Transcript
Is our military way behind other countries in terms of using the latest technology with AI, with drones, with biotech, with cybersecurity? I think for many years, it was. We're we're we're behind we know we're behind on supersonic weapons. Are we behind on AI? How did Hamas send undetected a 1,000 or so paragliders into Israel without Israel detecting it? Are we behind on, you know, the AI that's in sensors? Like, what is going on? So with the help of Chris Kirchhoff, who wrote unit x, how the Pentagon and Silicon Valley are transforming the future of war, we answer these questions and more. This isn't your average business podcast, and he's not your average host. This is the James Altucher show. Let me start off by asking, like, where is the US failing now? I feel like other countries are moving past us in a lot of a lot of ways with modern technology. Like like, we recently heard that on hypersonics, the ability to send missiles, you know, at speeds of, like, Mach 10 is is surpassing the US's ability to do it. On AI, we know places like China have no ethical qualms about pursuing the most, you know, aggressive forms of AI in or in order to create better weapons. Like, does the US have a chance in this new technological regime we're in? You know, it sure does, but it it is I I agree with you totally. It's counterintuitive. Like, how have we gotten to this place where all of a sudden, other nations are beating us at our own game, like, in a lot of dimensions, not only as you mentioned with, like, leading edge emerging technology like artificial intelligence, but also potentially on the on the battlefield, which is shocking. Right? Because, I mean, ever since the end of the 2nd World War, the United States has had sort of unquestioned conventional military superiority. But there are some reasons that have led us to this sort of astonishing juncture, and there are, I think, some emerging answers about what to do about it. And yeah. So and it's interesting you use conventional warfare. So, like, when I think of the the military race, the weapons race, we think of nuclear weapons. Like, oh, who has better and more nuclear weapons? And then it's just a sheer numbers game. Like, who has more planes and ships? And and I guess on the in terms of planes, who has faster planes? And none of these things turn out to be important in terms of winning a war. Like, who has more nuclear weapons doesn't make a difference in terms of who will win the war. And who has faster planes? It turns out, you know, we made all these, faster than the speed of sound planes, like the like the f fours, the f one zero fours. It doesn't really those aren't the planes that get into dogfights because now there's there's smart missiles guided, but there's drones, and so on. So it's not like these we don't make plane fast planes now. Like, the US doesn't make Phantoms or or other faster than sound planes because it's not necessary. So all of these things that were kind of measures of superior military superiority with conventional warfare are useless metrics now. Well, it really is true. I mean, just the other day, I was in a room, and there was a US senator, griping about, the fact that the number of ships the Navy fields had gone down. And, of course, you know, that's sort of like measuring your military effectiveness by how many cavalry horses you have in the barn. I mean, it's just not a a relevant measure. But it is important to step back in time and to realize that, you know, just like anything, the technology that drives military security ebbs and flows. And so, at one time, you know, the cavalry were the dominant force of war. And then around World War 1, you know, you got mechanized tanks, right, that clearly proved superior. And World War 1 was a war fought by battleships, but World War 2 is a war won by aircraft carriers. So that's another incredible transformation. And so I think the the thing to think about now, like, why is it that our military is no longer as dominant as it once was? You know, for a very long time, you know, this starting probably with the end of the Cold War, the First Gulf War, I mean, dropping bombs down, you know, chimneys. Right? The the US had grown up this, it's known as a precision strike complex that was just astonishing. But that was, you know, in the early 19 nineties. And, our adversaries have now had more than 30 years to study how we are successful and to purpose build their militaries around ways to be able to defeat our strengths. So that's part of the reason why we're arriving at this moment where all of a sudden, our margins on the battlefield have eroded substantially. Yeah. And, I mean, you well, there's a couple things you mentioned in the book that I wanna bring up, but let's actually talk about a recent example with how did how did the Hamas send over whatever it was, 1700 paragliders into Israel without Israel knowing? Like, Israel's got if anybody's gonna be ahead of us in technology, it's probably gonna be Israel. And yet somehow or other, their entire security complex was down. All these paragliders came in and just raped and pillaged for the whole day and started this devastating war that's still going on. How did that happen? How did we not notice with our satellites? Israel not notice with all of their sensors? Like like, just a basic technology thing seems to have gone wrong. You know, I I got a chance to visit Israel the 1st week in June in in Gaza as well. And, I mean, first of all, it's just such a tragedy because what happened on October 7th has kicked off a, you know, a cycle of violence that's just only spinning out wider and that, you know, has engulfed not only Israel and Hamas, but now some of the wider Middle East and and and Iran. Yeah. Lebanon. It's a really tragic situation, you know, moving away from a viable peace process rather than towards it. And so, you know, that does make you look at the the day what happened on October 7th as as even more tragic. And, you know, it goes to show that people that don't like each other will will come up with ingenious ways to harm harm the other side. And on October 7th, it started by Hamas using quadcopters, commercial quadcopters. They had literally bought off Amazon. Drones. Drones. Cheap drones. Not military grade drones, like drones that you and I would use to, like, film ourselves skateboarding or skiing. And they mounted grenades to those drones, and they dropped them on the generators powering the surveillance towers that looked over the border wall separating Gaza from Israel. So that's how the attack started. And then there were other tactical things they did that were quite unique, like, for instance, using paragliders to drop small amounts of infantry troops across the line. So So so when when they blew up, though, these, the the sensors on the surveillance towers using these drones, yes, it was basic technology, but doesn't Israel have technology to, hey. Our our surveillance, software just went down? Well, you know, they were pretty crafty. So as I understand it, they essentially took out all their surveillance towers at once. So, you know, like, that that Mission Impossible movie where all the security, you know, television, go blank at the same time. And, you know, Israel, you know, not you know, going back to the Maginot Line. Right? This, you know, fortification built to essentially prevent, you know, the attack that happened in World War 1 from being successful. And, of