Power, Prediction, and the AI Infrastructure Race: AMD’s Keith Strier and Rhombus CEO Dr. Anshu Roy

Welcome to the Washington AI Network Podcast, I'm Tammy Hadad. We're coming to you live from the 2025 SCSP AI Expo in Washington, just one day after our inaugural AI Honors Gala, an unforgettable evening where we presented awards to eight extraordinary individuals shaping the future of AI, technology, and innovation. That day, the White House announced the new Center for AI Standards and Innovation and Commerce Secretary Howard Lutnik joined us at the Gala and unveiled more details of the administration's plans. To view the full AI Honors Program, visit and subscribe to the Washington AI Network YouTube channel. In this episode, I'm joined by the leaders of two companies who are at the Center of the Global Race for AI Infrastructure, Keith Strider, AMD Senior Vice President for Global AI Markets, and Dr. Anshu Roy, founder and CEO of Romba's Power. Let's begin. I'm here with my dear friend, Keith Strider, from AMD. Let's hear it for him. That's right, because if anyone knows anything about chips, it's this man, the history, the industry, you know it all. I'm so glad to have you here. We hung out here last year at the AI Expo. You're back and the whole world has changed in the last year. So, tell me what's ahead. Oh no, I'm not going to do that to you. That's not fair. But what I am going to do is I do want you to talk about the history of the movement and chips because now it's everything we were together last night. Howard Lutnik was at the AI Honors and he talks about chips and he talked about wafers. I never thought it'd be there with the Congress Secretary talking about it. Yeah, it's a little unfair question is, you know, I'm not an engineer. That's right. Not a chip designer. That's right. And if I'm totally giving it on the front lines, I have to string. I have to the last six years. But the first time I stepped into a data center and actually looked at a supercomputer, what was in 2019. Okay. So before 2019, I was focused on AI for a couple years, but really chips didn't come into view until about 2017-2018. But the story, but it's important to talk about it because when the world sort of, when AI became a mainstream topic in my, and this is my opinion, it was around 2016. And if you remember 2016, that was the year in January in Davos that the World Economic Forum released that white paper on the fourth industrial revolution. And they kind of coined this term, like, hey, we're in this new industrial revolution, there's new error. AI wasn't the only topic in that, but AI was really the one that grabbed everyone's imagination. And throughout that year, a number of things happened. That's when Google slew to South Korea and AlphaGo beat the total on that famous Go game. And that was almost an alarm. That actually set off alarms all across the world that maybe this AI thing is a little bigger than we thought. And so forth. So a bunch of things happened that year. That's when sort of AI hit my radar. And shortly after that, I made the observation. Well, okay, everyone's excited about AI and models and so forth. But you know, you really can't have AI without computers. And people don't seem to be paying attention that much to computing. It was sort of a dismissive topic. Oh, yeah. Computers are easy. You just get that from the cloud. You know, people really weren't thinking about it. So that's kind of one and hit my radar. I see. Okay. So let's talk about the cloud because the cloud changed everything. Cloud changed everything. Nothing is democratized access to computing and really just sort of technology as the public cloud, right? And I think it was a fame as the story famously goes. It was an engineer at AWS, right? Who said the Jeff, hey, we got some extra computing capacity. Maybe we should monetize this. You know, and that was like what 2007, 2008, you know, and and so that changed the world in many, many, many ways. And so we love the cloud. But, you know, computing isn't one thing. And one of the observations I made is that that, you know, everyone sort of has this impression that computing is infinite and it's easy. And that's only true for some kind of computing. It's not true for all computing. And obviously, the more advanced the computing gets, the less, the more finite it gets, the more complex it gets. Okay. So we were together also. People think we're going to travel with the world together. We don't, but we were at the AI summit today together at the Grand Palette. Oh, empires. Yes. Yes. And there were so many people there from around the world. It wasn't just technologists. That's right. It wasn't just companies. It was, you know, the civil society groups. You name it. They were all there. So now this has gone from the last thing you talk about AI to everything, to this expo that we're in right now. And how does that change what you're doing? Yeah. I mean, the conversation initially with AI was pretty sciencey. You know, it was just really, really nerding out on really cool models and how they could do cool things and solve complex problems. But that's just not the conversation today. That still happens. There are still, you know, a lot of nerdy conversations to having, but it's obviously much more complex and more nuanced now. And you have to have a really good conversation about AI. You have to involve all those different groups. And actually, I think President Biden did a pretty good job when he built the National AI Committee at the US. I served on that committee. I met Ely at that committee. And it was, you know, reasonably balanced between sort of civil society, industry, you know, that kind of government, you know, had different different sort of factions that kind of get all those voices in the room, right? And that's that's kind of the conversation required is a multi voice. It's a multi-stakeholder conversation. The world is not necessarily always good at that, right? You know, but that is what it is. So not to get political, but now we have a new president. Yes. And Ely, by the way, remember worked for President Trump and put together the National Security AI