Ukraine destroyed Russian strategic bombers deep inside Russia with help from former DJ, a tattoo artist, and cheap drones hidden in trucks. What does that say about who holds power when drones are everywhere? Has China already caught up with the U.S. in technology? And what happens when a single private company can switch off the communications a military depends on?

Steven Feldstein is a senior fellow at the Carnegie Endowment for International Peace and a former State Department official. He joins Jon Bateman on the world unpacked to discuss his new book, Bytes and Bullets, and explain how drones are reshaping wars from Ukraine to Sudan, why export controls are slowing China but not Russia, and why technology may be the single most important force in the rise and fall of nations this century.

Transcript

Note: this is an AI-generated transcript and may contain errors

Jon Bateman: Many Americans, certainly people here in Washington, D.C., would have some version of the following story in their head. America has long been, and still largely is, dominant technologically. China is a rising technological power, but mostly it's not innovating on its own. And so America's job is to kind of hold on to its technology, to control it, and to prevent it from slipping it out of our fingers and kind of leaking into China. And so in essence, the competition between these two superpowers is a competition over control in which America is trying to hold on to control and China is trying to loosen that control. Is that what's happening?

Steve Feldstein: And I think that's a little bit of an old story. That would have been something that, you know, I think was emblematic of as we kind of entered the 21st century, China was clearly a laggard. They were clearly looking for ways to make up ground in almost any critical industry. The U.S. At that time had a pretty significant lead. Today, I don't think that is true anymore. I think in a number of critical industries, and this is one of the things I lay out in the book, you now, researchers find that China is either at par... Or even ahead of the U.S. So you look at a range of things, whether it's robotics, whether it is biotech, whether it looking at clean energy technologies. And China clearly has mastered a good portion of those. And at this point, its companies often have monopolies on the market. So it's not completely even, but this idea that China is a laggard, that they don't innovate, they copy. I think that's far outdated. It's in our interest to recognize that and understand just how much the landscape has changed.

Jon Bateman: Steve Feldstein, welcome back to The World Unpacked. Thank you, John. Happy to be here. You're actually my first ever repeat guest. I vowed to never do a repeat guest, I had my own boss on the show as a guest, I didn't invite him back. But you are here again. How does that feel, is that? It's a privilege.

Steve Feldstein: I enjoyed our last conversation and I'm looking to build off it, so thank you for having me.

Jon Bateman: Of course, the real reason why you're here on the show is that you've written a wonderful book. And when I got a digital copy of the book, I controlled F for my own name. I did find it. And so that's why you are here today. As you should. Steve, you have written an important new book on how technology, whether it's AI, drones, ships, rockets, is reshaping the global balance of power, including actual warfare and... Economic warfare, and I don't think I'm putting words in your mouth when I would say you seem to be arguing that technology may be the primary factor or the single most important factor in the kind of rise and fall of nations in the coming century, which is an important and provocative idea. So, we're going to talk about the Ukraine War, the Syrian War, semiconductors, so much more. The book is called Bytes and Bullets. Why not Bullets and Bytes, by the way? Did you consider that? I like...

Steve Feldstein: Bytes and bullets seems to roll off the tongue better. I don't know, that was what we settled on from the very early days.

Jon Bateman: Iambic pentameter, and also makes the orthography a little bit easier. You began this book with a remarkable story about the Ukraine War, a country you have visited during wartime. The story is about something called Operation Spiderweb, an incredibly daring covert action that Ukraine conducted behind enemy lines deep inside Russian territory. I'd love for you to tell the story because I had read about this, and yet reading the account in your book was so much more astonishing than what I had known, and I think tells us a lot about the themes that we're going to explore today.

Steve Feldstein: Yeah, no, I think it's a good starting point. And I mean, the operation itself, when it happened, was a real stunner. It was something that, I mean as the events were unfolding, I don't think we quite realized just what the Ukrainians had pulled off. But really what it was, it was an operation months in the making. And the idea was that Russia increasingly was putting Ukraine's defenses at a disadvantage due to its long range strike bombers. And so the goal for the Ukrainians was, how can we strike these bombers? And the problem was that they had no way to get to them with the existing weaponry at hand, whether it was missiles or drones. And some of these bombers were based thousands of miles inside Russian territory. Exactly, yeah, so they were dispersed amongst five different bases, as you mentioned, very, very deep in the Russian territory, so the idea that you could simply fly something in and strike them with any effectiveness in light of both the range as well as air defenses was pretty unlikely. So the Ukrainians came up with another idea, and this really was a prime example of subterfuge, but they found a couple former Ukrainian citizens who had relocated because of economic reasons to Russia. One was a former DJ, another is a former tattoo artist. And after-

Jon Bateman: of careful recruiting. And many tales of intelligence adventure begin with characters like this, I have to say. Absolutely.

Steve Feldstein: Those details, it only came out months later from others reporting that I was able to kind of piece together. And so after months of kind of recruitment and discussions with the intelligence services in Ukraine, they decided that they were trustworthy enough. And then the idea was, well, how can we put together a plan to destroy or really kind of create havoc amongst the long-range bomber fleet? So the idea was essentially... We need to find a way to sneak in quadcopters with explosive devices. We then need to have those assembled and then we need to transport them somehow towards these bases. And as it turned out, it wasn't as hard as you would think. I mean, essentially what they did was the Ukrainian intelligence services were able to sneak over the border, disassembled pieces. They kind of held them in these long trucks. They put them inside, it was hard to tell exactly what those were, and they were able to get them across the border without too much of a problem, and then get them to the DJ and the tattoo artist. So once they got the equipment, a couple things came into play. One, they had to figure out how to assemble the drones. And that wasn't easy, especially for people who didn't have the technological skills to do so. So I think they had to use different forms of phone calls and other kind of messaging. In order to give them very careful step-by-step instructions in terms of making sure they could assemble them correctly. So they were able to do that. A next step was that they actually had to recruit truck drivers without telling them exactly where their end destination would be and what the cargo was that they were carrying to transport these drones to those destinations. So over a long period of time, they were to do at different points when they actually recruited the truck drivers and the drivers were moving. You know, the drones on the road. I think at one point, one of the trucks, like the opening popped open accidentally, and the driver said, wait a second, these are drones. I didn't agree to carry this. What is this for? And they had to quickly make up a story on the spot that was good enough for the driver to say, okay, I guess I'll continue.

