The Pentagon-Silicon Complex and the Rise of the Neo-Primes

Welcome to the Cogs of War podcast. I'm Jonathan Panter, Executive Editor of the Cogs of War Vertical, brought to you by War on the Rocks and Booz Allen Hamilton. Today, I'm joined by Sharon Weinberger to discuss her recent book, Valley of Death, How Big Tech is Fueling the Future of War. Sharon is the Editor-in-Chief of the Bulletin of the Atomic Scientists, a role she assumed in August of this year. She was formerly National Security Editor at the Wall Street Journal from 2022 to 2026, and has contributed to many publications, including the New York Times, New York Magazine, and the Washington Post. She's also the author of three other books with incredible titles like A Nuclear Family Vacation, Travels in the World of Atomic Weaponry. Sharon, thank you for joining Cogs of War. Hey, thanks for having me. First of all, congratulations on the new role at the Bulletin. It's a publication I admire very much. I'm curious what brought you there and perhaps why now? So a couple of reasons, some personal, some personal. I had been at the Wall Street Journal for four years, a great publication. I'm now sort of on the other end of rather than an editor, a reader. I think they continue to do a great job. But my background as a reporter and also as an editor has always been very focused. As an editor, it's been broader, but as a reporter, it's always been very focused on the intersection of national security and science and technology. That's been sort of my sort of lodestar for probably much of the past 25 years. And at the end of my time at the journal, I loved it and was enjoying it, but I wanted to step back. I had a looming deadline for the book and was offered a fellowship and was thinking that I wanted to step back from daily news as much for a time period and really focus on some of the underlying issues I cared about, of which the book was one. The book was in some ways a sequel. Sort of each book I've done is in some ways builds on the other. The last book was on DARPA. This book was looking at where things were going, where I saw things going in the world of national security and technology. So I decided to leave the journal, go on a fellowship, finish the book. And then my thought was, I want to go back to a smaller newsroom focused on national security and technology and really go in depth on the issues that I care about. I thought maybe a year, year and a half, something like that might come. And literally, I think, around the same time that the book was put to bed, this opportunity came along and it was exactly what I was looking for. For those familiar, even for those not familiar with the bulletin, it originally focused on nuclear weapons, nuclear issues, but its mandate is existential risk, anything that could pose a risk to humanity, ending humanity. So today, those include AI, other forms of disruptive technology. Biothreats, climate. And for me, these were exactly the issues that I have focused on in my career. What does technology mean for us, for national security and as a society? So it was just kind of perfect timing. And it's a it's a great publication. I know you've only been there a short time, but have you noticed any differences from working at a nonprofit focused on existential risk versus being in a large newsroom? A place like The Wall Street Journal brings tremendous resources. To bear on an issue when Russia invaded Ukraine, when Hamas launched attacks on Israel, and then the outbreak of the war in Gaza, a place like The Journal, which has so much talent and resources, can really cover the issue from so many angles. The advantage of a smaller newsroom, and I've been at a number of them at Foreign Policy Magazine, which has a small newsroom that I led for a couple of years, The Intercept back in the day. When it was focused on investigative journalism, it still does to some extent, and still some very good reporters there. But at the time, I was leading the national security coverage and the Snowden archive, but it was a small newsroom. So you don't have as many resources, but you can concentrate them on a specific area. So when I was at The Intercept, a lot of the focus naturally at the time was on issues related to surveillance. When I was at Foreign Policy at the beginning of the first Trump administration, it was looking at sort of the massive changes happening on the National Security Council at the White House at the time in areas like U.S. Development Assistance, not comparable then to what happened in the second Trump administration. So you're much more nimble and you can pick and choose your issues. So both big and small newsrooms have advantages. At The Bulletin, it's again another focus, this idea of existential threats, really thinking about what are the types of technologies and issues that can really bring an end to the world as we know it. You know, every newsroom is different, but I enjoyed being in a large room, and I'm also enjoying once again being in a smaller focus newsroom. Well, I won't ask you whether you believe that AI is as existential a threat as nuclear weapons, though we might get back to that. I'd like to get to your book. So again, the title of the book is "Valley of Death," and you just mentioned that you see it as part of a trilogy, with your previous books. In some respects, yes. So my first book was what I would call