initiative, right? So now we've got states. We've got countries. We've got federal guidelines that members who we saw last night, members of Congress are working on. We don't have any answers. So how do you, how do you move ahead when the industry is just struggling to know what's coming? Not what's coming in the AI. That's all another conversation. But know what's coming in terms of regulation, how you work, and each, like I said, in each individual state or country. So you referenced the National Security Commission on AI and that's really, that is a underappreciated document. You know, Ely was the head of that. Of course, I was sponsored and I think catalyzed by Eric Schmidt himself and a number of other people and and that document was hundreds and hundreds of pages. But if you just read the first, say, 10 pages or whatever, the executive summary, it, it, that doc, and I don't remember exactly when that came out, right? It was a good four or five years ago that that became the basis on which I think a lot of our conversations today and legislation and the actions that are being taken even President Trump's EO on AI. A lot of that can be, you can look back and trace it back to the content in that original document. So that was a really, really seminal piece of work, frankly. And it also opened the country's eyes to the, frankly, the risks of not doing this well, right? It's not just, hey, we need to do AI because it's beneficial. It's a good thing. But if we don't, then in this broader context of strategic competition, right? Oh, and not just with China, but just with countries in general and just the, the, the, the, you look, you look at the title of our little session, right? The new calculus for competition. What leads to competitive anything, whether it's a country, whether it's a student competing for a new, for, for, for getting to a school? I mean, the nature of competition is changing because we have this ability to simulate and scale human-like cognition. That changes everything, right? Country, and, and that document was one of the first documents that very definitively said, if you don't do this well, it puts our country at risk. I mean, really, no one had said that clearly. So that was the beginning. So where we are today, right? And I think Trump, the first administration, the Biden administration now, you know, the new Trump administration has continued the momentum, thankfully, you know, of leadership and Michael Kratio's coming back and then Parker and all these great people coming back into the government. We, we have to lead, right? It's have to, right? And we have to do it not just, not just computing. We have to lead in computing, but we have to lead in energy production. We have to lead in the responsible deployment of AI, being mindful of the risks. I mean, you just have to lead in every possible way. So you're obviously at AMD, leaning the way, and you're not seeing on behalf of the company. But just to the point, again, the Commerce Secretary mentioned last night, power. Are you seeing changes in power so that there's enough power out there to make things move and to advance? >> We don't want power power or electricity power. We're talking, now I'm talking about electricity power. We're not really talking about it. Are these systems in place to support the AI that's being developed? Because even last week, keep just thinking the announcements from, you know, the hyperscalers. >> Yeah, they're huge. So, so it used to be when you talked about data centers, you talked about 100 megawatts as being big, right? Right? In fact, my colleague, Thomas, who's not here, Thomas Nicaria, who's a fellow SPP at AMD, who used to run Oak Ridge national labs, so one of the most prestigious important labs, and overseas, you know, the big computer systems like Frontier, he tells a story, because he's been there many, many years, and he tells a story, how I think around 2000, he pitched his boss at Oak Ridge like, hey, you know, I need 20 megawatts. And the guys like, you're crazy. I mean, who could possibly use up 20 megawatts for computing, you know? And he's like, you know, anyway, he tells a great story. And, of course, today, you know, we talk about 500 megawatts, gigawatt data centers, multi-gigawatt AI campuses, you know, we're in an older magnitude. >> We're in out of words, right? >> You need more words. >> Now, we did something that's above a gigawatt, you know, so, but you have these mega AI complexes, okay? And just to give you a perspective, one gigawatt data center is equal to a nuclear power plant. One nuclear power our plant, powers 30, 40,000 homes, right? So 30, 40,000 homes to one building, powering a centrally one cluster of computing capability. And we're talking about multiple of them, right? So yeah, this is interesting. Now in the world today, only three of those have been announced. You can call them Stargate level, okay? Stargate is sort of the term right there, open AI coin, right? There's three. There's one in Abu Dhabi, there's one here in Texas, and there's one in Paris. I'll put it this way. If you think, if you look at the whole planet, and you look at some very important geopolitical distinctions, you have things like how many nuclear countries do we have? But I think it's nine, is that correct? I think about nine countries, all right? Okay, so that's nine countries, and it took 50 years to achieve that. If you look at how many countries can launch a rocket that can break escape velocity into the outer atmosphere, that it's 12, okay? It took 60 years, and we have, well, I'm sorry, I did more than that, 100 years, essentially, from first flight, but we have 12 countries that can launch a rocket, that's it, right? Well, in the span of 18 months, we went from zero to three countries that will have Stargates. Wow, and the chances are there'll be at least three to five more in the next year announced, so what every country won't have a Stargate. So maybe 10 or 12 countries will be able to have the energy, the overall resources to accommodate a Stargate scale data center, right? So it's gonna be at that level, and so yes, at the end of the day, having that kind of