course, you know, the Germans just drove right around it rather than going through it. And that sophisticated fortification then, you know, 3 generations ago had fostered what turned out to be a very false degree of confidence that, you know, we're using advanced technology to protect it. There's no way for this advanced technology to be defeated. And, of course, that hubris led the Israeli military not to have troops in reserve and, you know, not to take a lot of basic steps that, you know, most militaries in that situation would have taken guarding a, you know, a a hostile border. So so okay. So this is one example. But now let's take the US. Even despite all the work you've put in and and is described in the in the book unit x, where is the US still lagging, or or where are you worried the US is still lagging on technology should that tensions result with China or North Korea? Like, in the book, you mentioned how you even thought we were we had round the clock satellite surveillance of North Korea so we could see what they're doing. And then they're and who would think North Korea is technologically savvy? But the reality is we we can't really see them with satellites as they're covered by clouds 200 days a year and and basic technology. We haven't to get past that, we haven't incorporated into the US military. Well, I I mean, we all wish that it worked like it did in the in the movies. Right? You know, like Mission Impossible and James Bond where, you know, there's a a super secret satellite that nobody knows about that swoops in to save the day. But, you know, of course, it doesn't work like that. And, you know, there has been a real shift. So, you know, in the book Unit x, we we begin the story in 2016. And if you had gone up to any 4 star general in the US military and said, I would bet you a steak dinner that in 10 years, drones are gonna be able to defeat tanks. They would say, you're on. You know? I'll bet you 10 steak dinners. And, of course, you know, they would have turned out to be wrong because here we are, in Ukraine, seeing drones, you know, demonstrably destroy armored vehicles. And not only that, but, you know, just, a few weeks ago, the Ukrainians had to withdraw from the front lines all 31 of the m one, a one Abrams battle tanks that the US had given the Ukrainian military because more than a quarter of them had been destroyed by Russian kamikaze drones. I mean, this is the most advanced tank in the US arsenal, in the arsenal of our adversaries. So I guess what I worry about now is, you know, no no force no 4 star general would would would take that bet today that a drone can't defeat an armored vehicle or a tank. But, if our most advanced How do they defeat a tank, by the way? How how does a drone defeat a tank? Well, you know, it turns out that, tanks are built essentially to fight other tanks and artillery. So they're armored to take strikes in particular from the sides, but some of their, topside parts are quite vulnerable. And if you can get a smaller charge up close, you might not leave the tank as a sort of smoking hole in the ground, but you'll disable it from being an effective fighting vehicle. So, you know, here again, inexpensive, you know, drones. We're talking, you know, a 1,000 or $2 of hardware plus, you know, $20 worth of high explosives taking out a tank that, you know, is is a multimillion dollar armored vehicle. And so, you know, that that leads me to ask the question, are we at one of those moments, right, where, you know, in the First World War, the cavalry, clearly met its demise, you know, with the rise of the tank? But, you know, could we be similarly today at the end of a century of mechanized warfare, of man mechanized warfare and shifting into a new paradigm. And if that's the case, then, and this is where, you know, we arrive at this paradox of how is it that the US military got into a very different place. You know, we've made a heck of an investment in tanks. We have a lot of them. You know, our allies have a lot of them. We have, like, planes and and ships that are designed to ferry tanks around the world to wherever they need to go. But if they arrive and they're not a viable weapon on the battlefield, then we've made a massive investment in something that doesn't doesn't any longer have the combat firepower it did just a few years ago. So let's say you were starting from scratch, and you could you could outfit the military with any technology you want and and ramp up whatever you want, for for these situations. What what would you do right now? Well, you know, some smart people in the Pentagon, I think, have, some great ideas already about what direction to head. And they've just started, about half a year ago or so, something called the Replicator Initiative, which is an initiative to build low cost autonomous weapon systems. So instead of building, you know, fighter jets, like on the cover of of our book, we have the f 35. You know, that's, you know, a $300,000,000 piece of equipment. Start with inexpensive but still fairly capable weapon systems and use mass, in other words, numbers to make up for perhaps a smaller number of weapons platforms that cost a lot more and get you just a little bit more capability, but but in the end could be overwhelmed by ma*s. And so there is a directional shift in the Pentagon now to embracing, you know, a much more autonomous, distributed, driven by AI set of weapon systems. But this comes along with a big debate. That's not only a debate in the era of the Pentagon, where all the 4 stars have their offices along with the secretary of defense. It's also a debate in Congress because it's one thing to have a consensus that, yeah, we should try out these new weapon systems, but it's another thing to say, we should probably stop buying tanks. We should probably shut down the tank plants, you know, and all the manufacturers in, you know, God knows how many states that produce them. I'm right. There's a whole complex here. There's a political story. You know, there have to be members of Congress that are willing to say, you know, it's gonna be really hard for my community to give up this plant that we've, you know, had for 2 generations cranking out tanks. But I'm gonna take a vote to do that, to, you know, take a bet on the future of war. That's hard. Yeah. So so what will happen? Well, I mean, we'll find out. Right? I mean, this part of the story is is not yet told. But as excited as I am as as all all of the things that Defense Innovation Unit has accomplished, particularly in shifting thinking, I mean, DIU now has a mission that Congress has actually codified in law. Congress, this year appropriated the budget of Defense Innovation Unit to be a $1,000,000,000, which is you know, that's real money. Right? Take a quick break. If you like this episode, I'd really, really appreciate it. It means so much to me. Please share it with your friends and subscribe to the podcast. Email me at alcatra@gmail.com and tell me why you subscribed. Thanks. And so let's let's describe that. So so this this unit, you started with your partner, Raj Shah, started in 2016. You got $30,000,000 in funds to start off with. Roughly, I mean, it was a battle. Like, everything with government, there's a there was a bureaucratic battle once you even got the approval. And the the job was to basically bridge this gap between Silicon Valley, which is working on all this amazing technology, and the military, which