School of War - Ep 165: Shyam Sankar on a Defense Reformation

Episode Date: December 17, 2024

Shyam Sankar, Chief Technology Officer and Executive Vice President of Palantir Technologies, joins the show to explain the broken Defense Department acquisition process and how he believes it can be fixed. ▪️ Times      •      01:24 Introduction     •      01:39 Employee #13      •      03:14 Palantir      •      06:22 Monopsony       •      11:18 Messy and chaotic      •      14:40 Dual purpose companies      •      17:18 The buying process      •      23:50 Pushback     •      25:59 Competing efforts     •      27:37 Heretics and heroes      •      31:22 Thinking about future war       •      35:05 A changing selection criteria      •      36:47 “The future is software defined”   Follow along on Instagram or YouTube @SchoolofWarPodcast Find a transcript of today’s episode on our School of War Substack

At the end of the Cold War, the Department of Defense famously hosted a, quote, Last Supper for defense contractors in which these contractors were told that there would be lean years ahead and that it was time to consolidate. Well, how has that worked out and how do we get our warfighting tools back into shape for the fights to come? I have one of America's sharpest thinkers and leaders on these and other problems with me today. It is a perspective for war This Iraqi invasion of Late December 7th,

1941, a date which will live in history. A bloody experience of Vietnam is to end in a state We continue to face the grave situation in grand

We shall fight on the beaches, we shall fight on the landing grounds, we shall fight in the fields and in the streets We shall never surrender. For more, follow School of War on YouTube, Instagram, Substack, and Twitter. And feel free to follow me on Twitter

at Aaron B. McLean. Hi, I'm Aaron McLean. Thanks for joining School of War. I'm delighted to welcome to the show today. Sean Sanker. He is the chief technology officer and executive vice president of Palantir Technologies. He's an author and commentator on defense and defense tech issues. Sean, thank you so much for joining the show. Thanks for having me, Aaron. So if my information is correct, you are employee number 13 at Palantir. Tell me how you came on to Palantir's radar, how Palantir came onto your radar, and how your two stories got intertangled. Yeah, that's right. I was lucky number 13. I joined about 19 years ago. And really, you know, in late 2005, I heard about a group of a handful of people who wanted to work on issues of national security.

And, you know, YouTube was a fledgling startup still independent. And you could look around and you can say, what am I going to spend my life on in the valley? Like, I'm a technologist by trade, and I believe in the power of technology, but I didn't want to build another web calendar app that somehow made the world a worse place. And really, this, you know, once I met these folks, once I saw what they were working on, I would much rather have sailed working on something of national import than have succeeded on working on exactly the wrong things. And I think that was, that was something that was deeply set into. I was a college student during 9-11. I tried to drop out and join CIA. I think they had many people trying to.

to do that. And, you know, so that didn't quite work out. But even before that, my family came to this country as refugees fleeing violence from Nigeria. My parents almost lost their lives to an armed robbery. And dad raised us with a deep understanding of the counterfactual, but for the grace of this nation, you'd be dead into ditch in Lagos. And so finding some opportunity to give back was always, always something important to me. For the lay listener who probably aware of Palantir and that it does stuff involving computers and defense. By the way, this would have characterized me, say, 10 years ago, like Marine Infantry Officer Aaron. Maybe you've seen clips of Alex Karp delivering mic-dropping moments at various conventions and events. But you're not really sure

what the heck this company does. For that kind of audience, what does Palantir do? Well, today, our business, you know, 50% of what we do is commercial. We build every Chrysler car, every Airbus Airframe, every HD Hyundai ship. And the other 50% of what we do is we work with the U.S. and her allies on issues of freedom. I think about it as freedom and prosperity, the two sides of our business. And they're actually more related in deeper ways that I think we can unpack as we go through the conversation here. But it starts with this fundamental premise of, I don't know, you look around the world today, you could say we have a legitimation crisis. Like, why do doors fall off planes? Why do we not believe fundamentally in the institutions, public or private and their competency?