like a smaller book. It was about sort of a crazy idea that came out of the Defense Advanced Research Projects Agency for sort of a nuclear hand grenade, a next generation nuclear weapon, looking for energetic materials that are far more powerful than conventional weapons, but not quite as powerful as nuclear weapons. And it was really about one crazy project gone wrong and what that told us about when pseudoscience or fringe science or even just controversial science takes hold in the Pentagon. And I looked at it as a metaphor for larger issues and controversies. But some of the pushback I got from that book from people inside the defense establishment was, well, you cherry picked one project out of DARPA and sort of drew lessons out of that. And I said, fair enough. So then my next book was an examination of DARPA, of its entire decades of history. I went into its archives. I interviewed all the living directors. I sued the government and won for interviews of DARPA directors at the agency themselves had commissioned. I went through Vietnam War records at the National Archives and created what I wanted to be. And I hope I achieved a very complex portrait of an agency that has contributed amazing things to national security. I think most notably ARPANET, which became the internet, stealth aircraft, the origins of the drone revolution of driverless cars, but also introduced the rainbow agents, Agent Orange to Vietnam, had a disastrous legacy in many areas of counterinsurgency in Vietnam and has some notable failures. And so the idea was to look just as people told me, well, don't cherry pick the failures, which I agree with. You can't cherry pick the successes. You have to look at this agency as a whole. And I think I came to the conclusion that it has some amazing lessons to be drawn from it. But there are also roads that went down that were very perilous. And the most perilous of all is looking at technology as the ultimate fix or cure for your national security problems. So that led me into the current book. Great. So would you give us a quick thesis of the current book, and then we'll talk about some more specific parts of it? The DARPA book came out about 10 years ago. And as it came out, I always look for what's different, what's changing now. And at the defense shows and arms shows I was going to, I was beginning to see, along with big banners and advertisements for Lockheed Martin and Raytheon, now known as RTX, Northrop, et cetera, I was seeing banners for Amazon Web Services and Google and then Palantir, which probably many of your listeners know was sort of a Silicon Valley founded defense company. And I thought that's interesting. And I think that a lot of people who follow national security know that Silicon Valley is founded on the backs of military funding. At some point that there was sort of a rupture of sorts in the relationship. And now what we are seeing and what we were starting to see around 2016, 17 is another pivot of Silicon Valley back to the Pentagon. And so that in a sense was sort of the third part of the trilogy of turning back to this era when Pentagon and Silicon Valley work hand in hand. And just as with DARPA, I wanted to know, is this good? Is it bad? What's good? What's bad? What to make out of it? I took that same approach going into this book. And you refer in your book to what you call the Pentagon-Silicon complex. How does that differ from the military-industrial complex in Eisenhower's terms? It doesn't differ in its fundamental makeup. It differs in sort of who's involved in it. So the military-industrial complex, this term that came out of Eisenhower's farewell speech, which is probably one of the most at times quoted, misused, abused speeches of all time. But he talked about the need, the development of a standing armaments industry, something that was new and unique in the United States and the need for it. And he did believe there was a need for it, but he was also worried about the dangers it posed. And it wasn't just industry. He was also worried about scientists. He referenced scientists of people making claims on the national budget of pushing us in directions that perhaps we shouldn't go. And so his message was that the public and our democracy needs to be engaged in those issues to pay attention to those issues and to attention to those issues. This new sort of military Silicon complex that is emerging is sort of different. Rather than plowshares, you have companies that had been involved or an area of the country. And of course, when I say Silicon Valley, I don't mean just the geographic area, but the ecosystem of venture capital and whatnot. But a part of the economy and technology, and to some extent, a part of the country that had been involved for the past number of decades, primarily in consumer and commercial technology, now wanting to become major military contractors. And some of the same concerns that apply to the military industrial complex of sort of revolving door, the concerns about people pushing for certain things, people and companies that were true of the military industrial complex for many decades, I think are now the same things that we need to pay attention to when it comes to Silicon Valley. So thus far, you've traced in this discussion, sort of a descriptive argument. I think it's important to note that there's a lot of history in the history, the relationship