Jon Bateman: Is this like the Pink Panther or Mission Impossible?

Steve Feldstein: Yeah, no exactly I mean it was like one thing if this had gone the other way just one single human being The entire operation could have unraveled. Yeah, so at any rate they were able to get the trucks and they were instructed They instructed truck drivers park each of your trucks You know adjacent to these bases and wait and then at that point they were to remotely open the trucks You know the kind of like the the panels on the top Fly the drones and then use artificial intelligence and computer vision to identify the specific targets that were pre-programmed in advance and to then use those drones to detonate on impact and destroy those long-range bombers. Now, I think my understanding is four out of the five bases that worked well and one, it malfunctioned. They weren't able to destroy the planes. Nonetheless, the damage was significant. In the billions of dollars, they destroyed long-ranged bombers that have made... Decades ago that no longer were even in production, and yet were essential to Russia's nuclear strike capability. And so not only was this a visible and very public embarrassment, but it also set back Russia's long-range strike capabilities to a significant degree.

Jon Bateman: One of the reasons why this is such a great story to illustrate the ideas from your book is it's a David and Goliath tale, where David's slingshot is the new technology of drones that enables them to flip the script. So you've got, on the one hand, these Russian strategic bombers terrorizing Ukraine. Basically, Russia is living off the military-industrial legacy of the Cold War. And then you've got these scrappy Ukrainians that through ingenuity and a few dozen quadcopters that did not exist a few years ago, that only cost a few thousand dollars at most, maybe 18, 24 inches long, right, very light, very small, they were able to destroy these $100 million planes. And actually, as you say, affect Russia's nuclear deterrent. I mean, what better illustration is there of how technology is rewriting the rules of who's a winner and a loser today? What do you make of that?

Steve Feldstein: Yeah, no, it is, you know, it's interesting. One thing I would say, it certainly, what we've learned, it not a one-off, because the US military itself is learning some of these very same lessons right now in regards to the Iran War, where you're seeing a similar dynamic play where essentially these very cheap drones, used creatively, are destroying 100 million plus dollar radar installations, bombers, AWACS, surveillance planes, and so forth. And so to me, this is sort of emblematic of a bigger shift that's taken place, a post-Cold War shift, where increasingly the know-how, the expertise, the ability to do creative things in very destructive ways with technology is spreading across the world into all sorts of hands. Now, Ukraine is one example. They still have a formally constituted military. They have a pretty significant defense industrial base, maybe not akin to Russia's, but still something that. Is a reasonable legacy of the Cold War. And to some extent, there is a predictable aspect to it. They are a relatively advanced country. They have great engineers. And so one would think that if they were able to withstand the initial assault from Russia over time, they would be able to come up with creative ideas. But what we're finding out, it's not just Ukraine. It is countries and insurgent groups that are not formally constituted, that many would consider to be sort of underdeveloped places like. The Houthis in Yemen, Hezbollah in Lebanon, even insurgent groups in Mali that are doing all sorts of creative, innovative things to disrupt far more powerful, better-resourced militaries due to this adaptive technology.

Jon Bateman: I think people know that drones are now an important part of global militaries. I was still surprised to learn, if I read this right in your book, that over a five-year period you have calculated that perhaps 140,000 fatalities worldwide from drone strikes in four or five different conflicts. Is it at that scale?

Steve Feldstein: Yeah, it is. Yeah, that's right. So that's data I actually took from ACLED, which is a well-resourced database. And, you know, what I did is I aggregated over this five-year period both the number of drone strikes and the number civilians who were killed that were attributed to those drone strikes. And each year, each successive year, last year, I think it was something like 36, 37,000 killed. The year before was even higher, something, you now, I thing in the high 40,000. Right, I mean this is very much a mainstay in war. And we can look at a few of the biggest conflicts that we're facing, that we are looking at right now, Iran, Ukraine, Gaza, as prime examples of how mostly they're being fought via drones. More than artillery, more than any other kind of weapon, it is drones that become the mainstays of those wars.

Jon Bateman: And I guess a lot of these wars where drones are being most heavily used, I'm talking about Sudan, some of these other places, they're a little bit outside the attention of some people in the West. So where you could get this curious situation where the number of fatalities globally is at the scale of a major war, and yet it's almost hidden.

Steve Feldstein: That's right, you know, Sudan is a great case in point, Myanmar is another case in point, but you have these conflicts that, I mean, Sudan, is, you know, is essentially like a hidden conflict when it comes to the attention span of the American public and, you know, the conflicts in Sudan which date back decades and frankly, which I worked on when I was in government in different forms, either in South Sudan or Sudan, persist today. I mean they're in the midst of a brutal and horrifying civil war, perhaps the most significant humanitarian crisis in the world. And you essentially have a proxy conflict that is being fueled on the one side by Turkish and Egyptian support via TB2 drones for the Sudanese armed forces, which is essentially the kind of government military. And on the other hand, you have, according to allegations, are reporting the UAE importing Chinese Wing Loong 2 drones to support the insurgent militant group, the RSF. And, you know, essentially each side is using these drones in different manner. To conduct long range strikes and frankly to terrorize and intimidate the population of the other side. More than anything else these drones are being used first and foremost to indiscriminately target civilians and the results have been horrifying. But because both the sponsors of both sides feel little need to kind of act in accordance with international humanitarian law or international human rights and because this is a conflict that has just. Escape the scrutiny of the public, it persists and continues in a really damaging manner.