infrastructure, the energy, the data centers, the computing, the ecosystem around it, this will be very determinative going forward. I went to the GTC conference this year, and you're an old employer, and I was shocked by the amount of people there, and it was more shocked because three of the people I met, and maybe it says something more about me, were pipefitters from different oil companies. Really, right? So I mean, I expected technology students, and all of that. So are the industries working together? Yeah, I mean, are you going to these massive companies and smaller companies who are LNGU name it, and saying, hey guys, we gotta work together. That's right, yeah, so let's talk about it. Let me start with the video, 'cause Nvidia's an important company, and I joined Nvidia in 2019 after going to the GTC in 2018. Oh, okay. And I actually met Jensen, he wouldn't remember that, but I met him at the time, and then of course a year later, I ended up working for him, but it was very clear in 2018 that he had a very comprehensive vision for the world as it relates to end-to-end AI infrastructure, and the rest of the world was waking up to it, but he was first, and when I joined in 2019, what I then learned is that he was also leaning into industries, and there were public about it, but not everyone was paying attention, 'cause no one cared about Nvidia in 2009, okay? So, but in video, not only was focused on AI, but they were focused on AI for industry, and so there was an oil and gas group, there was an energy team, there was a life science team, there was an auto team, and I ran public sector, I was hired to build the global public sector in the sovereign AI, as you know it today, and so there were different industries, of course, now we just expect that, but that's why you ran it to people like that there, so the industries are absolutely leaning in. They have to, it's the future of industry. >> No, I think that's true, so let's talk about governments, how do you deal with all the different governments? I mean, you're going around the world every week. >> Yeah, so, again, going back to the dozen, I mean, every company, are they asking you, hey, what's going on in the US? It's the latest, so here's where the conversation started. If you go back in time, to 2019, when I started on this journey, every, you know, the first conversation every country had in the world on what is the GPU, and how to use it started with me, and I had those conversations, there were really simple conversations. It went like this. Who are you again? (laughing) In Navidia? Is that a gaming company? I think my grandson, you know what I mean? And what's the GPU? That's how every conversation went, right? 'Cause no one really knew, no one really cared. So, you know, a lot of them-- >> There's no one believed him. >> No one believed him. >> I am believed that there was no urgency. >> All right, okay. >> So, so from 2019 to 2020, to November of 2020, well, no, we'll go all the way to the spring of 2023, essentially, in that time period, all those conversations were friendly, informational. There was momentum building, but no fire was lit. You know what lit the fire? Yeah, Chatchy PT, right? Okay, so, you know, a Chatchy PT hit, right? That hit in November of 2022. Two weeks later, and this was just a coincidence. I happen to be at a meeting called the OECD Digital Ministerial, which happens every few years, they bring all the ministers of digital together. This was in the Canary Islands off the coast of Africa, and Chatchy PT was not on the agenda, but when I showed up to the meeting, it's the only thing they want to talk about. I think this was like December 6th. So again, two weeks later, right? It was completely dominant. In fact, when Carmée, the Secretary of State of Digital for Spain, opened up the meeting, she literally opened up with, have you used Chatchy PT yet? So, it was a dominant, I came back for that meeting, and I sent an email to the leadership team, and I said, I think this is bigger than we think it is. Right? I said, this is the only thing people want to talk about. Because it was understandable to every one, relatable. It was easy, right? If you look at the laws of diffusion, and you look at what leads to the rapid laws of diffusion, right, you make something accessible, relatable, easy to use, there's certain things you do. Open AI crushed it, and so they put a really simple interface and a really complex technology, they made it free, and everyone was like, it took off, right? And you know all that goes, 100 million people in 60 days or whatever it was, and I think as of today, they've six, 700 million users a week, and it's crazy, right? The adoption level is fast, but it did blow the hinges off of our imagination, as to wow, I guess this is actually a real thing. This isn't maybe not a toy, not a game. I could, people could see the future, actually Aaron Levy, the CEO box, so I do not know, but I respect, and I read what he writes, and he has this great quote, he goes, when he experienced that, it felt like that rare moment, when you could actually potentially see the future, even before the future's arrived, you know? And I agree with him, that's how I felt. Obviously, Jensen felt the same way, and he made his famous quote in early February. I think that's what it was, that this, yeah, I phone moment of AI, and then everything changed, right? And so frankly, to answer your question, up into that moment, the conversations were friendly and easy. After Chatchy PT, in the spring of 2023, there was a fire, right? And there was real urgency, and people were in a mad rush to buy GPUs, which was a good thing, right? Obviously, yeah, go through your business. For the business, yes, yes. So things scaled very fast from there, and then, you know, that's how history is written. I mean, basically, it went from a couple hundred million to tens of billions, like that, as far as government sales go, and every country went from, okay, AI is important. Glad we have some policies to holy cow, we need a scale, and that's what led to where we are. Okay, so what, by the way, for those in the audience, just so you know, the