And that's very different than the world we had coming out of World War II. Our fundamental for this is that a big part of the problem is that the C-suite, our leaders, have been given a steering wheel that is actually a prop from the jungle cruise ride in Disneyland. It's not actually connected to anything, and they're diligently trying to steer us in certain directions, but actually there's a fundamental and deep failure that the factory floor is not connected to the boardroom. These mark-to-market moments happen during crisis. You recognize people spent tens of billions of dollars on their supply chains in the lead up to COVID-19, and they all fell over like paper tigers within two weeks. So you listen to the CEOs in their in their earnings calls during

COVID. They would talk about how technology had saved them and what were they talking about? Zoom and Teams. What a damning indictment of the software industrial complex that it couldn't produce anything that actually met its moment in this crisis. And so we wanted to build a company that imagined this stuff backwards. Instead of thinking about how do I build a company principally based on what, you know, how sellable is my software. I need to define it in a very tight box. I need to really think about what do the buyers want, how about we build software that actually matters? Let's work backwards from the problems they exist, which means you're not going to think of the great ideas eating strawberries in Palo Alto. You've got to be on the fire cells of Djibouti and the

factory floors of Detroit empirically understanding what does not work. Why does it not work? What does that mean about the software that needs to exist and go build that? And really, that was my I joined as a 13 employee to build what we later called the forward deployed engineering team, heterodox methodology of doing product development in the field. You're not just a commentator on these issues, but as listeners can probably tell from your answer just there, you're an impassioned commentator on this issue or a set of issues. A lot of your work and the work of some of your collaborators can be founded at firstbreakfast.com,

and we'll get into why it's name that here in a second, or maybe you just will directly as a response to this question. When you look at the defense sector, When you look at the Pentagon and the way the Pentagon buys stuff for America's soldiers, sailors, airmen, Marines, et cetera, guardians now to use. Big picture. What has gone so wrong that we are at the kind of situation where, as you very vividly put it, it's like the Disney ride? For me, it's the Millennium Falcon, which I recently did, which is a lot of fun, but I'm pretty sure as much as it seemed like I was steering the Millennium Falcon, I probably wasn't. I suspect there were actually multiple cockpits in that complex, even though it looked like I was specific.

specifically selected to fly the Millennium Falcon? I think, well, it goes a long way back. So I think one of the hard things about the department's role in trying to buy and acquire these capabilities is that it is a monopsonist. It is the sole buyer for the vast majority of things it wants to buy. There's not really a market for aircraft carriers. There's not really a market for many of the weapon systems we want to buy. And when you're a monopsinist, it's really hard.

Everything is your responsibility. You know, you don't get to benefit from Steve Jobs thinking of the iPhone. You have to think of the iPhone. And that is a ungodly burden. And I think we understand in this country, we believe at some fundamental level in free markets, or you don't. And so how are you going to structure the sort of chaotic, messy innovation that often seems wasteful and duplicative? How are you going to get that? Which is what we believe. It's like empirically, our wealth, our prosperity has been powered by this. And I think the failure of the Soviet Union was powered by their central planning. But if you really look at the Department of Defense, it does use central planning. It has a fight

it. It has a five-year process for the end. It's this idea that actually top-down, we're going to be able to allocate resources, make these decisions, we're going to push towards, it's a human aesthetic. Like, you know, the Soviets aren't unique. I think there's a whole cadre of humanity that prefers the unitary effort aesthetic. Like, we should just have one effort, put our weight behind it and do that. That would all be great if it was actually noble, what you were even going to build. But when we were at our best as a country, you know, when we were building ICBMs, we had every service competing. The Minuteman wasn't a birthright for the Air Force. All the services put in their hat in the ring. They were fighting each other to see who could make the best ICBM.

When Admiral Raybourne was building the submarine launched ballistic missile, he himself had four concurrent competing programs underneath them. The Polaris was not the birthright winner. But look at today. You know, the F-35 was always going to be the fifth-generation fighter. We didn't think maybe we need a 35, 36, 37. Maybe we should go to the legacy platforms and say, you know what? you're not legacy. Here's a little bit of money so you can get me to cancel these new programs. And how do we start to approximate the forces that we need? Competition. You know, we look at the Last Supper and the sort of conventional critique of the Last Supper, this dinner in 1993 in the Pentagon where we said, we need a piece dividend. We can't spend as much as we have been spending.

We're going to cut it by 67%. Not all of you, 51 defense primes are going to survive. You know, now we have five. We look at the great consequence of that conventionally as we lost competition in the industrial base. I don't think that's right. Sure, let's have more competition in industrial base. What we really lost is, you know, this consolidation, it bred conformity. It drove out the crazy founders, the innovative engineers. It was the beginning of the financialization of defense. And as a consequence, you just don't have as much crazy for lack of, which I think defines the U.S. the vibrant U.S. economy. A stat I'll give you, which I always loved, Mario Draghi just issued a staving critique of European competitiveness. He was the former European Central Bank

President and Prime Minister of Italy. In the last 50 years, Europe has created zero companies from scratch worth 100 billion euro or more. In the last 50 years, America has created all of her $1 trillion euro companies from scratch. So the performance of these two economies, you know, it couldn't be any more different. And then when you look at, you know, European capitalism is essentially categorized by the same actors who have been around for hundreds of years, you know, with essentially very comfortable positions. And if you look at American capitalism, it's very dynamic. It's often founder-driven.

