From Silicon Valley to the Pentagon: The Future of Defense Innovation

0:00If we were designing a defense procurement system from scratch today, we would not design the one that we have. When we talk about two, three, four years to build an aircraft carrier in two billion dollars, and it being destroyed on week one of any hot conflict with an adversary, you say, is that really the force that we want to build? It's innovative approach and innovative technology versus not. The authorities and policies all exist. You can do it. It's about making it so that this becomes more commonplace, more understood, and these patterns of how to acquire differently become kind of established.

0:38Part of the show intelligence, biotechnology, autonomy, energy, the pace of change in those spaces is, and frankly, likely always will be faster than it is an purely bespoke defense -oriented way. We talked about how prepared we are. I don't think he'd be asking that question and we wouldn't be all here if we felt really prepared. The last few decades have been a period of transition for defense and increasing number of startups have begun to rival large defense crimes. The industry has also gradually become a calculus of both hardware and software and exponential technologies have forced the DOD, of course that's the Department of Defense, to rethink how it's traditionally done business.

1:22In fact, these changing conditions were some of the inputs that resulted in the Diu, the Defense Innovation Unit starting in 2015 within the DOD focused on accelerating the adoption of technology. And today you'll actually get to hear it from the director of the Diu. That's Doug Beck, alongside founders and funders, equally looking to advance this partnership between Washington, D .C. and Silicon Valley. In fact, joining Doug for this conversation recorded in the heart of Washington DC back in January at A16Z's American Dynamism Summit are A16Z General Partner and co -founder of the American Dynamism Practice, Katherine Boyle, Plus, Brian Schimp, co -founder and CEO of Anderl, Defense Technology Company, specializing in advanced autonomous systems, and finally, Brandon Seng, a company building the world's best AI pilot.

2:17So, what is the next wave of defense innovation really look like? What is America realistically sit competitively? And how do we build a technology pipeline that poses us to effectively defend against our adversaries? Let's find out! Oh, and if you'd like to get an inside look into A16z's American Dynamism Summit, you can watch several of the stage talks from the event, featuring policymakers like Congressman Jake Ochenklaus or Senator Tadyon. and of course both founders and funders building toward American dynamism. You can find all of the above at a16z .com slash 80 summit. All right, let's get started.

2:59As a reminder, the content here is for informational purposes only. Should not be taken as legal, business, tax, or investment advice or be used to evaluate any investment or security and is not directed at any investors or potential investors in any A16z fund. Please note that A16Z and its affiliates may also maintain investments in the company's discussed in this podcast. For more details including a link to our investments, please see A16Z .com slash Discoachers.

3:30So Doug, DIE was founded as DIEUX in 2015. Why was it founded and how was it evolved over the last several years? Yeah, so back when DIU was founded, which is back to 2015, as you said, Secretary Ash Carter was truly prescient in seeing that there was no way for us to effectively counter the rising challenge of the people's world, China as well as the other China's world, without fully harnessing the power of the incredibly vibrant commercial tech sector that we have in the United States. And back then, it was sort of ironic because the Silicon Valley that any of us grew up in was actually formed out of the collaboration between government, the private sector and academia, its support of national security, that's actually how Silicon Valley got its start.

4:16But despite that, we have lost that connection largely. And so, D .I .U. X then was really started in order to build that bridge kind of app all between the commercial tech sector and the department in order to begin to harness that. And I think of that as D .I .U. 1 .0. D .I .U. 2 .0, which followed on it relatively quickly, was about proving that you could take a real military problem and commercially derived technologies, bring those things together, use other transaction authorities that we already had on the books, but didn't know how to use, and use all of that in order to put real concrete prototypes in the hands of the warfighter and do it within weeks or months, not years and years.

4:58and that has now been done lots and lots and lots of times. DIU has done it between 16, 80 times, spending on how you think about it. When we built all these other innovations and it is around the department to help do it too. And that is amazing, but it's not good enough yet. Because we have to do now is take that capability and apply it for true strategic impact, which means that first of all, we've got to focus on those most critical problems that we need to solve in order to ensure that we can deter major conflict or win a forced to fight. And then we have to do the hard work across the department of scaling those solutions so that they can have that level of impact.