audience had, it's called a silent audience, they're wearing headsets, but they're nodding, and they're cheering again, right? (laughs) Yes, they're here, and we're gonna come, and if you see that microphone over there, you can ask Keith questions, but this is my question for you now. What is the craziest thing anyone ever offered you for a GPU either at NVIDIA or at AMD? Doesn't everyone want to know that? What's the craziest thing you were ever offered? Where do they want to take you out to dinner? What did they offer you? Did you get a car? No, no, no, no, a house. It wasn't that they offered me something, it was more of an awkward moment, okay? I was very new to NVIDIA, and I had not totally clued into how, no, this is more of a gaming thing than a country thing. Okay, I have two answers to the question. So first, I was very new, and I was wearing an NVIDIA backpack. It is a true story. I was standing in front of the White House, 'cause I was actually in Washington DC, and I had my NVIDIA backpack, and a young boy came up to me and pulled on my backpack. I mean, he could have been more than 12, and I turned around, and I look at this kid, and he goes, "Sir, do you work for NVIDIA?" And I said, "Yes, can you get me a GPU?" No, I'm like, "I don't even know who you are." And you're asking me for a $600 gaming GPU. He's like, "Is there any way. " Like, it was so sweet, right? It was so innocent, but I didn't realize how impactful the logo was until this kid pulled on my backpack, 'cause there's any way you get. So that was first off. But now, jump ahead, a couple of years, now we're in the post-chat GPT craziness, and I have to be careful. I'm not going to mention the name of the country, but there was one country in Asia, it was not China, that went from not having any interest in GPUs to having true urgency, and the Prime Minister of that country sent a note to Jensen, and just basically said, "You must give us GPUs," and insist it on a letter, where basically we agreed to give them what everything needed. And honestly, I thought, at the time, I thought our answer was very balanced and responsible. I said, "Look, we are happy to work with you, "and we want to be responsible. "When the purchase order comes in, we'll fulfill it." Like, but it was a very strange letter, because it was almost like, "Please give us GPUs." And I was like, "Of course, we didn't even know what to say." But I think it was indicative of this sense of urgency at the time that what if we can't get what we need? And honestly, I think in video and AMD and the industry have stepped up really well, and I don't think many countries are necessarily, like we're not the problem, okay? It's a resource issue, there's an energy issue, but it's not the chip industry. [BLANK_AUDIO] We're not failing to provide the chips they need. Are they ready? Do they actually have readiness to take them and use them and scale them? That's the bigger issue. >> Right, you should talk about that a little because I don't think people that aren't deep in the industry what you need to make those GPUs work. >> Yeah, I mean, these are the most complex things in the world, right? I mean, we're talking about things, you know, the size of the palm of your hand, that have hundreds of billions of transistors that only work inside of Michigan. I mean, people focus on the chip, right? And the computer has tens of thousands of parts, right? You have incredibly complex interconnect and high bandwidth memory and obviously CPUs and GPUs and so forth, and it's no longer just even about the server. It's about the entire data center as a computer, you know, and then connecting up and operating at unbelievable scale. The amount of capability and technology and expertise you need to operate these systems is mind blowing, which is why CoreWeave, that one public this year, and the only thing that they do is stand up and manage large scale clusters. They're worth nearly $50 billion today, and that's all that they do. So this is a hard thing to do. Right, but it's also reminded that there are so few people in the industry, back to your earlier example. There's like three companies, two people. Hey, guys, you guys can get into this business. It's unbelievable. Okay, if you have a question, please walk up to the microphone. If you would, we should put this on and say, for any questions that are coming in. Yeah, you know, I think it's, look, designing chips is one thing. Manufacturing them at scale, at the quality needed, the yield needed to do it, which is why TSMC is so so powerful and well regarded company. It's because, you know, it's super hard to do. Right, you can throw lots of money at it, and that's not a bad thing, but that does not guarantee you success. And that's why there's really not that many places that can do it. So in these early months of the Trump administration and the announcements that have been made of companies working in the US, well, that's speed things up. I don't mean for the US, but overall for the industry. Oh, we're going to get to you next. Yeah, look, it's actually been really good. I mean, the Trump administration policies on a few different fronts have been very welcomed by our industry, right? I mean, the lay on the AI to fusion roles was very welcome. I know there's going to be more roles clearly, you know, but the way that they were written was not really welcome to most of us. We didn't think that they were productive. The executive orders on AI and publication and literacy, huge, they just, it backly should have been done two years ago. Right. Yeah, I mean, we're pretty supportive of the movement so far. Okay, sir, can you please say your name and your affiliation? Sure, Daniel Bennai Middle East Institute. Okay, great, hi. Hi, thanks so much. I just wanted to ask a question because it's a question that I think is driving a lot of the policy debate around the engagements in the Middle East about where we are when it comes to cutting edge chips with relation to China. How far ahead does AMD feel like it is compared to Huawei ascend and some of the others who are making the kinds of chips that you would ultimately be competing with in the