between Silicon Valley and the government, in as much as there is this general consensus view of the history, you can correct me if I'm wrong, wherein initially the government was a major investor in the 1950s and 60s and so on throughout the Cold War. At the end of the Cold War, end of history, in many ways, a lot of tech companies pulled away ideologically from the notion of supporting the armaments industry. There was this rift that you mentioned. And now we have a convergence again. People attributed this to many things. There's obviously a profit motive there. There is the Russian invasion of Ukraine. There is the rise of what some people call patriotic capital or patriotic investment. And many people would say, including, and I think I would be among them, that overall, this trend has been good. I'm glad that some American innovations heretofore had not been accessible to the Pentagon might now be for our national defense. But you seem to suggest that there are a number of pathologies as well that could be harmful themselves to national defense that have developed in the past few years. Would you elaborate on some of those for us? Sure. Let me actually, if you don't mind, I'll start with the rift because I think people, they have their own, you know, Silicon Valley has myths about itself. And so the rift being ideological. So I think part of the myth that we have of Silicon Valley, particularly the sort of the Steve Jobs, the start of the personal computing revolution is sort of counterculture, separated from the government, being almost anti-government. So the opening chapter of the book, the first chapter, starts with a story that I had been told originally by a former DARPA director after my DARPA book came out and had told me the story about Steve Jobs stopping by DARPA in the early 1990s, pitching Next. So to take people back, Steve Jobs had been exiled from Apple in the 1980s and had formed his own computer company, Next, that built these beautiful, sort of very distinctive black workstations, graphite workstations that were ahead of their time, had networking capability at a time people weren't really thinking of that. They were also incredibly expensive and were not finding a market. He had hoped that universities might buy them. They really weren't in quantities. And so this DARPA director said that Steve Jobs had pitched him and he decided to buy a few dozen of these workstations for DARPA, which of course was involved in computer networking at the time. But I also heard from one of the DARPA employees that the National Security Agency had a much bigger contract. And that this contract, this federal contract, NSA, and then also to some extent DARPA had saved Steve Jobs and then enabled him to sort of leapfrog back to Apple. So he's like, I saved Steve Jobs. And he was sort of half kidding. And I'm like, the National Security Agency saved Steve Jobs? You know, this person that we think of as a counterculture figure, we don't think of Steve Jobs as having military, let alone sort of intelligence connections. And tracking that story down, which I thought might be apocryphal, became sort of the starting point for me to think of as the starting point for the book. And then about, I think about four years ago, I met someone who worked on the contract. And it turned out that it was an even more intriguing story because he said, yes, that contract did save Steve Jobs and enabled him to go back to take over Apple. But it wasn't actually an NSA contract per se. NSA was a cover for the CIA and that it was actually a CIA contract. And it turned out that the CIA for satellite, as part of program B at the Madison's office, had bought tens of thousands of workstations and really kind of given Steve Jobs at Next a few years of runway to prove himself as an entrepreneur. And then, but when he went back to Apple, he never pursued any contracts with the military, with the CIA. And the reason for that had, I believe, very little to do with ideological differences. He didn't need them anymore. That critical time in the early 1990s, when, that intelligence contract sort of saved Next was, you know, at the end of the Cold War and the beginning of the peace dividend. And by the time Steve Jobs went back to Apple, the consumer and commercial marketplace was taking off just as the military budget was going down. There wasn't any benefit to him or Apple in working with the military. So part of my theory of the book is what we consider ideological differences between the Silicon Valley and the military. I don't want to say that you have genuine beliefs. You have people who are anti-military. You have people that are pro-military. And I don't want to say everything is based on your checkbook. But it is no doubt true that the general operating environment and the beliefs that you get out of that environment are tied to the economy. And so you had a generation of software and tech workers in Silicon Valley who came of sort of their workforce age and development in the 1990s when the military wasn't a major customer and commercial and consumer software and then hardware was. And you had workers who thought, grew up thinking, you know, you don't work with the military. And those become ideological beliefs. They become ideological beliefs that you can hold in part because there are so many jobs around that aren't tied to the military. So now, sorry, circling