5:35And doing that is what DIU3 .0 is all about. That's why the Secretary elevated DIU to be a direct report. That's why I messed up my perfectly good life to come join the team and why I'm so excited to be here doing it. Yeah, and what you're describing, I mean it's so remarkable for people who founded companies in 2015, 2017. The interactions with startup was very, very different at that time. Brandon, I know Shield was founded in 2015. I'd love to hear what it was like interfacing with the department, just as DIU was being founded before there was really any emphasis on startups. There's a lot of hesitation back in 2015 in 2016 around this idea that small companies, you know, we had just raised a seed round in 2016 could actually make a material impact for the war fighters.

6:18But by working with organizations like DIU with some of the forward leading organizations in SOCOM who are willing to take a risk, take a chance with these smaller, call it more agile, more software -focused companies. I think they started to see the return that they were getting on their money. And that's not just in terms of the product, but the heft of venture capital that was followed, you know, whenever we would close these contracts, you'd see more money flow and from the venture capital that would allow us to build better products for the customers and customers, it really started to click with them.

6:51I remember the conversation in 2017 after our series A round with Andrews and Horowitz, the customers like, holy smokes. I just want to recreate this 100 times over because we just let a $1 million contract you're getting 10x in terms of the product development capabilities behind the company that's working with the customers. So it's fundamentally changed over several years. I still think we haven't penetrated all of DODs certainly, but instead of being 1 % of the customers maybe we're at 10 % of the customers in DOD. So I think about that adoption curve of technology, of processes too, and we're starting to make real headway in great credit to the venture capital community, the defense community organizations like DIU.

7:34And then, yeah, certainly I think the work of companies actually following through and delivering capabilities that are making a difference matters as well. Brian, I'd love to hear your take on this because I know one of the things that D .I .U and other organizations were really focused on was not only having more education about how startups work within the D .O .D., but also really educating founders who maybe had never worked in defense. They didn't have, say, your background of having previously worked in defense. Walk us through how that ecosystem has changed and how D .I .U. has been a part of that since 2017 when you founded Anderl.

8:03Yeah, the business model side, I think, is probably the least obvious to anyone who doesn't live in this world, but is probably the most important, which is the way that the the government typically acquires technology as a very planned view to what technology needs to mature over a very long period of time. And the whole system was predicated in a belief that you can curate and know where technology needs to be to accomplish your military ends. That was sort of the whole genesis of the current system. What has changed is the percentage of R &D happening in the commercial world, the vibrancy of the startup sector.

8:37All of these things have completely changed the dynamic where there's a huge amount of innovation to harvest that does not come from this sort of very planned structured approach. So while the kind of planned approach might still be correct for battleships, where we've got a plan this a decade in advance and it's a very systematic thing, there's a very broad swath of technologies that now can be done in a much more commercial fashion. In commercial here, meaning the way every other industry in the world buys, right? Companies invest their own capital at risk to do R &D and if they have a product that works and it meets the market need, they are successful and if they don't, they go out of business.

9:11It's a very simple and efficient system. And I think companies like SpaceX have dramatically transformed the perception of what are those technologies that are achievable in this new model. I don't think anyone would have believed rocket launch and the largest satellite operator by several orders of magnitude was possible on a commercial basis a handful of years ago. But just doing that shows that the space of things you can go after in this different approach is massive. The capital is there, the possibilities there, the risk taking is there. Now, where I see the gap for a lot of companies is understanding the war -fighting problems.

9:44These have gotten increasingly challenging and increasingly opaque in a lot of ways around the type of adversarial environment we're going to need to fight in, the ranges we need to operate, the complexity of the threat. These things are very, very hard if you don't live in this world to really understand how to make this work. And then the second part of this is sort of that acquisition, budgeting, contracting process and how does this play out. You really have to marry those things and really deeply understand what problems are going to have real war fighting impact and then how do I align that to limited resources, priorities and all those things to be able to go out and scale these technologies.