chips that you design and sell around the world? What is that gap? Are you worried that does China have the capacity to actually export the kinds of cutting edge chips that it has or are they all needed domestically? How do you see that picture? Well, first of all, I am not qualified to answer your first question. So I'm not a chip engineer and I've never evaluated Huawei's chips. I couldn't possibly tell you if they're good, better, worse, or almost close to ours, I couldn't tell you. What I can tell you is that I'm not sure it matters. And I've said this and other people have said this, it's not my, I'm not the sole holder of this opinion, but the performance of chips matter, okay? And I'm referring to like the actual efficiency and the amount of intelligence that can be produced by a chip geopolitically it matters if money and carbon matter to you. What do I mean by that? That means if you're in a company and you're operating on a budget and you're competing with other companies on a budget, then yeah, you need the most efficient performance chips you can get your hands on. If you care about the environments or you're a country and you want to have a green economy and you don't want to just burn coal and so forth, then yes, the efficiency of those data centers matter, right? But if you don't care about money, let's just say money's not the issue and you can spend as much as you need. And you don't really care about lighting up a few more coal plants and efficiency's not the main issue, then the performance of the individual chip matters less because if I need three chips or four chips or five less efficient chips to achieve the performance goal that I need and that cost me more money and more coal, that's fine. You know, it's so in a geopolitical context, that's different than an economic context and that's where I'm not sure it matters. - Thank you. - You're welcome. - Thank you, Chris Russo. That's Chris Russo from Hadad Media. Thank you, Chris Russo for being here. - Yes, Chris Russo from Hadad Media. And last night, we had an event with the Secretary of Commerce Howard Lutnik and he was talking about embracing our allies in the chips race. So I think it was kind of an answer to a response to the recent Middle Eastern summit. I think maybe talking across both companies, your former employer and now, like has that been your motto previously or is there a new push for that to embrace our allies? - Yeah, I think both in VDAMD are US companies that are loyal to the US. So I don't think there's any question about that. At the same time, the semiconductor industry and this includes Intel and Qualcomm and Qualcomm have been actively selling into the Chinese market for many years. It is the second biggest market in the world. It's an important market. It's only become complicated recently. Up until recently, it was not complicated in a sense. It was just another important market. So I can tell you that at Hadad Media, I did not travel to China ever. So I haven't been there since 2018 because I was very focused on serving the allies. That was the nature of the project. So the AI nations program was an allied program and I assume it still is today, I'm not there, but that was the nature of it. So we did not focus on China. And I don't focus on China and AMD either. So China is a discreet market that gets served clearly but I've never personally focused on it. And so all the things that I've done and the very nature of sovereign AI, its genesis was in working with the allies, quite frankly. So that's just, that's what I can say on that. - There you go, okay. Keith, we've run out of time. I can't tell you how much I enjoy this conversation. - That's been almost 30 minutes. It's been 30 minutes. Let's hear it from Keith Strider from AMD. Great work and I want to talk to you about a GPU. I'd like to get my son in the front wrap on the way. - Yeah, let's talk about my bag first. - Yeah, there you go, thanks again. Thank you so much. - Great to be here. - Thank you everyone for being here at the AIX stuff. - Now, Dr. Anshu Roy joins the conversation to explain how Rombus Power uses AI and commercial sensing to forecast instability, support national defense, and shape strategic decisions across sectors. - Our friends at the special competitive studies project have done it again. There are 14,000 people here and I am so lucky to talk to one of the most important CEOs here. From Rombus Power, Anshu Roy. Dr. Roy, I'm so glad that you could be with us today. - Thank you. - Thank you very much. Because when you talk about why all the people are here and the work that's being done, Rombus is ground zero. You started all, take us back to the origin story. - Oh, yeah, sure. Thank you, thank you for having me, Tommy. The origins of Rombus, I started this company in 2011. See what happened was I was well on my way to become a professor and I had some awesome job offers in 2010, I had just come out of a Nobel laureate's lab. I published very well. I've done everything that one is supposed to do. What I hadn't done was really give hard, it's hard to whether I really wanted to become an academic or not and whether I would be a good academic or not. And when I did, I realized that I would make a very bad academic because I don't have that level of patience that you really need to both teach well and do research well. So what could I do with all my restless energy? And I think the answer was obvious. I had to start a business. I come from an entrepreneurial background. My father's an entrepreneur, my brother's an entrepreneur. So it was a natural thing. I started the company in 2011. First five years were spent just inventing and commercializing a solid state neutron sensor in Fukushima Daiichi. It was very successful. So that's for industry, government. Who did you start with first? Well, okay. You guessed it. I started it with the government and it didn't work very well because I really had no idea how DC works. Right, I mean, I was just a kid, I had no idea. And I made multiple trips to Washington trying to sell my neutron sensor. I used to carry it in my bag. How big was it? I was pretty big. Oh, okay. And people are