You know, you have big personalities. And when I look at the industrial base when it was working, today we think about it as North Brumman. But it was Jack Northrop and Leroy Grumman. You know, it wasn't Lockheed Martin. It was Glenn Martin. And we would look at that today and say Elon and SpaceX, that makes perfect sense. Elon has delivered a thousand X price performance on launch. You know, and I think this is, you know, maybe capitalism doesn't work that well.

Only founder-driven capitalism works that well. And of all the nations in the world to appreciate that, we understand founders are special. There's a reason we call them the founding fathers. Yeah. There's a lot in there, and I want to kind of take it piece by piece. You have this great essay, Screed, Palaimic, the 18 Theses, which can be found online. This is a recent production. The whole thing is read to read, but I have in front of me,

this graphic that you have that depicts the post Last Supper consolidation. It's super compelling, and it drives its point home, and I'll just read one sub-list. So you've got GD Fort Worth, Sanders Associates, GEROSP business, Martin Marietta, Gould, GD Space Systems, Honeywell, Fairchild, Weston Systems, Laurel, Goodyear Aerospace, BDM International Libroscope, LTV, IBM Federal Systems, Unix, ComSat. I think I'm actually missing one. But anyway, point is, all of those, and I think one more, are today,

just locking. They're just locking. one, one enormous corporation that your case is, suffers from these maladies. You started making the case for competition there. Let's extend that out. Actually, let's flip it. Let's flip it to the competition amongst buyers, breaking the monopsony, where you have, in your revision, different services competing. I've read you, maybe it's in this essay on the 18th Eces, calling for combatant commanders,

potentially, or combatant commands to be able to compete and be purchasers, essentially. If you look at this from the sort of semi-religious perspective of jointness, where the point is to reduce inefficiency, to simplify, to bring into better coordination, things that would otherwise be un-coordinated, you could make a case that what you're saying is insane. You're saying, I should have multiple services, which from this point of view maybe don't even really need to exist at all. Like, maybe we should be moving to some sort of unified bureaucratic structure. These services are sort of historical artifacts. you're telling me that you want to empower them to run around and all buy the same thing from different people? Like how wasteful to the taxpayer is that? Tell me why that sort of obvious response to your point is wrong. Yeah, it's dialectical. So efficiency only matters once you've achieved effectiveness.

And you have to understand that the process of going from zero to one, of creating something of driving innovation, is very different than the process of going from one to end. So when McNamara brought his practices in the 60s to the Department of Defense from Ford, you know, here's how you managed. Well, Ford had unlimited demand. Literally every car they made, there was a buyer for. The sorts of techniques you come up with drive towards efficiency there. Efficiency is actually the only thing that matter for Ford where you had unlimited demand. But if you found, if we counterfactually imagined a different market scenario where actually you had to think about how do I build a new car that someone's going to even want to buy, how do I get a Tesla

at a market, how do I create a new category? That methodology will kill you because the reality of innovation is that it is messy and chaotic. In any attempt to get from zero to one through process will fail. That is part of our legitimacy crisis with our large, mature institutions. They can't reinvent themselves. They've been praying so long at the altar of process that they've forgotten what innovation actually looks like. Yeah. You know, this is the thought I'm having for the first time inspired by your commentary here today.

it's never occurred to me in quite this way before. I'm curious to know if you agree, because you're sort of pushing on an open door with me and this critique, and you have a lot of friends in Washington, just not enough at present to actually push the thing over. But the core problem is that we have this centrally planned monopsony, as you put it, that does what it does okay

and does a lot of things very, very badly. The core problem is that at the same time, our entire defense strategy, since at least the Second World War, has been premised on a qualitative edge, a technological advantage, really a series of technological advantages, constantly staying ahead, if anything, lapping our competitors such that we can get away with economies and other spaces. So, you know, first nukes and precision and reconnaissance and so forth, and now we're all looking for the next thing. So we're trying to combine a defense

strategy that is deeply, deeply, it requires technological advantage to succeed. Otherwise, we've got to rethink how we're doing things in a way that I don't think we're comfortable doing with an acquisition process that in your account is basically designed to make real innovation, or at least innovation at the pace you need, impossible. Is that kind of roughly where you are? I think that's, yeah, that's roughly correct. I think one of the problems here, you know, to keep with the religious theme is the great schism that happened. So if we look at the structure of our economy, World War II, early Cold War, at the fall of the Berlin Wall, only 6% of spending for major weapons systems went to defense specialists, what we would consider today as traditionals. The 94% went to what I call dual purpose companies.