back to your question. Exactly. So now that we once again have an interest from the civilian sector in working with the government and providing technologies, be they military use only or be they civilian use, in what ways might this harm national security? And what mistakes are being made, for instance? Okay. So let me actually talk about the good and the harm. The good and the reason why you have across Republican and Democratic administrations wanting to work with Silicon Valley is because Silicon Valley has a lot to offer the military when it comes to AI, when it comes to software. Some of the issues that the Pentagon has struggled with, software in particular. It is an absolute truth that Silicon Valley is going to be far ahead in those areas than the traditional prime contractors. So working with Silicon Valley in those areas is undoubtedly something that you should do, that you arguably have to do. Now, what are the downsides? So I think the biggest downside is a certain culture. It doesn't apply to every company in Silicon Valley, but I think undoubtedly a Silicon Valley culture of, you know, fake it till you make it is very, very different than the culture of the FAR, you know, the Federal Acquisition Regulations for all of its downsides. The idea that, you know, you need to have, you need to be realistic about what you're doing. There needs to be, to accept the failures, report the failures that these things have to work. There is a tendency, and I think anyone who has dealt with the VC-funded defense tech startup culture to include even the big companies would admit that there's a lot of companies that are just outright bullshitting. That, I think, is the biggest danger. Some of the other related dangers is just sort of the general, particularly as we move from software into hardware, there's a lot of sort of sloganeering that I'm seeing of, you know, let's build and reindustrialize and lots of really great, rather than PowerPoint briefings, I call it pitch deck warfare of just really slick, marketing of like, we're going to do, counter this and counter that, and we're going to build missiles at scale. And it may be that some of these new companies can do it, that they can do it better, cheaper, they can innovate faster. But the vast majority, with the exception of the big, I mean, I'm sure there's other exceptions, but the biggest exception of SpaceX, which has proved itself in some critical areas, most of these companies have not proven that they can produce, for example, you know, missile interceptors at scale. They haven't proved they can do it cheaper. They can't prove, they haven't proved they can do it faster. And so to just assume they can is very dangerous. I hope they can, but they haven't proved it yet. So it seems like there are two categories of problem. One is the one you just outlined, which is that if the companies can't actually deliver, that's a vaporware problem. And there might be some waste, fraud, and abuse associated with that. But then I think there's a separate set of problems that if they can deliver and the government is highly dependent, a few companies who aren't as dependent on the government in as much as they have a huge amount of revenue that is derived from commercial uses, commercial users. And so it's almost like a sorcerer's apprentice problem. Government ends up needing AI for everything. And it's the private sector providing that, that has a clear asymmetry in its capabilities to provide that. The government cannot build these products on their own. Is the sorcerer's apprentice problem one of the things you talk about in your book? Yeah. So what I talk about, and I think is exactly what you're talking about, is this idea that the government can't deliver these products on their own. And so it's the model of most of these companies, not all of them, but most of the big ones. I mean, most of that they will, using public capital, venture capital investment, that they will, rather than going through this arduous, decades-long requirements process where the Pentagon lays out exactly what they need, they will, with their own funding, develop technologies and sort of sell these as commercial items. That has been, you know, the model for Palantir for many years, Andral, although they are also getting funding, but that's been part of their pitch, that they will take on the risks, the development risks. Well, what does this mean? This means that these companies own, in most cases, the underlying technology. So even if they, there's the commercial issue, but even if they aren't selling to the commercial world, they own the technology and can, in most cases, decide what to do with it. So what are the downsides of it? Well, one downside we saw was in the initial months after the Russian invasion of Ukraine, when there was this big controversy over Starlink. Specifically, there was, I think it was in Walter Isaacson's biography of Starlink, and it's a sort of corrected part of it, but this story that the Ukrainians wanted to launch an operation in Crimea, and Starlink was at the time, and to some extent today, still the communications backbone for the Ukrainian military. And they asked, they wanted Elon Musk to turn it on over Crimea. Musk at the time was, you know, tweeting these things about, you know, concerns about nuclear escalation and red lines and what Putin would do, and decided, you