very gracious in America. They'll give you the time of day. They'll talk to you. But you have to know the process after that. And so I didn't know all of that. And so I was very, very unsuccessful in selling. Very successful in making unsuccessful in selling to the U.S. government. at that time. And then I got this phone call from the Japanese. We are Alan Header, who's my, who was my Nobel laureate, post-doctoral advisor. So you started with the Japanese government? Yeah, and then what I did was I realized that when in 2015 when D.I.U. started and just opened it's right, the defense innovation unit at Pentagon. And so I knocked the door and that's when I met my good friend, Bucky butto, who's now a major general butto, of course. He was Colonel back then. And you know several cups of tea later we became friends and it was very obvious that we had something to say and we could contribute in meaningful ways. And and D.I.U. did an amazing job at helping us connect with problem owners. And what's that? Well people who have in the D.I.U.D problems they want to solve. Right. So that could be the Navy, the Army. Yeah, the Air Force. Yeah, anyone. Intelligence services were right. Right. And we we were very lucky to have that interface. And when we did have that interface, some of them saw were able to express what they really wanted. Defining a problem is the hardest thing. If you can ask the right questions and actually answers come relatively easily. But but defining a problem is an iterative process as you know. And what I enjoyed very much in those years was this idea that we could iterate with them just on the question. This was a radically different way of doing business for the Department of Defense. And it's all thanks to Secretary Ash Carter's vision. And he he was really taking us back to the 50s, which is how we did business back then when all the breakout efforts like Corona and so on were happening. And in other words innovation comes first instead of here's what we've been doing all these years. And we're going to keep doing it. So I'll edit that a little bit. Yes, innovation comes first but innovation for urgent problems. Okay. That sense of urgency is absolutely critical because that ensures programmatic funding down the earth. Innovation's tend to die mostly because they are resource start and they're resource start because somebody in the budget making process decides that they're not important in us, right? Right. For what a variety of reasons political or whatever that might be. We were very lucky to actually be able to understand that through a man named Samaraki. He was my intellectual guard father. And he retired as CEO of Lockheed Missles and Space eventually. And he always used to say rush to learn to me rush to learn rush to learn. And that sounds like where it's to live by. Yeah, people say in the valley rush to fail and he's like, no, that's wrong rush to learn. Yeah. So I am blessed to have met someone like that. So now let's go to today. So now we have this. I was going to say brave new world. I don't think that's right. I think we have this world where everyone is rushing to work in public sector. Yeah. And even we're surrounded here at the AI Expo with people with all kinds of ideas. Yeah. You're already in there working with them. Do you take ideas from other companies? I mean, do you have people you want to work with? Because just to your point. And you can't say it enough. It's so hard to get into defense and intel and have the relationships. It's not about friendship. This is about standards. Yep. And rules. Yep. Right. So do you collaborate with others? Or how do you do it? We collaborate with over three dozen companies. We have a lot of different partners. Our system is fundamentally for what we call predictive decision support in real time. So we the ability to sort of see the impact of decisions that are being made and to be able to see them in real time using artificial intelligence in a variety of different ways. So give us an example. For example, let's take a warfare. In a recent conflict that happened in Asia, you're probably tracking that. Where would where would one country attack? Let's say Pakistan with their attack India. Where would they attack from? What kind of what kind of attack would it be? And let's say I could tell you that. I could tell you when that would happen. I tell you that three days from now it would happen. Okay. And then if they sort of increase their defenses, then how should they do it to be most effective? That is the kind of question we are able to now answer quite effectively and operationally. Well, sadly, you predicted the start of the war in Ukraine. Yeah. We did it to the day. We were off by one day. Right. And some of it is also, you know, some of most of it is just mathematics. I mean, it's a patterns of behavior that machines can pick up better than any human can at scale. And and the ability to sort of no matter which country you are, you cannot operate large militaries without certain. Right. And no matter how good you think you are a deception, you there'll always be things you'll give away. And so it's those pieces we pick up very easily through commercial sensing, no government, including the US government. We'll ever be able to match the combined power of the commercial world to pick up what's going on on our planet or in space or wherever or in the cyberspace. Even China. Not even China. Not even close. They cannot afford it. They would go bankrupt because the total corporate strength, the entire sort of the investments on collection for not for intelligence, but for business purposes that has been done is so spectacularly large that no government can afford it. This was not the case in the 80s or the 70s or whenever, you know, these when when it was quite the opposite. But with the with the internet came the cloud for rapid communications, incredible open source explosion of software. That that there was a monetary incentive to create this kind of ecosystem of very small companies that could eventually become some would become very big companies and so on and everybody wanted to become that big company and everyone