had nothing. What were some of the first successes of this? Like, how did you sort of prove out that this was a good idea? Well, I mean, first of all, the the military had a whole lot of stuff, and it was it was pretty amazing stuff. Right? So going back to the, you know, the f 35, the the fighter that's on the cover of our book Unit X. I mean, this is like it's a flying supercomputer who that, you know, is stealthy. Right? It has super advanced weapons. It can strike targets, you know, well over the horizon. It it's supersonic. I mean, it's an amazing piece of hardware. Right? It it it will take down any other plane in the sky, with with a comfortable margin of error. But the way it is produced is through a very bespoke kind of production, right, that the US grew up in the Cold War and that still exists today, such that the design of the F-thirty 5, the most advanced aircraft in the world, was finalized in 2,001. And it didn't become fully operational until 2016, 15 years later. And of course, a lot in technology had changed in those 15 years. So the whole purpose behind Defense Innovation Unit and having the Pentagon become a part of Silicon Valley, begin to access the consumer technology ecosystem, is that that consumer technology ecosystem, which is really separate I mean, we we have sort of 2 systems of production for technology here in the United States, one for the military and one for everything else. And that other system is faster. It operates under much fewer rules. It's much more free market. You can do things typically for much lower costs. It's much more agile. And the whole goal of the office was to bring these two systems closer together. I mean, it's interesting because it reminds me of of both the post office and and the arguments about the the the the 200 year old argument about communism versus capitalism. Is that basically and I'm not trying to take one side. I guess I am taking a side here, which is that when you have government, you know, planning things, it doesn't work as well as when kind of the private markets, you know, create things and and plan things. Like, central planning doesn't work. So if the government says, oh, we're gonna need all these fighter jets 15 years from now, that doesn't work as well as just kind of a bunch of venture capitalists getting together, funding something, and then seeing what what the market produces and delivers. Well, it's a I mean, yes. Of course. But, I mean, the cut the story is more complex because these 2 different technology ecosystems do fit together in very meaningful ways. And so if you, you know, grab the, you know, your the iPhone from your from your pocket, you know, about 80% of the technology in that iPhone wasn't developed by Apple. It was developed by DARPA and other federal labs in the, you know, sixties, seventies, eighties, nineties. Even Siri started off as a as a DARPA research project. So there is a a synergy between these two systems. So that's important to know. The second thing that's important to know is that, you know, this defense system of production that has strengths and weaknesses is purpose built for a reason. You know, you can't really have a free market per se for aircraft carrier builders. Right? You can't go on Amazon and, like, comparison shop. It turns out there's, like, 1, maybe 2 companies in the world that can build you an aircraft carrier. So what that means is you're not in a free market. You're in a system that economists refer to as monopsony, where there is one buyer and just a couple sellers. So the market won't set prices. You have to set prices in a different way, in a way that involves cost accounting and auditing procedures that make sure the taxpayer dollars are being used fairly in building this incredibly complex technology that takes years to assemble. And by the way, it might be around for 30, 40, 50 years. Right? The lifetime of some of these platforms is quite impressive. And so that system, as you have to imagine, has to run under a lot of rules and regulations and requirements that specify exactly how things should be built in advance, not knowing how technology is gonna change. And then there's, you know, an army of auditors, and each defense contractor has to have an accounting system that will plug into the Pentagon's auditing system. So, you know, in effect, that creates a system not unlike a public utility where, you know, we all have to have electricity. Right? You can't let the free market decide that we're only gonna give electricity to people in cities because there's a density of profit there that would motivate private capital to, you know, put up electricity wires. So these two systems, they do exist understandably in very different ways. And what happens in the military industrial complex isn't by itself crazy or without logic. It's just that we're moving into a world where that other technology ecosystem that moves much faster is now gone global. I mean, the world is flat in the sense that, you know, the microelectronics market is globally distributed now. So our adversaries are now able to buy a lot of the same technology the U. S. Military is. You know, if you go back to the Cold War, we were spending a lot more money on federal research and development. So you sort of assume that what DARPA had in the closet was 6, 7, 8 generations more advanced than, you know, what what we would carry around in our pockets. And now the reverse is true. The, you know, the iPhones that we carry around in our pockets are actually far more advanced than the processor in the f 35 because the consumer technology ecosystem has grown so much larger. So, basically, this is where you guys came in. Again, you built this this, like, a VC fund or, you you know, a a company to basically bridge this gap between this fast moving technology and maybe the slower moving decision makers who still need to make these decisions that only they can make because they understand what's going on in the war. And but did they recognize they have to move fast? Like, obviously, they did enough to fund your operation, but what happened then? So you started out. You're in Silicon Valley. Were you treated by other VCs as if you were a legit VC? Hey. We're Andreessen Horowitz. Come to the table with us and hang out, and we'll look at companies together. Well, showing up in Silicon Valley and saying, you know, hi. I'm I'm from the government. I'm here to help, typically doesn't get you very far. Although, like you said, there's one massive buyer there, and if you could get into that food chain, you're gonna do very well. Well, it's it's tricky, though, because even though, you know, the Pentagon's budget is, you know, $800,000,000,000 a year, right, which is enormous, and its weapons spend, what it spends on technology and systems is, you know, like, 200 and 80,000,000,000 or so. You know, the consumer technology market is $25,000,000,000,000. It it dwarfs, right, what the Pentagon spends. And not only that, but, you know, until we got there, if you wanted to do business with the Pentagon, you had to compete for contracts under what are called the federal acquisition rules or more specifically the defense supplement of the federal acquisition rules. And so if the federal acquisition rules are like the Encyclopedia Britannica, the defense federal acquisition rule supplement is like the size of a dictionary. So you have to figure out how to master that. And, you know, the typical