know, wouldn't allow Ukraine, he wouldn't do it, he wouldn't turn on Starlink, he wouldn't do it, he wouldn't do it. So, you know, wouldn't turn on Starlink. It was later, the story was walked back a bit of, well, he couldn't because, you know, Crimea was under sanctions. It became a more ambiguous story. But there were, and are today, still in the Pentagon concerns about dealing with Elon Musk and about the idea that you need permission. You know, people sort of outside defense think, oh, the government can just force a private company to do things. It's really, really hard. There's limits to the Defense Production Act. I mean, even going back to Truman in the Korean War, trying to take over steelworks to stop a strike, and the Supreme Court, you know, knocked that down. So, no, we can't just force private companies to do things we want them to do with their technologies. And you saw this again with the dispute between the Pentagon or the Department of War, as they call themselves now, and Anthropic after the Maduro raid. When Anthropic expressed concerns to their partner, Palantir, about how their models were used, Palantir then reported them to the Pentagon. Emile Michael, the chief technology officer, kind of went, "Well, I'm going to go back to the Pentagon. I'm going to go back to the Pentagon. I'm going to go back to the Pentagon." So, you know, I think that's going to be ballistic and the Pentagon designated Anthropic supply chain risk. And that's still being, you know, that's going to be battled out in the courts for a while. I think the Pentagon may lose, although unclear, TBD. But I, even if I disagree with how the Pentagon handled it, sort of going all in and just like, okay, we're going to cut off Anthropic and designate them a supply chain risk, I'm sympathetic to the dilemma that they face, which is, what do they do when companies have technologies that are critical to the military and they wish to place limits on them? I sympathize with the frustration and fear that the government and the Pentagon have with that. And I, my, what I argue in the book, you know, in this cases of SpaceX and Anthropic, I think we're going to see more and more of those conflicts as the Pentagon becomes dependent on technologies that they do not control. Well, may I push back a little bit and we'll see what you have to say, but I think in a competitive market, there will be some sort of a selection effect, wherein those companies that are willing to allow the government to use their technology for whatever purpose the government wants will realize they can make money on that. And those companies like Anthropic or anybody else who resists will be selected out of that market. That might be. And to some extent, that has always been Palantir's pitch. Indeed. Palantir's pitch is we'll do exactly what the government tells us to do. But there's a few dangers you, you get into there. First is does Palantir just become the new prime? I always tell people, you know, there's all this, you know, accusing Peter Thiel, one of the co-founders of Palantir, this, this, or that. I don't really engage in the fascism, not fascism debate. I don't find that useful for what I do, but I do tell people like really read his book zero to one, his book on entrepreneurship, where he just says outright what, what he wants, which is our goal is not competition, our goal is monopolies. We want to create, I mean, he calls them creative monopolies, but we want monopolistic profits. So the goal of Palantir is not to become one of, you know, many enterprise software systems that AI systems can deploy on. They want the Maven smart system and they're well on the way to become the software platform. So at one way or the other, I think in a lot of cases, first, you will become dependent on one of these companies and, And same thing, let's look at SpaceX. So you could say, well, if Musk does something we disagree with, we can just turn to Jeff Bezos. Like, he's not launching anything yet. So it's very, yes, in an ideal world, we would have competition. And then if one company doesn't perform the way we want, we can turn to the others. But the other thing I argue in my book is I, and that's my fear. Like, I do think we need Silicon Valley, but my fear of what is emerging are going to be new, and people are already calling them the neoprimes. SpaceX having, you know, and I know there's argument that's a de facto monopoly. Palantir, perhaps eventually having a de facto monopoly. Anderle clearly wants to be in its own way a prime, you know, whether monopolistic or not. They wish to displace the Lockheed Martin. So I think your pushback is valid, but I still think that the danger we face of very, very powerful companies that own the underlying technology, and even if you're not concerned about sort of the greater fears of democracy accountability, then you can be concerned about the situation we have now of, you know, these very small number of prime contractors being over budget, behind schedule, and all those same things. Yeah, to be blunt with you, I think if you and I were. We're having this conversation in five years. Instead of saying the term five primes, we would be saying the term seven primes, but that's the direction we're going. And I agree with you. There's no doubt in my mind. I don't even think that it's much of a secret that some of these new entrants intend to pull up the drawbridge