10:22So I think DIU has been an excellent bridge at kind of solving a lot of those letter problems. So how do we take these commercially modeled business models and apply it to war fighting problems, but try to bridge that gap with education, support, process to enable these companies to actually be successful. First thing I'd say is yes, yes, and really where do we need to go from here? Because if you think back to those days when a lot of us kind of grew up together helping make this happen, and I was a part -timer at the time while I was in my day job at Apple, there's been an enormous amount of progress in moving the ball forward, and back then there There was a relatively, you know, when I thought about, for example, friends who were saying, I'm going to go and start an investment focus in defense tech kind of at this intersection.

11:08It was a relatively small group of people, super pioneering. Now, there's a ton of capital, a ton of interest, and a lot of people who really want to be part of this. So we've really reached a tipping point on that. But we still have a long way to go on taking some of what Brian was just talking about really to the next level. Because while now, if capital is not really the scarce commodity, an interest isn't the scarce commodity, there are plenty of both founders and funders in this space, or toward this space, we have to be a much better counter -party for them. And that's really about clarifying the demand signal so that you don't have to have a PhD in how the Pentagon works or what exactly the need is in order to get there.

11:47And it's just about a couple of very simple things. First, it's about clarifying the demand signal, First, which problems matter? Because it's kind of hard to tell if you're out there, even if you're reasonably well educated, which things matter the most of the department? So it's, which things really matter? Second, it's about clarifying the demand signal, not necessarily in terms of a very specific requirement of this is exactly what I need to build, but it's, what's that problem that we're trying to solve? And how might we get after it? But then it's about, what is the path to scale or is there a path to scale?

12:17So that it's clear that if I make the investment, if I'm a founder or a funder up front, if I make this investment, that there is a pathway to scale there, because we need that scale from the department's perspective to deliver this strategic impact. Well, the companies need that scale in order to deliver the ROI. And so we've got to get a lot better at clarifying where the demand is and where there's a true pathway to scale so that those people out there who assess, price, take, and then manage risk for a living can do so on our stuff. And so that the success models and the reference cases are larger in number and deeper in impact so that that flywheel gets going in a truly American way.

12:59And that at its heart is what we've got to do here, which is that Brian was alluding to. This is about getting that way that is very simple, that normal businesses work all the time. That's built what, for example, Silicon Valley is into this incredible engine for the world. Get that online in a way that works for delivering this kind of capability. Yeah, we'll go back to something Brian said about how there's large exquisite systems, battleships, things that can take 10 years to build. And then startups are really good at that sort of 12 to 18 month cycle, moving really quickly from prototype to production.

13:30Where do you see D .I .U. Shepardy companies to say these are the types of technologies we expect startups to build. The Primes can do X, Y, and Z, but we want startups doing this. And how does that relate to something like Replicator, where we really is a focus on startups? So I actually don't think it's about a Prime's versus Startup thing. I think it's probably not the right split. I think there's every reason in the world that a Lockheed could show up with counter drone systems or kind of autonomous technologies that D .I. is really looking at and be able to be very successful. I think the dissonance often is in the business model piece where I think a lot of the Startups have embraced the business model of I can prove myself through delivery and I'm willing to take on the risk that is very different than the incentive structure.

14:15We, as a country, have put on to the primes and they have built a route, right, which is very much the government is going to absorb the risk and kind of steer where they want to go. And maybe explain cost plus to those who don't know what it is. So cost plus fixed fee contracts are how a huge percentage of the DOD procure systems. And the basic structure of this is the government will reimburse the contractor for all incurred costs, so bearing all the risk and pay the contractor a fixed percentage margin profit. So this includes not just what you spent on the thing, but all of your costs for production facilities and overhead and HR and all these things.

14:52So what's the natural incentive of this, right? It's to drive you to look at how much labor am I putting on to these projects, and it drives you towards a model of often taking a lot of risk on behalf of the government. And the government values this at time and encourages it, where we have these longer and longer duration programs. And so the natural incentive is to shoehorn as much technology as possible into these generational platforms, which makes them more expensive, more risky, and more challenging. And at the top of that, when you get into production, there's no incentive to reduce your cost, because you literally will make less money if you reduce your cost.