contract process for FAR based contracts, you know, they run typically 18 to 24 months. So if you're a startup that's backed by a venture capital firm and you tell your investor, you know, I have a good lead on a government contract, but I'm not going to know if I get it for 2 years, They're gonna tell you, that doesn't sound great to me because I need to know that you're gonna be profitable, in about another year or 2, year and a half if I'm gonna give you your next round of funding. So the 2 systems were simply incompatible. And for that reason, in 2016, when we arrived in Silicon Valley, venture investors simply told the people they funded, hey, don't even take a meeting with the government. It won't help you get your next round of funding. In fact, if you try to break into the government market as a young startup, we actually will withdraw our funding. We'll take it away because we don't think you can win. Wait until you go public and you're much bigger and you can hire lobbyists and specialists and, grow a government business. Don't do it now as a start up. And so what would you do? What what worked for you? Well, you know, there were people that were back in in Washington in the acquisition community, the people that run this massive system of of how the Pentagon buys its its technology, beginning to imagine how can we make it easier? How can we break into this other technology ecosystem? How can we reimagine a different world where we're not the prime mover and we have, you know, 5 or 6 defense primes who are custom, whose whose whole mission is to is to meet our every need, right, and to cater to us? How can we enter a market where people are skeptical of us? How can we make a contract on commercial terms that doesn't require you to have an auditing system that's not commercial specific, that's, you know, bespoke for the Department of Defense? How can we we waive all of the red tape that we ordinarily would impose on a public contract and know that it's safe to do that because we're accessing technology where prices are generally set in a free market, where you don't have to have that system of cost accounting to ensure taxpayer dollars are are are being invested wisely. So, you know, and this is an incredible story. It goes to show how one individual can can really make a difference. But on our staff was a 29 year old woman named Lauren Daley. We met the 1st week in the office as we were getting to know the people, that were at the unit. And she said, hey. You know, there's a bunch of people in the acquisition community that have this kind of wild idea for how we could do things differently. And it all relates to a single clause that was inserted into this year's National Defense Authorization Act, the act of Congress that funds the department. And I think this could allow us to actually work with startups on terms they would they would be willing to do business with us on. And so my my coauthor, Raj, asked Lauren, you know, well, it's great. Like, can you maybe, like, write up a paper that describes how we could actually do this? And and so she she hands Raj a 20 page paper. So I've actually already written one. So we read it, and we realized that Lauren was really onto something. There was this powerful new authority that a renegade senate staffer had, like, literally snuck in the dark of night into this, enormous law. And so a couple days later, Lauren and I flew to Washington, and we met in quick succession with the head of acquisition policy in the Pentagon, the head acquisition lawyer in the Pentagon, and ultimately with the general counsel of the Department of Defense to make sure everybody saw what we saw in the new law and, and grew a consensus quickly that we should try out a much faster means of contracting. And so within 2 weeks, we codified a new policy that allowed us to let a contract between 30 to 60 days and to do so on on commercial terms, which was remarkable. It had it had never really been done before in that way. And and so what what did you do with it? Well, we then went around to firms we knew had technology that could be useful military missions and and and useful right away. I mean, not something that you'd have to, like, engineer to make work on the battlefield, but something that would be kind of turnkey. And we rapidly set out to begin technology pilots, and we did, I think, 10 or 12 of them in our our first 100 days. And I'll be darned, but, like, a few of them worked really well. Like what? Well, you know, of all things, we discovered a flying car company in Silicon Valley called Joby Aviation. Joby Aviation was at that time a stealth startup, so they were not advertising themselves. But of course, we knew people who knew them and we got introduced and they were developing an electronic powered VTOL aircraft. So literally a fly in air taxi that can take off and land vertically but fly horizontally at a very high rate of speed. And we knew that this technology could be really interesting to the military. Right? I mean, you don't need a runway to land it. Unlike a helicopter, it's actually quite quiet. Also, it's smaller. So rather than loading a lot of people on a helicopter, which can be very vulnerable and we've had some very tragic incidents of losing large numbers of people in helicopter shootdowns, you could desegregate forces. And Joby, they were very well capitalized. So they actually weren't interested in at first in working with the military because they asked the smart question, like, what's in it for us? Like, we have plenty of money to keep developing our technology. But it turned out we had something they did need, which is that we had military test ranges. And it's much easier to get on a military test range than it is to get on a civilian test range, much less paperwork, much quicker. I I would think you had something else that they needed, which is actually legal viability for in use cases. Like, it's unclear. Like, there's 100 and 100 of flying car companies out there, and and if you Google if you go to the Google's patent site, there's gazillion patents on flying car technology. No no city is ever gonna at this point, years years later, no city is and even close to approving flying cars. Maybe San Francisco, but that would be about it. And but the military can actually use these things. Yeah. So that's true too. So, you know, we did we were able to look into some hey. There's a there's an orthogonal mission set. And, by experimenting your technology on the mission set, you're gonna get a lot of data. You're gonna get user feedback, and you're going to be able to fly quickly on our test range. And so began a multi month period of experimentation. DIU actually acquired the first Joby aircraft. Fast forward 8 years later, the Joby eVTOL aircraft is presently undergoing operational testing with the Air Force. It's about to become part of the year the US Air Force fleet, which is an And what will it do? Well, it will perform a number of potential missions. You know, one is transporting people. Another is transporting supplies. And because the vehicle is can be both piloted but also has autonomous capabilities, you can just imagine how this would be useful to move and resupply troops or forward positions that are in areas of danger and risk, where you wouldn't want to ask a helicopter crew to go in if you didn't have to. So there's a world of possibility that the Air Force will have open to it, as they begin to acquire Joby Aircraft at scale. It seems like also you've done a lot of stuff with drones. And like you mentioned with Ukraine, these kamikaze drones in mass are are taking out the last generation of military innovation with the the 100 year old technology of tanks. So so it it seem and and drones is something that's been fairly big in Silicon Valley over the past 10 years. What what are some of the successes with drones? So very early, at at DIU, we were aware of, you you know, things that were happening on the battlefield in places like Syria, Afghanistan, and Iraq. And, there were 2 things happening. 