was running for that. And so that's that's capitalism and that's that's great. So with that led to what we are now beneficiaries of, which is we we are able to source over 5,000 streams of data into our platform from satellites to sensors and cell phones and, you know, you name it. It's in their social media, a dark web, all kinds of things. When you combine all of that intelligently for addressing questions that everybody is asking, there are some basic questions we're all asking, right? Where would certain countries attack? When would that be? Which countries might go politically unstable? Which where would conflicts happen? Where would black swans happen? What could they be? Everybody's curious to know questions. So way before Sam Altman came up with chat GPT, you were collecting data. Yeah, we were. And you pile it on. Yeah. So foundational models like Sam Altman, OpenAI, Claude, all of these mistral, whatever. They're all very very important. What what's needed underneath them though is data. Very curated connected data in order for you to generate reliable intelligence. Well, the reason I bring it up is because chat GPT did kind of change everything. I wonder if it's made your industry more competitive. Sure did. And it probably helped you with your business because you're probably saying like, hey, we've been doing this for years and these guys only have this tiny part, right? Did you have a lot of meetings like that? Yeah, I mean, we did. Initially it was, it was different. Initially it was like, oh well, you guys are irrelevant now because we'll go to OpenAI for everything. Very soon, of course, that lasted for about a couple of months. And then people quickly realized that the chat GPT and Claude are very powerful genetic capabilities. But for doing specific things, accurately, you do need other kinds of face. So last night, we gave an award at the inaugural AI honors to Booz Allen engineers who created something called Space Lama and essentially with NVIDIA chips and with the help of meta with Lama 3.2, they actually created a, for Black and better way to say it, this chat GPT or Space Lama for the International Space Station. So I [BLANK_AUDIO] to Senator Mark Kelly, who was an astronaut, and I said, "Hey, have you heard about space, Lambe?" He's like, "What are you talking about?" I said, "You no longer have to carry manuals on the International Space Station." And he said, "This is an incredible idea, because when I went up in space, we had seven feet of manuals. We had to get in the capsule, right?" So it sounds like you're taking even more. My question is, with open source, has that chow, has that changed things for you? Because that's the little guys that can jump in, or you don't see that as competition. Well, no, no, no, no, no, no problem, and little guys should jump in. That's what an ecosystem is for. We've been doing it for 10 years, we have a head start, if we don't stay, work hard enough, then we won't have that head start for too long. And that's just the nature of our business. So can I talk about the GPUs? Because I just talked to Keith Thryer, who used to be, he said AMD, he used to be at NVIDIA. And I love this story of NVIDIA that Jensen Dwang was talking about these chips and GPUs and everyone else was talking about the CPUs, and they had this huge bat. But other people did not have that bat. Were you aware of that? Did it, I mean, just not towards your business, but just in the overall ecosystem, AI, did it make you nervous, or did you think, man, that guy doesn't have any idea what he's talking about? Well, well, so a lot of stories get built in hindsight. So, you know, I think, I think we should be careful of, I'm in that business, by the way, be very careful. I can sell to HBO films. That's one of the things we like to do, but go ahead. So it depends on the narrative you're trying to portray. So I think it's a, that's an interesting way to look at it. I didn't look at it that way. You know, the man was making GPUs because the gaming industry was exploding, and that's all there was to it. And it's so happened that the neural networks were far more efficient with GPUs because they can operate things in parallel, and that's all there is to it. Parallel computing, the world of parallel computing was already in place, but it used CPUs, and it was a tiny market where nobody within their right minds would ever invest. Right. Gaming, highly addictive, completely consumerized. Of course, let's make chips for that, and that's how, that's how this business works. And then, of course, you know, machine learning came in, and when the AI explosion happened, and now we're seeing what's, we're seeing it won't last though. You don't think? Well, good. It's cute. It actually becomes commodity rate. That's, that's… Oh, I don't know. You say that. I don't know that. It makes sense, though, but I don't know if you know Jack Hittery from Sandbox, AQ, and he talks about GPUs. Yeah, sure. So what happens with that? What happens with that is that the nature of problems you can solve will change dramatically. The nature of problems that quantum computing is built for is quite different from what artificial intelligence tries to do. AI is, is a massive correlator, okay, and it does a very good job of that. Not all problems are correlation problems as it turns out. So what we will see happening is that we will understand our sciences much better. I think we will be able to use that knowledge to do compute more effectively, such as optimized logistics much more effectively. That's one example. Good. Yeah. What's that? Medicine. Oh, yeah, absolutely. It will be an incredible revolution in medication. Do you know science doesn't understand how talent all works on the human body. It just works. Yeah. So that's why we call it drug discovery, not drug creation, right? We discover. And so in as long as it is discovery, correlation tools will help you. So AI will help you. When you start to develop an understanding at a scientific level of how drugs can be created in more intentional ways to do certain things through chemistry and you understand biology and physics to that level, then it will be