15:26They're going to buy a fixed quantity. So it's a very challenging incentive structure where it can work for high -risk programs, where there genuinely is the government saying, hey, I do want to take the risk. But once you get into a lot of these technologies where the cost to actually build them isn't so prohibitive that private capital can't support it. Right? That is a very different world that we're living in now. And so I think the class of technologies that D .I .U has been looking at is how do I take lower costs, more autonomous systems, and make them deploy at scale? That is something that does not cost you a billion dollars to invest in, right?

16:01We're talking tens to hundreds of millions of dollars of private capital to make these things go. And even a large class of modern weapons fit into that same category in my opinion. I think there's a much broader array of these technologies where if you had a framework where every three to five years I'm going to buy the best that exists. So we know a market exists. We're getting clear, consistent demand signal. And there's value to continue to invest in the space. And also I know if I don't invest, I'm going to lose next time because someone else stole my lunch, that is a hugely incentivizing way to get the best out of the industrial base.

16:33So I think the incentives on this case really do matter. And so we focus a lot on trying to drive the government. And it's very natural with the DAU, but is less comfortable in other places towards these more firm fixed price where we want to take the risk. And it's notable that actually a lot of the primes right now are saying they will not do firm fixed price contracts anymore because they've often taken a significant loss on most of the programs that they're doing with So it is a hard model to shift to if you're not used to it, but if you built your company from the beginning to be good at risk taking and operating this model, you can operate this way.

17:06It is possible as how every other company in the world works. I just add to something you said about these exquisite systems that the PrimeZip built, what's really interesting to me is how the defense market is transforming and moving away from exquisite capital assets like aircraft carriers, like ships, like very, very expensive of fighter jets or very, very expensive drones. And they're moving away from it because there's wide recognition that these are some of our most vulnerable assets. And that on day one or week one or month one of any conflict that they would be targeted and eliminated very, very quickly.

17:42And so when you talk about two, three, four years to build an aircraft carrier in $2 billion and it being destroyed on week one of any hot conflict with an adversary, you say, is that really the force that we want to build? And I'm not saying you're getting rid of every single exquisite asset that we have in our inventory, but what's happening in the market is this hybrid force architecture where you're getting lots and lots of smaller, more affordable autonomous systems, and you can pair them with larger exquisite assets when required, but that is what's really opened up the opportunity for a lot of companies to make a material impact without as much capital as required to build in our craft carrier a nuclear submarine, the F -35 you name it.

18:27And so that's something that I think is really interesting in the marketplace and why you see more and more companies able to solve the problems of the defense department without requiring tens of billions of dollars of CapEx to do so. Just to build on that, first of all, I agree very much with Brian that prime versus startup isn't for me the right way to think about this. It's innovative approach and innovative technology versus not. And so, D .I .U. does about 15 % of its work with primes and works with startups and works with everybody in between. For me, it's a little bit more about making sure that we're harnessing the very front edge of the very best technology, some of which may be developed with defense in mind in that last mile, but much of which is reliant on, and Brian referred to this earlier, on the incredible innovation that's happening in order to meet the demands of the commercial universe, right?

19:22The billions and billions of consumers out there and the enterprises that serve them, innovation in those spaces is driving relentlessly forward speed in 11 of the four tiers, we track the department, things like artificial intelligence, biotechnology, autonomy, energy, the pace of change in those spaces is, and frankly, likely always will be faster in some of those spaces, then it is an purely bespoke defense oriented way. And so we've got to be able to harness both simultaneously. We've got to make it easier to do that so that the investment, for example, in leveraging technology that might be, that's why I call it commercially derived, might have been in some ways commercially derived, and put that extra bit that we need on it, for example, for hardening and challenging EW atmosphere in order to make it be able to serve that critical defense need.

20:10Yeah. So a lot of the conversation we're having right now seems to be very much the impetus for replicator. A lot of our listeners have probably heard a lot about it but might not understand the details or I mean it would be great Just to understand what is replicator and what are the measurements for success? What do you expect to see from the program? So replicator is a critical example of how DiU3 .0 is being put into practice and more importantly of the broader Initiative across the department at the sector in the deputy or leading for how we bring more transformative capability and much more quickly.