1, ISIS and Taliban fighters were actually acquiring commercial drones, you know, commercially off literally off Amazon, primarily through the company DJI, the Chinese drone maker that makes the most popular drones. And they were using them against US forces, And they were using them to do things like disrupt special forces raids because the drone would be able to spot our assault teams coming. They also, at times, early in our existence, used them to drop grenades and actually kill US troops. So that called the attention of battlefield commanders. Hey, our fairly unsophisticated adversaries that don't have, you know, their own predator drones or their own helicopters or their own fighter aircrafts or nevertheless finding ways to defeat our very sophisticated technology using cheap consumer drones. And then the other thing that was happening was our forces had access to very sophisticated high end drones, surveillance drones, and they had access to some tactical drones, but nothing as simple and easy to operate as a DJI drone. So some units were literally, taking DJI drones with them into combat, some US units, and, using them for reconnaissance, using them for surveillance. And that was a problem because DJI drones being Chinese made, we couldn't be sure that they weren't recording where they were in the world and what they were looking at and then uploaded that information ultimately to to Chinese servers. So in response to these developments, Defense Innovation Unit, and this is something we disclosed for the first time in in the book, started a special task force called Rogue Squadron. And Rogue Squadron was the military's 1st commercial drone unit. And it was started by 2 real renegades, Mark Jacobson, an Air Force at that time, lieutenant colonel, and Ryan Beal, a Navy helicopter pilot who was also a coder. And both of them were drone hobbyists. They loved drones. Like, between the 2 of them, I think they owned over 50 different commercial drones that they just had been, like, playing with kind of Mad Max style. And so they set up literally in the parking lot of Defense Innovation Unit a test range where they could fly the drones they had, drones, prototype drones from startups in Silicon Valley, companies that were testing out counter drone technology and began they began to experiment. What could we do about adversaries that are using DJI drones? Are there ways to disrupt them? Are there ways to, if we capture one of them, exploit it and figure out where the operator launched it from? Can we take DJI drones that are being used by American forces? Because they're the only consumer drone on the market, you know, that has so much capability. They have, you know, 90% of the market share. And can we make them safe so that we can be sure that the locations where they're being used won't be disclosed, won't slip out to somebody else's computer servers. And in short order, they did all of these things. And that led to some really important programs that are still ongoing in the military today. One program to make commercial drones safe, to literally declare, okay, you know, these drones are safe to be used by the U. S. Military, by other government entities because we can be sure there there's not a counterintelligence risk, and then other efforts to create both offensive and defensive drone technology that, today is being used in Ukraine. Yeah. And then and then, Palmer Palmer Luckey, the guy who who started Oculus, which sold for 2,000,000,000 to Facebook as a virtual reality headset maker, he started a a drone company? Yeah. I mean, so, Palmer Luckey, you know, clearly, one of the most incredible technologists of our generation, came, to visit my coauthor, Raj, at Defense Innovation Unit along with another venture capitalist, Trey Stevens, somebody that had, a lot of national security experience, had himself worked in the intelligence community and then worked for Palantir before going to Founders Fund, one of the more prominent venture capitalists here in Silicon venture capital firms here in Silicon Valley. And they were trying to, imagine, hey, if we could create a new company focused on this whole new generation of advanced defense technology, how should we do it? What are the most important problems that software and hardware designed by leading product engineers could solve? So they came by to have a chat with Raj to kind of brainstorm together and that brainstorm before they even sort of decided to formally build a company and ultimately named it Anduril, gave birth to today what is one of the most important new companies in the world in terms of the technology it's producing. And, you know, fast forward just a few short years, Anduril, in something that really was a surprise, an astonishing development, won a major contract from the U. S. Air Force, probably the most important contract the U. S. Air Force has let in the last 20 years to develop something called combat collaborative aircraft. So these are supersonic stealth drones that will fly alongside f 30 fives and other manned aircraft, and give the Air Force whole new capabilities, you know, swarming a fleet of swarming supersonic stealthy drones that can both perform reconnaissance missions but also can perform strike missions. And, the 2 final performers, as that contract competition went forward, were both nontraditional. 1 was General Atomics, not, you know, necessarily known for building aircraft, and then the other was was Anduril. And this is a huge success. This shows that what Defense Innovation Unit set out to do is actually happening in that we now have venture capitalists that are viewing the military as a viable customer, as a customer that's not going to just make its startups go bankrupt if they try and compete for contracts. And that in turn has led a whole new generation of founders to say, you know, I actually wanna focus on building the next generation of military technology. And so you have not only Anduril, but other companies like Shield AI, like Joby, like Saildrone that are becoming unicorns, you know, whose valuations are now north of $1,000,000,000 joining the first, you know, league of defense startups, SpaceX and Palantir and this whole new generation of of companies that are now, selling, really advanced technology developed in modern ways by leading Silicon Valley engineers to the military. Well, you know, and you mentioned you mentioned Palantir. Wasn't Palantir initially funded by In Q Tel, which was kind of the CIA equivalent of what you were doing with DIU? So so there was kind of this something already happening, but there's it was more on the CIA side in terms of of of, you know, funding startups. Yeah. In Q Tel is a really, amazing and and interesting, in institution. It it is actually a a venture fund, unlike Defense Innovation Unit, which simply invests in r and d pilots. So