Jon Bateman: I want to come back to Sudan and some of these other active conflict zones. They're so poorly understood. And also, it's such a strange evolution that we're saying, I mean, you use this term, long-range precision strike. This is a term that's only been associated with the United States military in recent years, perhaps Russia. The fact that we are now talking about militant groups in a country like Sudan and using that same phrase, because of the rise of drones, is remarkable. But I want to turn our attention to a topic that is maybe closer to the attentional zone of a lot of people watching, which is U.S.-China competition. I actually flipped through your index. I was looking for what are the things you talk about most in the book. Top item, China, America's competition with. So let's talk about that. Let's do it. Um. Many Americans, certainly people here in Washington, D.C., would have some version of the following story in their head. America has long been, and still largely is, dominant technologically. China is a rising technological power, but mostly it's not innovating on its own. It's kind of stealing and using unfair tools like subsidies to catch up. And so America's job is to kind of hold on to its technology, to control it, and to prevent it from slipping it out of our fingers and kind of leaking into China. And so in essence, the competition between these two superpowers is a competition over control in which America is trying to hold on the control and China is trying to loosen that control. Is that what's happening or do we need a more complicated story? Yeah.

Steve Feldstein: I mean, I think that's a little bit of an old story, honestly. I think it would have been true 10 years ago. That would have something that, you know, I think was emblematic of as we kind of entered the 21st century, China was clearly a lagger. They were clearly looking for ways to make up ground in almost any critical industry. And, you now, the U.S. At that time had a pretty significant lead. Today, I don't think that is true anymore. I think in a number of critical industries, and this is one of the things I lay out in the book. Uh... You know researchers find that china is either at par or even ahead of the u s so you look at a range of things whether it's robotics whether it is biotech whether it' s looking at clean energy uh... Technologies uh... In china clearly has mastered a good portion of those uh... And and you know at this point its companies uh... Often have monopolies uh... On the market and so And then there still are a few other areas where there is a lot of contestation, and there certainly are some areas where the U.S. Remains ahead. A couple of them I would point to would be one, reusable rockets, in terms of putting up space rockets. That's one area where China has yet to kind of master that technology. Another is leading edge semiconductor chips, which China sort of continues to struggle. I think it's making up ground, but continues to struggles at some point, whether it's three, five, or seven years behind, what the U.S. Is able to do. So, it's not completely even, but this idea that China is a laggard, that they don't innovate, they copy, I think that's far outdated. I think China is leading the innovation side on many critical technologies, and it's in our interest to recognize that and understand just how much the landscape has changed.

Jon Bateman: You mentioned it's a 10-year-old idea. I can actually remember precisely 10 years ago, I was working in the Pentagon, and this was the time when America's concern about Huawei and 5G telecommunication equipment really spiked. And what crystallized to me about that episode was American policymakers were not used to the idea that China actually was the market and technology leader. In. A strategic tech space of growing importance worldwide. And it almost happened without a lot of people in Washington paying attention. And so all of a sudden, they wake up and realize, wow, they're not just stealing from us, they are not just copying from us. There ahead, we, the collective West, are behind. And then that was also around the time when people started noticing China's rapid progress in AI. This was prior to large language models and the kind of generalizable AI tools that are most prominent today. People were looking more at narrow AI algorithms and simply counting the amount of scientific research, often of high quality, coming from Chinese universities and labs. Right, well, there's also computer vision.

Steve Feldstein: Too, right? Because they had the surveillance systems. That was one of the things I wrote about in my first book, you know, both this kind of idea that you had these Chinese behemoths like Huawei, ZTE, that were kind of very quickly emerging and selling all sorts of exporting, all sorts to surveillance equipment to other countries in doing so with pretty advanced technologies that our own companies really hadn't done much in terms of marketing. But actually on the 5G question, that was also pretty interesting because the counter that we had to Huawei. And some of the other Chinese companies' 5G offerings were essentially European companies, right? It was Nokia or Ericsson, but U.S. Companies weren't even making this stuff anymore. So when we were going around diplomatically, and I was in the State Department when you were at Defense, so we were kind of in other jobs, in other lives, making the push about, well, use the U. S. Alternative. I mean, it wasn't even a U.s. Alternative, it was a European alternative under this kind of larger allied umbrella versus the Chinese options that were on offer.

Jon Bateman: So now we're 10 years past that. And as you say, China is now no longer the leader merely in 5G telecommunications. And there's long and twisted saga that we could talk about with Huawei. But now I think it's fair to say commercial drone technology, electric vehicles, advanced batteries, solar panels. You could kind of go down the list and these are significant. Sectors. These are strategic sectors. So, okay, so we have rebutted the old story that America is trying to kind of hold on to this leaky sieve of its technology leadership. What then is the new story? How should we understand the current state of the competition and the right... Framework or paradigm through which Americans and our leaders can formulate some kind of response.

Steve Feldstein: Well, I mean, one of the things, the theory of the case that I hear is that essentially there are some technologies that appear to matter more than others. So that even if China is at par or ahead of the US in some areas, on certain things that are fundamental, particularly AI technologies, that's an area where the US maintains a lead. And that it maintains a league both on the hardware side, you know, in terms of leading edge chips, you know largely designed by Nvidia and made by TSMC. And it retains a lead as a result of that on the software side when it comes to putting out frontier models where Chinese alternatives remain a little bit behind. And so I guess one question you can ask is that is AI going to be kind of more important than other technologies? And if so, what is the theory in terms of how the US can protect a very fragile, somewhat diminishing lead in an area? Is that possible? And is that something that... Is in the U.S., America's best interest, and I think you can have an interesting argument on that in sort of two directions, which I can get into.

