything breaks first. Of course, we’re moving up the power density curve very, very quickly, exponentially. And so from 12 kilowatts to 40 kilowatts, to 120, 200, and it’s going to go beyond that. And so we’re trying to compress, densify computing
Jensen:[25:08]
as much as we can. When we do that, of course, liquid cooling becomes more efficient. When we do that, we can use copper for longer. Copper using electricity for as long as you can is good. So that you don’t have to
Jensen:[25:18]
hop cross electrical to optical. We’ll ultimately have to go optical, but it we’ll stay with electrical as long as we can. And so as much of the data centers we have, it’s more cost effective, it’s more energy efficient, it’s more reliable. And so that causes us to densify. The other benefit of densification is that all of the GPUs that are in the particular rack
Jensen:[25:39]
or in adjacent racks can behave as one giant GPU. It’s really quite amazing.
Rene:[25:43]
Amazing, yeah. One of the things I’ve always been curious about Jensen, is keynotes that you did at Computex that I remember watching that when you did and it was on Sunday night, and the sheer volume of content that you go through is not only unbelievable, but as someone else who does keynotes that are not nearly as long or is in depth, I just marvel at how you pull it off. Do you
Rene:[26:03]
do massive amounts of rehearsal for – remember back in the day when we worked together, I remember changing them at times the evening before and you still pulled them off. But now the level
Rene:[26:13]
of depth that you go into, particularly given that you’re talking about the datacenter architecture, you’ve expanded it out. How do you prepare for it?
Jensen:[26:21]
Well, we’re preparing for it every day. You know, that’s kind of the nice thing about our job is we’re not actors. [laughter] And so so those are kind of our job. We kind of live in it. And so we’re number one, we’re preparing every day. But a lot of what you and I do, frankly, is teaching. In order to shape an industry, in order to shape the market and
Jensen:[26:41]
to introduce new ideas like what we’re trying to do, a lot of it is teaching, you know, it’s not advertising, it’s not, you know. And we’re a platform company, meaning we can’t
Jensen:[26:51]
really do what we do without other people doing it with us. And so we’re about teaching, inspiring, showing, maybe demonstrating, and hopefully step by step by step, you know, we get more and more believers in CUDA in the beginning and NVIDIA accelerated computing today in joining us in our journey and AI and now the next big thing that we’re working on is
Jensen:[27:11]
physical AI and how do we teach AIs that obey the laws of physics on the one hand, and then also understand the physical laws on the other hand. So I think the journey is fairly long and so,
Jensen:[27:21]
you know, GTC and Computex are opportunities for us to do that, to celebrate our ecosystem and the work that they’ve done, teach them, you know, we’re sort of inspiring about the next.
Rene:[27:31]
Quite similar, we’ll do QBUs, I’ll do presentations and my chief of staff will say, gosh, the slides are easy. That’s kind of what you say all day. And I’m thinking, well, how could it be different?
Jensen:[27:40]
It’s still hard, though it’s to be honest, because we don’t practice. So, you know, it’s not because we choose not to practice. By the time we get all the stuff put together, there is no time to practice. And so we just grip it and rip it.
Rene:[27:53]
Jensen, thanks much.
Jensen:[27:54]
It’s great to see you. Good job. Everything you’re doing. Proud of you.
Rene:[27:57]
[chuckle] Thank you.
[music comes in]
Rene: I’m Rene Haas, CEO of Arm and it’s been a pleasure to have you with us today. And thanks again to Jensen Huang for joining us for our very first episode. We’ll be back next month with more exclusive conversations and insights from the world of technology. Make sure you follow Tech Unheard wherever you listen to your podcasts.
Rene: [28:18]
Thanks for listening.
[music rises]
Producer: [28:25]
Tech Unheard is a custom podcast series from Arm and National Public Media. Executive Producers Erica Osher and Shannon Boerner. Project Manager Colin Harden. Creative Lead Producer Isabel Robertson. Editors Andrew Meriwether and Kelly Drake. Composer Aaron Levinson. Arm production
Producer: [28:42]
contributors include Ami Badani, Claudia Brandon, Simon Jared, Jonathan Armstrong, Ben Webdell, Sofia McKenzie,
Producer: [28:52]
Kristen Ray and Saumil Shah. Tech Unheard is hosted by Arm CEO Rene Haas.
[music fades out]
Arm technology is everywhere technology matters, providing the foundation for advances in innovation and the future of AI computing around the world. See how we’re helping our partners around the world create a better tomorrow.