it doesn't actually, you know, buy a stake in a company. It it it just buys their product to try out. In Q Tel was created in the wake of 911 by some visionary people who realized, you know, as our boss secretary of Ash Carter did, that the US government's monopoly on advanced technology was starting to dissipate, that the rise of companies like Apple, Amazon, Microsoft, each of whom had, you know, Google, each of whom had a market capitalization larger than the entire defense industry combined, meant that the locus of innovation was shifting out to the commercial sector. And so unless the intelligence community was also able to access that technology ecosystem, the sort of James Bond Q shop of special toys, you know, would eventually be out outpaced by by what you could buy in the consumer market. And so In Q Tel has been out in the valley for a long time and invests fairly small amounts of money as a way of observing what technology could be used by the intelligence community and then making introductions between those companies and potential customers within the IC. And so that model was important for us to replicate in many ways and build on, but for the Department of Defense, which is a very different creature than the intelligence community and has a different mission set and as well as a different scale to it. And and where did you guys intersect with AI? Like, how did you start bringing AI I know AI was a big focus for you. So how did you start bringing AI into up to speed for the military? Well, when we arrived in Silicon Valley in 2016, that was a year where, in particular, machine learning was really coming of age. And so, you know, machine learning was beginning to be, you know, as accurate as a human in spotting certain things. So that's really incredible. Right? There's, you know, a machine almost as good or even in small use cases better than human eyeballs at processing data or information. So so right away, we realized, well, you know, here we are still sort of fighting the wars in the aftermath of 9:11 in Iraq and Afghanistan and facing, insurgents that are implanting, you know, IEDs, which which can still be destructive even to our most advanced armor. So we built this massive surveillance system to surveil the battlefield and to try and track insurgents, spot them, plant in IDs. And so there's one device in particular called GorgonStair, which which is a drone or you can actually put it on a balloon as well, an aerostat. And it will essentially use hundreds of cameras to look at an area the size of Disney World and take, you know, multiple snapshots a second of what's going on, you know, both video and photography. So this is a massive amount of data to process, you know, way more data than any team of human analysts could, day in, day out, keep track of. So, we, along with some others in the Pentagon, began an initiative called Project Maven, which was an experiment to see, could machines and machine learning algorithms begin to process this data in ways that could flag places where human analysts might want to go back and look? So can we tell the difference between a tank and an ambulance? Can we identify markers that you would want to go back and have an actual human intelligence look at from this, you know, these gigabytes and terabytes of imagery that's coming off multiple platforms? And that led to a real breakthrough, which which showed that, yes, you actually can use machine learning much more efficiently than the limited number of intelligence analysts you have. And it ultimately led to the creation of something called the Joint AI Center in the Pentagon, which was an attempt to centralize a group of experts, technologists, and military strategists to seed AI across the military and to take, take it into many more different mission sets. And what what other successes have come out of that? Like, what other technologies are we using now? Well, on artificial intelligence, I mean, everything from, using it to predict, aircraft maintenance. So, you know, every modern aircraft has got a lot of parts in it. They all break at different rates. You can replace them on a set schedule, but you're still gonna get surprised based on the conditions you're deploying the aircraft in. So, if you can just figure out a few parts that are likely to break, you know, the next flight cycle and go ahead and replace them now rather than have an aircraft have to return to base and not be able to complete a mission, that's a huge gain in operational performance. And so one of AI's other early successes in the department was helping increase the number of flyable hours that aircraft and unmanned platforms were able to achieve. AI is is now more recently being used in even more advanced mission sets. So one of the most successful defense innovation unit programs today that happened long after Raj and I left the unit happened with the Navy's underwater mine detection mission. So, you know, mines, which which are military technology that actually date to ancient China, I think, like literally to the Ming dynasty or, you know, 100 of years, maybe not that far back, but 100 of years ago. You know, mines were also used in the American Revolutionary War, in the Civil War, and they're still around today. And it turns out particularly in some of the most likely potential complex scenarios, including Taiwan, mining could be very decisive. So the Navy works hard at being good at having autonomous vehicles that can, as well as ships, can go and detect where mines have been planted and then diffuse them before they collide with a Navy ship or a submarine. And artificial intelligence underwater proves to be really important to increasing the efficiency of an underwater water vehicle, to be able to notice, hey, what's that anomaly over there? You know, maybe I should go take a look at it now and have an algorithm that will redirect, either this underwater vehicle in that direction rather than have to wait for data collectors to, you know, pull the tape from the from the vehicle, process it and say, oh, shoot, I wish we would have you know, now we have to mount another underwater vehicle to go take a second look. So little upgrades like that have, you know, in cases doubled or quadrupled the mission effectiveness of some of these underwater vehicles, which which, you know, is a very important development for for the US Navy. Are there any technologies that have kind of ethical issues in commercial use that you wouldn't have in, like, military use? So for instance, facial recognition. So, obviously, facial recognition in the consumer space, it was kinda floated out there and then quickly slapped down. You know, oh, we we think we might see you in a photo. I would get these messages on Facebook, but then that technology sort of went away from from commercial use for for good reason. But, obviously, with military, use, it could be important. Like, oh, we have this satellite footage of this marketplace in Afghanistan. Turns out there's the one of the big military leaders is there because we recognize them from the satellite using our facial recognition technology. So anything any initiatives like that, or or do you get slowed down by the commercial ethical issues? Well, I mean, it's always important with technology, whether it's in a consumer context or a governmental context or a law enforcement context or a military context to be very attuned to ethical issues. And as we experienced