Jon Bateman: Yeah, no, I mean, you're almost capturing my own personal vibe shift, because as a red-blooded American who had a career in the US National Security Establishment, eventually I became somewhat depressed about what seemed to be kind of an inevitable Chinese catch-up. And I sort of thought, OK, I guess this is just how technology works. It diffuses, and it advantages kind of rising powers and second-comers. And This may just be part of a long-lasting erosion and an aspect of American power. And then AI came around, specifically the LLM boom inaugurated by Chad GPT. And it kind of felt like America had gotten its mojo back. And I started to think, well, maybe there actually is something to be said for the dynamic, unruly, wooly, American-style venture capitalism. That has now produced what an argument I take seriously is a kind of master technology. Where maybe if we stay ahead in frontier AI, maybe that gives us the kind of cognitive capacity that allows America to reverse the tide. We could become the EV leaders, we could become advanced battery leaders, we could be the drone leaders because access to this frontier AI will become the most important. Engineering and design tool, the most important manufacturing tool, it will be the kind of universal salve that you can apply to any problem. Yeah.

Steve Feldstein: Well, that's, I mean, look, there's a lot of assumptions underlying that. You know, I'm not ready to completely dismiss the fact that AI is fundamentally more important than many other technologies, but I would poke a few holes into this. Let me offer a couple. So one of which is that, you know, with AI, there are sort of two ways to think about it. You have the frontier and you have deployment. And what we know is that yes, the US is leading on the frontier. Right, the US is the one that's ahead when it comes to the best models, the most cutting edge, the one with the most autonomous agents running around all over the place, breaking into different companies, right? Fine. But when it come to the actual take up and adoption of models for applications around the world, when it came to which models are being most used by other countries and their developers, you know, I think what we see a very different story in that Chinese models have actually proven to be fairly popular and successful on that halving. A model that is almost good enough, almost a frontier. And frankly, like, I mean, the question you gotta ask yourself is you're a small business in Kenya or Nigeria. Do you need the frontier? Or do you need something that is just about at the frontier that comes at a 10th of the cost?

Jon Bateman: Here's my respectful push back to this. This is almost the great debate in the kind of political economy of AI right now. That's right. Do you need to be at the frontier, or is it OK to be good enough? I think one thing is, from what I can tell, the frontier actually is still more popular in a literal sense. A lot of people look at data from these router companies to analyze which models are being used most. But... What I've come to learn recently is that it only captures a small slice of AI usage and that there's reason to believe that OpenAI and Anthropic are maybe still serving up about 90% of the tokens online. So that is a little bit of a market test that people are willing to pay for the best. And then maybe then bringing that idea to different countries. It certainly seems sensible that in some industries, some companies in some countries, which have fewer resources. Maybe if you go to a developing country, yeah, they don't have the best of most technologies, right? I mean, that's what it means to have fewer resources. But if you think about what does it mean for Volvo to be a competitive global enterprise? What does it mean for an advanced economy to function and compete? One might think that being an advanced economy or a high-tech company means you need the best. Cognitive tools. That's sort of the essence of who you are. Of who you are.

Steve Feldstein: Well, I would say, and I think that's a smart response. I would a couple things there, which is one of which is that a lot of places are using both Chinese and are interspersing. So I was just in Singapore about a month ago, I was talking to different officials and entities in that country. I mean, essentially what they're saying is that certain things we use cloud code and other things that where we need and kind of less tokens and more powerful use DeepSeek or Kimi or one of the other Chinese models and that we are not willing to choose one or the other. And so I think you're seeing a lot of that. I think the second thing I would say is that if you're making the advanced economy argument, well, one of the biggest advanced economies that I see coming around the horizon is China. And we know the kind of push that is being made internally there in terms of what types of models to be used. Even if today it's okay for Chinese companies to use foreign models, I would bet you in five years that will probably no longer be the case. In which case then, you can also ask the question, does fall will matter. In the face of BYD and all the other competitors that are coming from China anyway.

Jon Bateman: Yeah, that's a great question for Europe, actually. I think you're right. To zoom out a little bit from AI, but I think just to draw the deeper lesson here, American policymakers seeking to hold on control, I think, have often wanted countries to sort of choose between the US and China, like when America allowed advanced AI chips to go into the UAE. That was on the condition that the UAE rip out a lot of Chinese equipment and software in order to create that separation and to sort of use the leverage that Washington believes that it has. AI kind of challenges that, and that's been my experience, too. When I vibe code my own apps, and I've got several, immediately you realize that you want to use the fastest, cheapest on-device model that you can for whatever tool. You're trying to achieve, and that inevitably means it's a mixture of models. That's right. That seems to be a challenge to this American strategy of kind of forcing countries to choose and a strategy of control, generally.

Steve Feldstein: Yeah, I think that's right. I think you're sort of, you know, kind of rowing up upstream a bit on this idea that people will sort of conform to very static, baked-in choices, that they don't have their own agency, frankly, to make decisions about what they want. And that country's...you know, it takes...it's supply and demand. I mean, you can make whatever diktats you want as a supplier and say, you're gonna use our model, You're going to bake in our ecosystem and you have to make a choice. But, you know, there's a demand side as well, and countries can choose, and they can make their own choices about what works best for them. And there will be price considerations, there will geopolitical considerations, and innovation considerations. And those three are all ingredients in a very complicated mix that varies country by country. But I think this idea that you can divide the world in two, and it's preordained that it's going to be one or the other, it's not exactly how things are playing out, at least not so far.

Jon Bateman: Yeah, export control specifically. That has been the lead tool, the main action arm of a lot of this US policy. With varying results, I think you would say, and I think this is something I gathered from your book. I want to ask you about Russia in a minute. With China, there's been a lot of debate about whether the US export controls on chips and chip-making equipment going to China have been effective. My sense is most people now think that they have been effective in some sense. So if you go with the mainstream view on that, check the box. That's a win for the American control strategy in what some people think is the master technology, the master competition. But then you look at a country like Russia, where we have intensely thrown export controls at them because of the Ukraine war. What kind of results have we got there?