20:41And in a nutshell, what replicators about is about putting real capability, critical capability that solves a critical operational need, again, that's necessary for us to ensure we can determine your conflict or win a force to fight. In this case, this is a charitable autonomous capability in order to counter capability from the people's Republic of China primarily, and to do it within 18 to 24 months, so not years and years, and put that capability out there. That's the first objective of replicator. The second one, which is as important, is to do that in a way that we break down systemic barriers to change, that we identify and knock down things that get in the way of making that happen across the department, and by doing so make it easier and easier and easier for us to do that again and again and again on other problems.

21:31And this is why, in some ways, we talk about replicator itself all the time. I actually think almost as important is the government structure for Replicator, which is the thing called the deputies innovation steering group, co -chaired by the deputy and the vice chairman directly. We at DIU are the principal staff to that, as well as sitting on the team, and then we chair the defense innovation working group, which is its next level, so basically the same group without the deputy and the vice. That's the government structure for Replicator. And the reason that's important is that is the entire department coming together on a table of leadership level.

22:04services, combatant commands, the joint staff, the office secretary of defense. We're all there at the table so that we can be driving this forward and knocking down the barriers. And today, speaking here at the conference, the deputy mentioned we are on track to deliver that capability within that time frame. And that's 18 to 24 months, something that ordinarily would take three or four years. Yeah. And I know when replicator was announced, Andrew put out a statement about how revolutionary it was as founders, what was it like hearing that there's this kind of program now, and especially since you were all founded way before something like this existed, how does it affect scaled startups?

22:39One of the things that I love about replicators that it's rooted in first principles as it relates to warfare, as it relates to the customer's problem. And at its core, it's about getting after this concept of mass. And mass has been a fundamental war fighting concept as old as warfare itself, right? If you have more fighters, if you have more equipment, you tend to do better on the battlefield, you tend to do better in conflict. So it's rooted in first principles that it's core. What I like about it is it's a push from the top of the department for everybody to start thinking about it. And I was in Abu Dhabi last week, and I met with the Aft -Sent Commanding General and one of his quotes, which I love, he's like, look, I'm not interested in 1 ,000 drones.

23:20I'm not interested in 10 ,000 drones. He's like, we need to be producing hundreds of thousands and up to a million drones and munitions to be prepared to deter and if necessary when the next conflict. And so what I really like about that, again, it reinforces what the deputy is talking about, it reinforces the replicator program, it's rooted in first principles, and I think that's fantastic. To give you a sense, right, in terms of one of the challenges when we talk about scale, today when there's a disconnect between the program executive offices and the combatant commanders, the war fighting commanders, et cetera.

23:55We're on a program of record. If you asked this was brought up, they asked the commander, how many of these aircraft that shield AI produces? Do you think you're getting? He said, I hope we're getting a thousand. They're getting 60, right? Via that program, but that's the disconnect between what is actually required to deter and win conflict versus a program manager who might not necessarily want to raise their hand and say, I think we should build a thousand or 10 ,000 of these. Which is a challenge itself. So I appreciate the top cover, top line message of replicator and that we're moving towards this hybrid architecture.

24:28We're going to have thousands, tens of thousands, hundreds of thousands of aircraft, but we'll also have our exquisite aircraft for the missions that require those capabilities. Yeah, and I think the shift in the way that innovation has been talked about in the department since we started, I, go every year to the Reagan defense forum, and you kind of see how the tone changes every year. And the tone for a long time was about, here is the sheer volume of small businesses and innovative companies that we are dealing with. Here is the number of prototypes that we have done. And it was completely missing the point, right?

25:05The point was not, can we come up with technology? The answer is, obviously, yes, right? Like it is very clear the technology can be invented and we have a very vibrant base that can do that. The shift in messaging and focus towards how do we actually get this as deployed at scale or firefighting capabilities is the key one. And I completely agree that the focus from the top is necessary to drive change. There's two things that drives change in any organization and especially in the DOD. It is an urgent conflict and it is top -down leadership. And those are the two things that get changed to happen.