behind them because that's merely the incentive of the market. The question is whether I think that those companies which tend to be more ideologically inclined toward the defense mission and select their personnel for that, whether they would refuse the government. The use of their products in some of those cases that to me is an open question, but I'm very sympathetic to your argument that that's the direction we're headed. They'll do whatever is in their financial best interest. So if it is in their financial best interest to supply the government Pentagon and not have any guardrails, they'll do that. But that also is sort of the dilemma of AI because government isn't going to be the only customer of it. We just spoke about ideology a few times, and then you brought up Peter Thiel. You also brought up Palantir and Anduril. So now I have to ask the question because full disclosure, even though I follow defense technology and claim to know a little bit about this, I've never read Lord of the Rings. Can you explain this Lord of the Rings universe companies to me, but specifically the extent to which you think that's just a marketing mechanism or whether the founders actually see themselves in that kind of light? So I when I first started writing about Palantir back in 2020, I went back and looked at went through a lot of issues of the Stanford Review. The conservative publication that Peter Thiel had founded when he was at Stanford, and he'd always been obsessed with Lord of the Rings along with sort of the these reading lists of sort of, you know, classic tomes of conservative books, they would include Lord of the Rings. So I am not the AI, I cannot do a deep textual analysis of Lord of the Rings, but the general idea that they appeals to them is this idea of a sort of an innocent, good civilization that you're trying to protect from outside forces. This narrative carries over for them personally into these companies, that they are in this battle with evil, that they are the protectors of Western civilization. So they are not just defense companies. They are, yeah, they are the defenders of the hobbits, I guess. I mean, but but so narrative, it has become particularly for Palantir, which is publicly traded at very high valuations, this really good narrative, and I have probably watched more. TV appearances of Alex Karp than any human being in their entire life should ever have to watch, and so I have watched where the narrative evolves, and one of the things that I find so fascinating about Karp is that he rarely on these appearances talks about the competitions they're involved in or the very specifics of Palantir software. talk about kill chain. We're part of the kill chain. We kill people. Sometimes we kill people. just kind of goes on and on. But he really harps on this narrative that has proved very popular, particularly with retail, meaning individual investors, that we are sort of superheroes battling the forces of evil. And that has become a really useful narrative for him. His narrative is shifting again now, which I find really interesting. In what way? So now he's going after the AI companies, this idea of sovereignty. This is a completely new the pitch that they've made. So Palantir was never an AI company. AI models deploy on the Palantir system. AI, they kind of were able to jump into the AI bubble. But then now the narrative is sort of very Karp is blasting the AI companies. It's just the evolution of narrative for Palantir I find absolutely fascinating. Well, I also find it very interesting the extent to which this ideology is very much a black and white, good versus evil, West versus the rest. Not that I'm not sympathetic, by the way, in a personal sense to some of those narratives, but there's something about it that's rather dystopian at times, rather dark. And rather it's this question of existential risk. It's like, if you don't, I hate to say it, but if you don't buy my products, or if you don't buy products like these, if you don't engage with the newest technology, the US is going to lose in some sort of massive existential race with our adversaries. I just find it be, I'm not sure if it's a good idea to say that, but I think it's a good idea to say that. I think it's a good idea to say that. What's funny to me is because I've covered national security for over 25 years. I don't think you cover national security if you don't have some belief that national security is important. I deeply believe in the importance of national security to our country, to our allies, to like-minded countries. But the idea that any one company is going to deliver that, I have hard questions. I have hard questions about their model for data, about what they're providing, about what other software providers are not being given the opportunity, about their contributions or lack of contributions to national security. So I just don't buy that they are the pinnacle or the absolute tool for defending Western civilization. They are a defense company, like any other defense company, and we should judge them and their products as such. Yeah, I agree. I definitely don't buy that any one company is a make it or break it, the hinge on which history rests. I think, though, what's interesting or what I would say is the claim is that if we fail to, let's say, uptake AI at extreme speed, that that is an existential threat. Our failure to win the