called drug creation. That's right. Okay, I have to ask you about a genetic because there's been a lot of conversations on the streets. I like saying on the streets about clean data versus like crappy data. Yeah. So crappy data like Twitter, right, when it's just like people yak and up a storm about whatever it is versus the kind of, I feel like you have all the clean data in your pocket. I can't see it. Can anyone see it? Does he have clean data in his pocket? And so with the gen tick, right, you've got to have clean data to be effective. How do you look at that? I look at it slightly differently. So the first of all, when you go to X or other social media sites, any data in our business, secured national security business is actually very meaningful because it reveals certain signals, even noise reveals signals. So what people blabber on the social media and other places or what different mainstream media bubbles do in their own little worlds is very, very important for us because those are nothing more than information narratives, those are that's the battle of the narrative. And we actually learn from our system learns from that. And then it tries to generate, depending on what kind of messaging you're trying to create, it can do that. Okay. Okay. In national security, that's a very important piece. And yes, we have a whole bunch of clean data that we have curated over the decades. You call it clean data. We call it intelligence data. Intelligence. Intelligence? Intelligence data. Okay. Yes. Because it's connected. It's curated. Yeah. It's just amazing. Only read media intelligent data. That's what you say to the staff, right? No more of that other data, right? And I can see you saying that to the team. You're right. You can see. Okay. Sure. No. We do that. So you ask me how I think about a genetic genetic can be applied to any one of those things. If it's generating what you see as crap, then you probably asking the wrong question. That's what it is. And it's just a mismatch of what you were asking, this is what it's giving. There's the signal in everything. What do you think the most important thing is for people to look, look for? Upcoming. Like what a piece of it. So let's talk first about the national security piece because there is so much going on geopolitically. We're not getting into specific things. You pointed out one of the incidents. What should people look out for that? And then if you don't mind, even though it's actually a primary area in the larger picture of what you see on AI applications, but let's start with national security. Let's head. And what are you looking for, Tip? I think a lot what I'm looking forward to is more, more of, more adoption of the capabilities the V at these companies like us have already developed. And I think in the current environment, it probably is going to work. And that will be a very exciting next step for us, where people actually use would be. But is that because the more sophisticated people are, the more you can accomplish or it just helps the world to see how it works? I mean, you're talking about business, but I'm also wondering if you think it helps everyone because I feel that AI has, you know, people worry about jobs, I worry about jobs and all of that, but I just feel that it's moved everyone to the higher level of more potential success. I agree with you. I think what I'm looking forward to is better messaging of that. Okay. Better messaging. Oh, we have a communications person the front row, we'll talk to her about it after. Yeah. And I think that would be very, very important because everybody's very concerned as they should be when they don't know something. So those narratives will be very helpful at a consumer level so that that will help everybody including businesses do their thing much better. And then on the national security side, on the national security side, I think things will get more predictive, more anticipatory, winning in the cognitive domain will become easier. The point is not to go fight wars, I mean, it's to win it if it's must be fought, but the point is to actually stop wars if we can. Do you have an opinion in human in the loop? I do. Yes. And the cost of the decision is very high, then you absolutely need humans in the loop. Cost meaning in human lives or whatever that might be where you, it's even not even measurable cost, right? For low cost decisions, I think it can be delegated to machines like buying stationary or whatever. Okay. One more question because we're running out of time and I know you're very busy man, China. What do you worry about? Where do you think their biggest opportunity is to catch up with the US? China is very hungry, and when a country is very hungry, it kind of figures it out. And so with the leadership that's so top-down, I think that's what I think. where it is going to catch up with the United States is I think it'll begin to catch up in innovation. They're very much an innovation starved country and generally culturally they are not really programmed in that way because of a variety of historical reasons. That may begin to shift. But well because of the incentive structures that come top down and I think we need to pay closer attention to that part of their efforts because I think they could outpace the United States if we are not careful. Well they're outpacing by the way they may outdo us. I see because you do you have at least those 5,000 inputs coming in at Rambus right so you could do a little prediction. Absolutely absolutely absolutely if you're willing to pay for it. Yeah, if you're willing to pay for it I like that and she Roy let's hear it for him. Thank you so much for being here. I appreciate the work you're doing. Thank you for being with us here at the AI Expo and thank you for listening to the Washington AI Network podcast. Be sure to subscribe and join the conversation. The Washington AI Network is a bipartisan forum bringing together the top leaders and industry experts to discuss the biggest opportunities and the greatest challenges around the AI. The Washington AI Network podcast is produced and recorded by Hatted Media. Thanks for listening. [BLANK_AUDIO]