with Project Maven where there was a large protest by some Google employees who were caught off guard by the fact that their company was engaged in this technology pilot and also had a misunderstanding based on a fairly sensational headline. You know, this is a story about how one headline can really, impact a lot of a large number of people into into into thinking about something the wrong way. The headline implied that Google was working on an offensive drone warfare project rather than a defensive surveillance analysis project. But in any event, you know, technologists that are producing this technology have, of course, a special responsibility to being attuned to its use cases and to speak up when they see potential harms because they may well see those harms before others, you know, as do, of course, you and I as as citizens. So, yes, certainly, there are there are examples like face recognition technology that, you know, you or I would not want to be deployed, you know, by our local mall security, right, or potentially by police forces in our city, but that you would imagine, you know, we would we would both feel comfortable deploying on the battlefield. But there's actually, you know, it cuts another way as well, which is that, you know, if if if, you know, you're you and I are using a commercial laptop and it crashes and has to restart, you know, it's no big deal. Right? I mean, maybe this podcast gets interrupted. You know, maybe you get caught out of a Zoom meeting for a few minutes. But if that's the computer running an airplane that you're flying in, it crashing and restarting, could be a real problem. So, you know, there is a phrase that's fair that says beta testing can kill people. And that speaks to a part of the skepticism that is fair from people in the military that say, you know, for so long, consumer technology, sure. I mean, it was fun. Like, my kids have iPhones, but do we actually know it's going to perform on the battlefield in different conditions, in high temperatures, under extremes, under high g loads? So there has to be a way also to make sure that commercial technology that may work well in everyday life will also work well on the battlefield. Yeah. So with with facial recognition, is this something that is starting to be considered for military use or or, you know, these types of technologies? Well, there's, yeah, there is a, you know, quite a history to this, and this goes back to, actually, the Iraq war where, you know, during an incredible insurgency, the US military was doing its best to try and secure the Iraqi population and make them safe by figuring out the you know, who the small number of insurgents were, and and and trying to to take them off the battlefield. And, you know, it's it's really interesting. I I got a chance to serve as a civilian in Iraq in in parts of 2,000 6 and 2,007, so the most violent years of the war. And, you know, just 1 or 2 people out of a 100, it turns out, can really cause a breakdown, you know, in sovereignty, can take a, you know, a village or a town from a place of relative peace to one of extreme violence where everyone's stuck inside their homes. And so even then, there was a question of, you know, biometric identification. How do we know when we capture somebody in a, in a house we just raided who they actually are? And, you know, have we captured them before and released them because we thought they weren't actually part of an insurgent cell, but then we find them again with explosive residue on their hand? So, yeah, indeed, these are the kinds of technologies that matter because on the battlefield, they help you separate combatant from civilian. And so you would want, you know, a military, particularly one that abides by the laws of war and does its best to minimize harms to civilians to be equipped with technology that's very good at identifying people. And and is the technology getting there? Is the is the technology good now? Well, this is not this is a little bit outside my domain, but, this has indeed been a huge focus of of the intelligence community and and the US military. And, I have to imagine that, the technology today is a lot better than it was a generation ago. I mean, given what you know and and your experiences with the with the DIU and and, again, bridging this gap between Silicon Valley and the military and all the things you discussed in your book, if you were to start a company today to bridge this gap, what would you what would you focus on? Well, we're in this interesting moment, right, where, you know, again, the civilian technology economy, the consumer market has grown so much larger than, the military one. So so I would start a company that would take areas where the consumer technology market is clearly in the lead and apply that to technologies that would help us make us more safe and secure. So, you know, artificial intelligence is a sort of become a canonical example. Right? I mean, this is arguably the first major technology that, yeah, sure, there was some government r and d much earlier, but the generative AI wave that we're now living through, which is extraordinary. And it's like a Thomasin, as in light bulb movement that was produced entirely by the private sector, by 3 companies, 4 companies, 3 of whom are within 10 miles of where I am right now. And so I would want to position myself to take those leading edge technologies and then imagine the ways they can be used differently than our, you know, the existing technologies we use to keep ourselves safe and secure. Like like, what specific like, okay. If I were to throw something out, I would wanna create something that could scan all emails sent to anybody to see what might be sus to flag what might be suspicious, you know, terrorists talking to each other about plans to attack something. And, again, ethical qualms aside, like, that would seem to be a useful thing in the military and and possible, like, in the sense that you could probably there's probably plenty of examples of suspicious emails over the decades that you could analyze and learn from to to create an an AI language model that could at least help flag potential dangerous emails? That's, I mean, certainly an intelligence community mission. I mean, the things that I think about as somebody, you know, as a national security strategist, you know, that are certainly top of mind are, you know, how do we make sure we don't get into a great power war, you know, a war between truly advanced militaries. Right? I mean, we haven't had a great power war for 3 generations now since the 2nd World War. And what happened there was horrific. It was on a scale unlike any of the other conflicts that we're seeing today. So, and that, you know, one certain worry that I think, you know, more people are talking about now is is is the United States and China, You know, as our relationship, which is one still today of incredible economic interdependence, is running into more tensions, tensions about Taiwan, tensions about China's behavior to its neighbors, tensions about China's economic coercion of other nations. So as we try and think about rising global tensions between 2 superpowers that are both very advanced, how can we how can we keep the peace? And, of course, one of the ways of keeping the peace is to field military systems that alter the calculus of your adversary, that make them think it's just not worth it to try and go to war, that even though for our own