Steve Feldstein: Yeah Yeah, no, it's a great question. You know, the one thing I would say very quickly in China before we shift to Russia is that I think I agree that export controls, by and large, as it relates to the leading edge chips have worked for now. And if you look at it in sort of the very short time horizon in terms of saying, what have we bought ourselves in terms of time today, I think that's a fair argument in favor of export controls. But I would also say, look a little longer. Look at a 10 or 15 year horizon. To what extent. Will these conditions persist? Will this lead persist? Are we not also incentivizing Chinese companies and engineers to come up with new ways to design chips, new adaptations that potentially will rival what we have currently? I mean, I've seen just reports even this morning, I was looking where Chinese companies are coming up with new software offerings that rival CUDA from NVIDIA when it comes to the software needed to run advanced chips. Now again, maybe that's not quite where NVIDia's software is today, but Ask that question again five, seven years, and potentially you'll get a different answer.

Jon Bateman: Well, and just to say, this is the challenge of analyzing and forming policy on technology. You have to plan for a five, 10, 15 year horizon. You're our first double guest on this show. We've got a huge backlog. We actually do book about 10 to 15 years in advance. So I think, Steve, we should go ahead and set up the part three where we can actually test whether these export controls worked or not. But bring us to Russia, because generally speaking... My recollection of the mainstream narrative is at the outset of this war, the West, led by the United States, hammered Russia with unprecedented levels of sectoral sanctions and export controls. And there was a kind of exuberance. People were sharing these infographics about the collapse of the ruble and this, that, and the other. Four years on, I think maybe there's more of a frustration of, like, how are they still in this? Yeah.

Steve Feldstein: Yeah, so I would say, at this point, the sanctions regime, to the extent, you know, one is still being enforced in a robust way, has sprung a myriad of leaks. It is on life support in terms of stopping Russia's military machine from acquiring the components it needs to punish Ukraine. And it's a very tragic thing, but there's a lot of reasons behind that. I'll give you a few. Let's start with the fact that, like, even if you want to run an effective sanctions regime, you have to constantly update it. You need to have a robust enforcement mechanism and a institutional body, BIS, that will constantly find new components that are being substituted. And this is a part of the Commerce Department that enforces export control. They're the ones in the lead in the US government, and they've lost personnel, as most other parts of the US Government lost personnel under the Trump administration. So you need to constantly update it. So one thing we've seen is that the US has not updated their sanctions. So even if. You had the same regime in place, and barring other outside actors, the fact is if you're not keeping them up to speed, then over time, they lose their effectiveness. That's one. The bigger factor is that we were never able to get China on board with sanctions. At best, we pressed them not to provide direct arms shipments, even though they provided material support in terms of all sorts of components. All sorts of secondary assets, like bulldozers, all sorts of things that the Russians see and see machines that the Russians have used. And we were never not able to stop that. At this point, it's like an open border. Not only do we know that the Chinese and other countries kind of along those those lines, including the North Koreans are providing different levels of material support in the case that the North Koreans even artillery and and and missiles. But we also know that it's reciprocal and the Russians are giving in-kind IP and technologies. To those countries in advance. So it's a mutually beneficial two-way street. And it points to the fact that even if the U.S. With its allies were able to construct like an iron wall and stop anything from flowing outside the country, and they certainly have not been able to do that, the China factor would have stopped that. But the third thing I'd quickly point to is globalization, is that even when we have tried to stop companies from providing components, There is such an integration of supply chains across the board that run through all sorts of trans-shipment points in secondary countries that stopping that flow from leaking has proved to be almost impossible, whether it's going through Morocco, whether it is going through China, as we mentioned, or Mongolia, or some other place. The idea that you sell to a country, which is a very normal part of commerce, and that the Russians then come into that third country and they kind of repackage and then import. Those, the components they need to themselves, is like, is part of their, it's been part of their playbook and we have not been able to really put a damper on that at all.

Jon Bateman: Yeah, you know, often some sort of Russian military hardware will be captured by Ukraine and exploited and publicized, and people will tear it open and say, like, what are these components? And they'll find Western and US components in there. And then someone will go to that company and say what the fuck? You know, like, how are your components in their? And then the response, very understandable response, will be like, well, we sell these in every country in the world. You can buy them essentially at like Best Buy, or I'm slightly exaggerating here. And so there is no regime that can block those from coming into Russia. It's just too diffuse. I guess that's part of this control problem, right? That if more powerful technologies are just more widely available. Right.

Steve Feldstein: Well, I mean, one of the problems has been in the messaging. So the messaging from the Biden administration, coming from President Biden on down, was essentially, we are going to destroy the Russian machinery, it will be unable to access any kind of components whatsoever, we're gonna send them back to the Stone Age, that kind of rhetoric. That was never gonna be the actual result. I think the actual goal was always gonna be a bit more modest, which is that you will present a very high barrier, You will make it costlier, more expensive, and more difficult to acquire the components that are needed for Russia to prosecute its war effectively. And I think if perhaps we had messaged it differently and we had publicly said, our goal, we understand, we can't reverse globalization. Yeah. Like we just, we can, like we're in a different age right now, but that we could at least make it, like we could create huge pain points and really put a resistance in the machine. Of exporting and importing, I think that would have probably led to more effectiveness. But I think we messaged it wrong.

Jon Bateman: And here we are. Yeah. Well, honestly, Steve, I think that's a good broad lesson for foreign policy and, frankly, just for life on this earth. The tough problems can rarely be eliminated. The stubborn adversaries can rarely removed from the chessboard. What you can do is make life harder for them and create friction and costs. That's a very frustrating answer to Americans who are accustomed to being the power player, the hegemonic actor. We want to see 1989.

Steve Feldstein: We want to see a full collapse of the Soviet Union, the fall of the Berlin Wall, and the ascendance of a unipolar world, once again, with the U.S. At the top. And that's history. There's all sorts of reasons why that kind of dynamic is no longer in play right now. As I mentioned, China is one example. Globalization is another example. So we can't go back to that, and we shouldn't expect that we will.

Jon Bateman: Yeah, and if 1989 contained the seeds of 2022, when Russia had its full invasion of Ukraine, was 1989 all it was cracked up to be? Was 1989 really 1989, right? Right. Because it just set us on this path. Russia was not wiped off the chessboard, even then. No. They went into a kind of weakened state, but the geopolitical grievances, the military industrial base, the unresolved tensions with countries like Ukraine remain present.