AI race is existential. But simultaneously across the American population, you see it in the numbers that are coming out day by day. People believe that the excessive or too quick uptake of AI, that itself is the existential threat. It's a threat to the environment. It's a threat to the American way of life. It's a threat to children's learning and what it might be. So we have this huge question in AI playing out that split between those in national security who think it is absolutely essential and the domestic cases where there is growing consumer resentment. Do you think that that latter part, people's, the civilian public's concern with the proliferation of AI in their lives will ever bleed across into defense tech? What do you mean by bleed across that they'll they I mean, so let me take a bit of a broader issue. I think what you're seeing is a right. The public is smart. Like the public doesn't always have all the facts. They're not always, they don't, nobody, no people don't have the time to be informed on all of the issues that sort of affect us in modern society, but they know when people are BSing them. And so I think what you're seeing right now is rhetoric about, oh, you know, AI 10% chance it's going to kill us all, but we need to go really, really fast. They're seeing rhetoric about, you know, the concerns about human extinction, but we need another round of venture capital and public capital. And they're seeing the physical infrastructure being built of AI data centers. And there is a public backlash that I think is rightfully going across. It's one of the few things right now that's bipartisan of concerns, whether it's MAGA, Republican, Democrats about AI, about the companies and what's going on. And it's because I think the rhetoric that they're hearing doesn't make a lot of sense to them. Understandably, it doesn't make a lot of sense to me either. So that question of AI being so important to our national security, but it might kill us all. I think that's a hard one to resolve. I, for one, I'm certainly, I believe in investing in technology. I believe that investing in technology important for national security has had tremendous knock-on effects for commercial civilian economy as well. But again, I think there's a lot of rightful questions about that are being raised about the circular financing of some of the large AI companies about not just the technology, but the business models they're pursuing. So that's the bleed over, I mean, this general, this rising public resentment of the broligarchs, of the proliferation of data centers, of the foisting of social media on children, whether people, and now we're seeing it manifest in Congress with some skepticism towards the broligarchs, whether that will mean more regulation, whether that will mean more scrutiny over Pentagon contracts by whatever Congress comes next, whether that will mean antitrust legislation. I don't know which one of the things it will mean, but there's going to be a backlash. And I think it doesn't, of course, it matters which administration comes in. It matters what happens in the midterms, but I see this going across party lines. I think there's going to be a tremendous backlash. We're seeing the start of a backlash against Silicon Valley. And so, of course, what you hope happens is some sound oversight, regulation that makes sense. Will we get that? That part, I don't know. I'm a horrible prognosticator of the future. Well, I'm still going to push for some more prognosis. Let's put the bulletin hat back on to wrap up here. You're going to be covering this question of existential threat quite seriously in your new role, not just with nuclear, biotech, AI. I'm not going to ask you what happens in the year. What I'm going to ask you is, what are the big questions you want to get answered, if possible, while you're at the bulletin? I think that we are already exploring this question, sort of the AI doomerism, the risk. I do think that AI poses risks. I want to understand the science behind it, rather than just saying it poses a 10% risk of human extinction, because I do think those numbers are just drawn out of thin air. What are the ways that AI poses risks in the military operations that are going, such as in Iran, in one of the most essential areas for the bulletin, as AI becomes more integral and command and control for nuclear weapons? I'm not talking, per se, the worst scenario of AI launching nuclear weapons, but there's a very, very complex system for nuclear adoption of command and control, of choosing target sets. AI will become involved in some of those areas. What are the risks of that? The already risks you're seeing expressed sometimes, again, with that sort of doomer element of AI and biosecurity. In what ways can AI be used by rogue actors, whether it's in the military or in the military, whether it's in the military, in biological areas or the Houthis? What are the ways that AI's post-concrete risks? That, to me, is a really important area that we're exploring and should continue to explore over the next year. Well, I look forward very much to continuing to read the bulletin and your book. So, Sharon, thank you for joining Cogs of War. Thanks for having me. Thank you for listening to Cogs of War. Don't forget to sign up for our free newsletter that comes out every other week. Cogs of War. Keep reading, keep thinking, and stay healthy.