reasons we might want to, make Taiwan, you know, our long lost province come back to the homeland, you know, maybe maybe we need to shift the calculus so that the cost of doing that are simply too high. So those are some of the military problems that I think about when I try and imagine how how new technology could actually make a real difference. How would how would you make the the cost of invading Taiwan too high other than with nuclear you know, the threat of nuclear weapons? Well, you know, so Taiwan's an island. Right? I mean, you have to get there. Right? You have to cross the sea. And so I think that's where a fight over Taiwan would would take place. And so you have to imagine, under incredibly pressurized circumstances, far more so than today on the battle lines of Ukraine, you know, with massive electronic warfare capabilities on both sides, how in a very contested environment would you be able to stop an invasion? So a technology that's autonomous, that can fight through jamming, that can operate in areas where it's not gonna be able to connect with global positioning systems, which where it's gonna have to figure out where it is on its own. Those are the kinds of systems, I think, that will be most important to, deterring, conflict in in, in Taiwan. So, you know, it seems like it's gonna be very hard ultimately as technology improves. Like, let's say let's say drones could be the size of mosquitoes, and let's say they all they're all equipped with GPS and facial recognition software and explosives. So you you could imagine China sending in a a a a million of these mosquito drones into Taiwan. Oh, I recognize this person. He's the leader of Taiwan. Here's the generals of Taiwan. And, you know, in in one hour, you know, go at supersonic speeds from from from China over to Taiwan, high enough that it can't be affected by any, you know, ships that might be protecting Taiwan, and just take everyone out that they need to very quickly. And there's almost nothing we could do about something like that as the technology improves. Well, you know, this this brings to mind, in older literature in political science that is that is newly relevant on, something called strategic stability. And, you know, one of the counterintuitive outcomes of the proliferation of nuclear weapons, so nuclear weapons developed, you know, as we've all seen in the movie Oppenheimer here by the United States in Los Alamos, but quickly stolen by the Soviets. From there, nuclear technology proliferates in the next couple decades to a small number of other nations around the world. So, you know, the question the question is, you know, did nuclear weapons make the world a more dangerous place? I mean, these are horrifically destructive weapons, or did they actually make the world a safer place because they reduced the likelihood that nations that had nuclear weapons would actually go to war with one another because you know, nuclear exchange is no good for anybody. And indeed, in certain ways, the advent of nuclear proliferation has reduced great power conflict demonstrably. Now, you know, there are limits to this. Would you want everybody in the world to have their own personal nuclear weapon? You certainly wouldn't. And so one of the open questions that we should all be alert to as citizens is, is some of this new autonomous technology driven by AI going to increase strategic stability? Is it going to make great power war, make war less likely because of the high costs involved or war to break out? Or is it going to in different ways destabilize the international system where one nation gains a sudden advantage that changes the calculus it has about whether it will go to war against another in a particular circumstances. So these are all things that are now being very, very actively studied and debated, not only by national security strategists, but by a lot of other Americans who are you know, I've seen all the Hollywood movies, right, about the Terminator, and should be very focused on how technology is changing, our world. Yeah. I guess even in, like, cybersecurity, like, you look what happened a few weeks ago with CrowdStrike. Now we think that wasn't some sort of hack by a foreign, you know, actor, but, it could have been, and we see how quickly it it, you know, took down almost the whole infrastructure of, you know, many industries across the world. And we could see as as cybersecurity gets even more sophisticated, particularly when combined with AI, that could easily happen again and again and again. And I I worry we're we're heading into some kind of black hole where we we the technology gets so good of offense, and there is no defense, and we're just in nonstop trouble. Well, you know, this this brings to to the fore another concept that comes out of the Cold War and strategic stability and military history and strategy more generally, which is the offensive defense balance and how will technology change it. And, again, anytime you experience a major shift, something that gets the gives the offense dominance, you know, then you may end up in a situation where where countries, if they have a reason to go to war, will, judge that, well, you know, maybe we actually have a shot at achieving our objectives at a cost that is acceptable to us. And so, you know, you can imagine a nightmare scenario, right, where AI gets very good at hacking or creating cyber disruptions. But it's equally true that AI, you know, as DARPA has demonstrated in some very advanced experiments, including their cyber challenge, that AI can also be used to, autonomously patch systems and much more quickly detect vulnerabilities and code than human cybersecurity professionals. So, you know, again, it's an open question. Will really powerful AI models in the end be a threat to cybersecurity or will they enhance it? What do you think? Well, I think the jury is out. But, if I were an advanced country, that requires a lot of, you know, microprocessors to make my daily life work, I mean, my toaster downstairs has got a microprocessor I would want to make a lot of investments right now in AI tools that will enhance cyber defense because it's my interest to make sure my very vulnerable and tightly coupled infrastructure will stay safe. So all these things should hopefully be guiding the investments we make as a nation, as a society, And and we better get them right because if we get them wrong, we're gonna be exposing ourselves and our way of life to, to, risk. Well, from your mouth to god's ears or at least the government's ears as they start purchasing more and more of these technologies, and and you you you did a great thing by starting this this DIU to to bridge this gap between, the technology of Silicon Valley and the needs of the of the military. And you and and Rod Shai, your your partner in this, wrote it write about it in the book Unit x, how the Pentagon and Silicon Valley are transforming the future of war. So really great stuff. The we only touched the surface of it. There's so many technologies that were being left in the in the dust on, and now we've been catching up, and and you describe a lot of this. And and it's also a good textbook on how to deal with bureaucracy. It was really fascinating how you you worked around some of the issues and some of the problems that you almost found yourself in. So very riveting story, unit x. Thank you so much. Thanks for coming on the podcast. Thanks for having me on.
Comments