Steve Feldstein: That's right. That's right. And you know, look, there were all sorts of questions about, you know. I mean, there was a time, and I talk about this in the book a bit, where under Gorbachev, a very messy period in Russia's history, I mean, it was actually a real desire to be part of a formal alliance with the United States. And that, you know, perhaps if the transition to a market economy hadn't been so messy and so corrupt, that, you know paving the way for someone like Putin to take over. You know, perhaps if the U.S. Had been in a more formal alliance and was able to exert more direct influence, maybe things would have ended up differently. It's impossible to know what an alternate history would have brought us, but either way, we're here now in part because of the legacy of how the Cold War ended. Yeah.

Jon Bateman: Yeah, it's hard to know and, you know, I want to keep us on this tech topic, but it's actually very relevant. A lot of political scientists would say that a hegemon like the United States would inevitably trigger other actors to come in and balance against us, right? And the biggest fear of U.S. National security leaders is that the bad guys, as we see them, Russia, China, Iran, North Korea, perhaps others, would somehow unite in this balancing act. And work as a concert against American interests. Now, typically, this has only happened in bits and pieces, but tech seems to be one of those areas where, as you say, Chinese supplies are filtering into Russia to support its war machine. North Korea is sending troops and technology to Russia. Even Iran, it's remarkable that Iranian drones are powering. Russia's attacks in some substance. So in this tech world, we now are seeing, you might call it an alternate ecosystem arise. And if China is gonna be a part of that ecosystem, that means this alternate ecosystem is always going to have a lot of technology to offer.

Steve Feldstein: Yeah, well, look, China will certainly have the ability to call the shots, as it will like. It is now the world's foremost manufacturing power, exceeding the U.S. It has an independent foreign policy that oftentimes is in conflict with, or certainly in competition with, the United States and its allies. And so it no longer is a country that you can write off because you can't write off China, all the interests it has which don't conform to U.S. Or allied interests. You know, run the risk of undermining US influence and US power. So that's a very real type of thing. I mean, what I would say, you know I hesitate to go too far in the direction of like you know an axis of powers that are aligned against the US and whatnot, because I think it's one of you know very specific transactional interests. I think there's very little so far behind the actual kind of, I mean there isn't a normative affinity. That Iran, Russia, or China really have for one another, I think it's one where they see a particular area where mutual cooperation can be beneficial, right? Iranians said, look, we'll give you the drone platform, the Shahed drone, one-way attack drone platform. We can build a joint factory, and then you can do other things to adapt that drone, and then, you give us something back in response. But I don't think the same kind of shared alliance that we've seen in conjunction with NATO,

Jon Bateman: Certainly not, certainly not, but in terms of tech transfer, it's a concern, I would say. I mean, let's put it this way, there's not a lot of a normative affinity between United States and the UAE, for example, right? Radically different society, radically different values and social structure. I guess you could say the Trump administration and its family have certain tastes, let' call them, certain peculiar affinities toward...

Steve Feldstein: An aesthetic, a gold esthetic of shared interest.

Jon Bateman: Strong men and cash and flash. But beyond that, it's not there. And yet, the UAE is becoming perhaps America's most important security partner in the Middle East. So these things can result in tech transfer, even if it doesn't reach that level of kind of cultural trust that we have with NATO. That's right.

Steve Feldstein: That's right, that's right. Although, I mean, look, I wouldn't go so far as to call the UAE an ally. I think they're hedging quite a bit still, even if G42, their AI champion company, has made a very specific decision to align itself with Microsoft and U.S. Interests on the AI front. I think the country more broadly is still relying on lots of Chinese technology when it comes to its networks and other sorts of things. So I would be—I'm skeptical that we're seeing kind of a full break one way.

Jon Bateman: Yeah, not calling it a break, but I will say, and yet, despite all the connections with China, the Trump administration did something very few people noticed recently. I'm sure you track this, Steve. There is a list for export controls of countries based on kind of how good buddies you are with us and how many security threats we see connected to you, and the UAE was recently to the top of that list, the so-called A-tier. Of countries alongside countries like the United Kingdom and Canada. So it is almost a free-for-all now in terms of technology transfer to the UAE. Now there's an element of that where it's like only a Trump administration would do that. But actually, I'm not sure I would go all the way. I think America is looking for new friends in the Middle East as ties with Israel's hours. The UAE has a lot to offer. So, the kind of hard-headed people in the national security state, I can imagine them advocating for this, retaining this into a democratic administration.

Steve Feldstein: Yeah, well, look, that's fair. And frankly, it was something that was started in the Biden administration. I mean, the whole idea between cultivating G-42, getting Microsoft to come in as an investing partner, forcing the Chinese, the UAE, the Emirates to make a choice between the U.S. And China, that was all started in The Biden administration and was continued on and got bigger. And, you know, there was the Gulf trip in May of Trump's first year in office in a second term that solidified things and also led to a bit of personal enrichment. But the ground was laid under Biden. So I agree with that. And I don't think at all that in a potential successor democratic administration that you would see a shift in policy in that way. I think the UAE is important. The Middle East currently is a mess. I think I think last thing either side would do would be to back away from a relationship that seems very critical. So yeah, we're in agreement here.

Jon Bateman: We spent the last little while talking about the perceptions and decisions of governmental officials in Washington, Beijing, and other capitals. Going back to the index of your book, some other entities that are highly cited in the index, Elon Musk, Jensen Huang, Mark Zuckerberg. These are not government officials, other than a brief tour of duty that Elon provided some service to his country. These are the titans of industry whom many people perceive to actually be calling the shots on US tech policy. This is a major subject of your book. I think you raise a question in this book about who really is controlling U.S. Technology policy and strategy. Tell us about that.

Steve Feldstein: Well, I mean, I don't know if you saw the news this morning or from yesterday, but apparently even on AI regulation and governance, it's all voluntary. It's all company's design. So, you know, the shift I talk about in the book is that we have moved from a closed technological ecosystem to an open technological ecosystem subsequent to the Cold War. So during the Cold war, the idea was government funding by and large was represent and the bulk of R&D. That proprietary innovations were created either through government-supported labs or government labs themselves. Oftentimes they were hoarded and kept within a closed ecosystem, but they were not commercialized. They did not make their way out. And that after the Cold War, you saw a flip when it came to levels of funding of R&D. So the private sector portion increased significantly. The government portion was slashed in part to take advantage of a peace dividend. And so... And a lot of these technologies, these innovations, were commercialized. And so you saw iPhones and other sorts of consumer electronic devices that used innovations manufactured in government labs like Siri, which is natural language processing, or the touch screens. These were all things that were commercialize and kind of put out. And so what that has meant over time is that now the bulk of innovation, the bulk capital and the design decisions have largely been outsourced. To private companies and their labs and their scientists, as opposed to being run from and influenced by governments. And so that has made Silicon Valley and companies like Google, SpaceX, AWS, and Thropic very wealthy and very powerful, but it also means that they are running in a direction that sometimes is in conformance with US interests and sometimes. Runs against those very interests. And with this administration and the deregulatory approach that they have taken, not just with AI, but that being a very good case in point, we're seeing that chasm widen even further where they are increasingly given the independence to make decisions that directly affect U.S. National security interests without the input of public officials.

Jon Bateman: So SpaceX is a really interesting case study here. This is Elon Musk's rocket company. It has built these remarkable reusable rocket engines. It runs the Starlink internet network from satellites. It has tens of thousands of satellites in low earth orbit. Its recent IPO was what briefly made Elon Musk a trillionaire. So it's worth more than a trillion dollars itself and is a major contributor to Elon Musk wealth. Um. In the original Apollo program, NASA heavily relied on private contractors to build spacecraft and software guidance systems and the like. But as you implied there, Steve, NASA engineers were actually still driving technical decisions. They were deeply in the weeds and essentially telling contractors what to build and how to build it and giving the specs. That's really flipped now, where... With different commercial relationships that SpaceX has been a part of, still largely funded by the U.S. Government, but the nature of the relationship has changed where the government now simply says, here's what we want you to achieve, you go and figure it out. And partly that's because we no longer have that intellectual capital in the U S government to make those decisions. But I will say in defense of SpaceX, they've done what other people couldn't. I mean, could you? Could you turn that around and say, someone like Elon Musk is only so powerful because he is providing services and products that others can't?

Steve Feldstein: Can't and or won't. Yeah, I was going to actually make that same point and I agree with you in that. And that there's the conundrum with Elon Musk is that on the one hand, he has broken through with his products and done things for the space program that otherwise were stymied by Boeing and the contractors that kind of operated sort of business as usual. That he was able to make some bets. He had to file lawsuits in order to allow his products to compete. For space contracts, and he did things that we didn't think could be achieved with the amount of money that was spent. So he deserves a lot of kudos for being able to do that. The flip side of this, however, is that at what point do you have to rein it in? At what point, do you say a private individual with their own interests who are making national security decisions about how their products are used is too much and goes against. The interests, the broader national security interests of the United States. And there's lots of examples. One example is his control during the Biden administration of Starlink's internet system when it came to Ukraine's ability to access it. And at different times during Ukrainian military operations, he decided in part after calls with the Russians to turn off Starlink access during critical military operations in Crimea. Now, you know, that decision uh... As far as and i can tell in the modern era has never been tested with a private individual to such a degree to the point where the pentagon was on the phone pleading with them to turn it back on later on they were able to come up with a contractual arrangement greater government control over starlink that could be used by the ukrainians but the fact is at a critical moment a proprietary technology owned by a private company uh... Was controlled by a single individual in a matter of national security. And this is far from the only example of that.

Jon Bateman: We are truly in uncharted waters here where the capital and sophistication required for cutting edge technologies like what SpaceX has built is so far beyond what the government itself can do now that the government is increasingly reliant on private actors for services that might be considered core governmental services, military communications, for example. So I grant you that it's disturbing to survey that landscape and say, private actor really has the government over the barrel here. Where I think I have intellectual conundrums is that the government sometimes isn't a trusted actor either. And so the Rorschach test on the other side of the equation is the dispute between Anthropic and the Department of War under Pete Hegseth. It's almost the inverse there, right? Where Anthropic as a private company. Was saying, we have our red lines. We don't want to be used for lethal operations. We don't want to used for mass surveillance. And there were people in other large AI companies who wanted to have similar stances. And in that case, a lot of people felt like they were the heroes and that it was the Trump administration who were the villains in seeking to bring to bear the long arm of the government and sort of forcing them to heed Trump's policies. How do we reconcile these two seemingly-conflicted case studies, where a lot of people seem to take opposing views on them? They kind of wanted the Biden administration to win over Elon, but they kind of want Dario to win over Trump.

Steve Feldstein: Yeah, no, I mean, it's a great one to point out, and I don't have a clear, a good answer for that. I mean part of it just goes to trust in government, you know, to the extent that you think that the Trump administration as it's currently run, the Pentagon in particular, is serving the national interest and best interest of the American people, then you would want to see its viewpoint prevail against a private company, to extent you are distrustful of their intent. And I am distrust full in terms of everything I've seen, in terms the dismantling of ability for powerful technologies in their desire. To harness and do whatever they would like with those technologies, then it raises a really profound question, which is that, you know, generally, I would err in the side of having a government which is accountable to the people through the ballot box, through congressional oversight, judicial restraint, and so forth. I would prefer to have them in the driver's seat. In this situation, I think the feelings were a little more mixed, where enthropping, at least at kind of a high rhetorical level, seemed to be on the right side of things. I will say this, that it's not as if... Anthropics technologies weren't already being used and contracted out to the Pentagon and being used for lethal operations. It was just that there was a very specific caveat that they wanted, which is no lethal, autonomous weapons use associated with Claude, and then no mass surveillance of American citizens associated with Claus. But everything else was essentially on the table, including very concerning allegations about how Maven has been used with regard to fighting the Iran War. Uh... And collateral civilian casualties uh... So you know i i think people like to have a clear villain uh... In a clear hero and i think as you kinda actually get into the details of this particular dispute it's a little muddier then one and the other

Jon Bateman: Yeah, going back again to my time at the Pentagon, this was during the time when Google had its big moral stand against working on Project Maven, which you mentioned, the major computer vision effort at that time in order to help with lethal targeting of drones and airstrikes. Google was a part of it. It then pulled out under employee pressure and came out with a set of public AI principles, including one which said it would essentially not be directly involved in killing. So it could do things like veterans care or help with analyzing data sets around pilot nausea or something like that, but it wouldn't be in the kill chain, so to speak. Now compare that sweeping moral stand to the much more narrow carve-outs. That Anthropic, OpenAI, and even Google's own DeepMind have sought and achieved with these language models. It's marginal by comparison. As you said, they're fine with being directly involved in the kill chain, and actually they are at scale. It's just not wanting to fully take the human out of the loop.

Steve Feldstein: Yeah, well, I mean first of all what I'd say about Google is that subsequent to that Maven Brujaha, it has quietly in the last couple years changed its policy so that it now does sell Technologies directly to the Pentagon and military contracting. So does open AI. They also change their policy and therapy changes policy You know a host of others Microsoft as well So so even those companies that initially took a stand in the late, you know 2010s have since changed their mind about... Taking military contracts, that's sort of one thing. And the second thing, you're right, when it comes to LLMs, I mean, there is no ambiguity about that. I mean I think every AI company has essentially signed up to have their models used in different ways to look through, sift through the data, to identify potential suspects, to be integrated within the MAVEN targeting platform, and then to be used for subsequent strikes And military operations. And it's not just the United States, it's also US companies working with other militaries, including the IDF.

Jon Bateman: It does somewhat complicate for me this question of like, is the government losing or gaining leverage over these tech companies? Who has whom over a barrel here, right guys? On the one hand, I grant you like, Elon kind of seems untouchable sort of politicians, particularly on the Republican side are quite afraid of crossing him. He has control over X, which has this remarkable megaphone and ability to influence politics. On the other hand, we just went through this tour of all the different AI and tech companies who have gradually given over more authority to the government and been willing to sign up for things that they wouldn't have signed up for before. And we see, of course, under Trump, he's taking equity in companies and bullying them essentially into giving over chunks of shares and all sorts of other unusual things. So it's kind of like there's a balance of terror between. The two sides and almost like this cat and mouse game. I don't know, how do you see that? That's a good name.

Steve Feldstein: For your first book, Balance of Terror, by the way, Big Tech and the Future of Humanity, right? There you go. Yeah, no, a couple things on this front. I mean, I think most importantly, you're right, that there is an interesting dynamic, although I would say this, that companies are willingly going into these contracts because they're lucrative and they want the revenue stream for a variety of reasons. So I don't know that there's so much being co-opted per se, is that they're making a voluntary choice to go into contracts. But The other point I wanna make, a broader point, is that let's step aside from the U.S. For a second. Let's look at every other country around the world, save China and Russia, a few of the other kind of allies of China. Look at what they're subject to when it comes to these other companies. Are they able to dictate terms when it come to cloud computing and hyperscalers? Do they have leverage to push back against them? And I would say the answer, by and large, is much diminished. So even if the U.S. Is sort of, you know, at par, and there's a question about who has more influence vis-a-vis the Pentagon or Anthropic. What influence does the UK have, or the Germany have, or, you name another country, or Turkey have, when it comes to Microsoft or AWS, or all the other compute infrastructure they rely on for very large companies, which make a lot of profits and are not beholden to those countries' interests. That sovereignty issue, I think, is a huge one. And so around the world, we might say, from Washington's standpoint, well, maybe someone even, but any other capital looks upon Silicon Valley's power and says, we have very little recourse to push back. We have very.

Jon Bateman: Leverage. That's a great point, Steve. I mean, you brought up the example of Elon turning off Starlink along the front lines of the Ukraine war. That's one thing. At least you might say, oh, the Ukrainians sometimes have other kind of forms of communication they can resort to. But if you imagine a future state where all aspects of military command and control are filtered through AI technologies. And. You would be foolhardy as an advanced military to use anything other than the frontier models, which are exclusively served by U.S. Companies. Then the power that an open AI, an anthropic, a deep mind, whoever has over you, is extraordinary. They could literally.

Steve Feldstein: Just cause you to lose a war. That's right. I mean, I know all sorts of levels. It's the AM models themselves. It's data centers and the chips that are running them. I mean it's all across the supply chain, you know, that for now is largely US dominated or Chinese dominated. Yeah.

Jon Bateman: Well, maybe that's a good place for our final question here, which is, where is all of this headed? We've hit on a bunch of different ideas here, from the US-China battle for control to the interior battle within the United States and other countries, public versus private. We also spent some time on drones and the proliferation, a dizzying array of relationships amongst many, many countries. When we have you back on in 15 years, what should we be talking about then? What will we be taking about then.

Steve Feldstein: There are two trends that I've highlighted in the book, which are seemingly contradictory. On the one hand, we've talked a lot about the centralization of power, whether it's in private companies or within the US or China, and that we see technology, particularly AI technology, only accelerating those pathways and kind of leaving the rest of the world behind. On the other hand, we've also talked about the diffusion of military innovations and civilian innovations around the world, the ability of smaller states, insurgent groups and others to access drone designs. To access just good enough AI models to do interesting things. And so the future is unwritten. Both trends matter. And which one will matter more, I think, is up for debate. So maybe in 15 years, we'll look back and say, well, was it the diffusion, the democratization of technology that ended up making the difference, or was it centralization amongst a couple superpowers and their superstar companies that ended winning the day? I don't have a good answer to that, but I'm curious to see how it pans out.

Jon Bateman: Let's find out, can't wait for round three, Steve.