I think too much is said about dual use, not enough is said about dual purpose. Chrysler used to build cars and missiles. Ford built satellites until 1990. General Mills had a mechanics division because they had to build machinery to process cereal. That Mechanics Division made torpedoes in artillery. When Pontiac took over the production of the Orlikon 20-millimeter anti-aircraft gun during World War II from the Army's organic industrial base, they reduced production time from three and a half hours per unit to 15 minutes. Now, they did that not because they were smarter humans.

They weren't. I bet the same talented people on both sides there, because they could leverage everything they learned about mass-producing cars to mass-produce weapons. And so this schism has slowly kind of, it's led us to put our defense industrial base on the Galapagos Islands. You know, exquisite exotic creatures that really can exist nowhere else. The things we build are quite special. But it comes at quite a cost that you're not going to develop produce a lot of them. You're not thinking about, you're not able to leverage any of the tangential commercial R&D knowledge and experience to make the defense industrial base better. At some point in time, every camera,

car and cereal box Americans bought was subsidizing national security and the lethality of our service members. And unfortunately, China has really learned from this playbook. If you look at Chinese primes today, only 27% of their revenue comes from the PLA. The rest comes from us as Patsy's buying cheap crap on Amazon. Now, we are subsidizing lethality against U.S. service members. Talk me through how it works if you want to buy something at the Pentagon. Let's pick a simple example, and I'll fictionalize it a bit because I know there are actually recent real handgun examples. Let's say you want to buy a new handgun. The old M9, Beretta, it seems a little clunky. It seems like we deserve something better. You know, it's the 21st century. So, you know, today is day one,

year one. You're somebody with authority and you've decided we're going to start the process of buying a new handgun. What happens next? A reasonable person who has no idea how this works might say, well, gosh, there's lots of people out there who make handguns. You go pick the one you like or, you know, maybe ask for something fancy and see if it comes your way and just buy the thing. I take it it's not quite that straightforward. Yeah, we've come up with a Byzantine structure, really, you know, everything that has gone wrong, we've tried to create a new rule for, and we have, of course, a duty to make sure we're spending our money well. So this process leads us to something that looks like, first, let's have a validated requirement. Let's write down

everything that it is that we think we need. And that structurally tends to lead to bloat. You know, First of all, there's just the fundamental uncertainty of do you even know you need this or not? And are all these requirements equally valuable? But okay, so we're going through the process. We're writing up a requirements document. It has to be validated. This itself could be a multi-year process. You know, if you go really fast, maybe it's half a year, but typically it's multi-year.

Then you're going to get that requirement is turned over from the warfighters to the acquisition community. They're going to develop an acquisition strategy. They're going to go figure out the resources. they need to actually buy this and fund the program and try to model what it'll take over time, then you're going to start executing your strategy. This has led to a world where it takes two decades to field new weapon systems. And it's all very slow. And I think what we're starting to realize is it's also maybe wrong. You know, obviously the world has changed tremendously in two decades by the time you figured this out. And because it takes so long, you know, you go through this a few cycles.

What incentive are you creating for the warfighter? Well, man, this is my one shot at riding this requirements document. I got to get every bell and whistle in there possible. These things are so slow that I got to make it super special. And it's actually, it leads to a pathology that increases a likelihood of failure. And if you think about the software world, obviously atoms are harder than software, but if you know the software world, the companies that do the best are the ones that ship changes the most frequently. So you actually want to be buying continuously in small tranches that allow you to learn what you need, adjust it, and create incentives for both comprehensive. petition in the industrial base and the ability for the warfighter refine their understanding,

you know, I don't actually need this. Or how do I engage in the tradeoffs of if I let go of this requirement, I could have twice as much at half the price. That doesn't exist today. It's just such a serialized process that you're unable to manage the tradeoffs along the way. Now, let's go to a hardware example, SpaceX. They co-locate their R&D engineers on the factory floor. These are two different functions, designing the product, making the product. But the reason they do that is there is obviously a feedback loop. And if you're going to make a ton of these starlings, or you're going to make a ton of these rockets, that feedback loop means the next increment can be better and better and better. You want to get into that process of continuous improvement that today's system

really deprives us of. It's one of my big reflections in Ukraine. So the obvious three lessons we could be taking from Ukraine. The first is you can't expend 10 years of production in 10 weeks and think you got you got it right. You know, I think it shows us that we fundamentally confused the stockpile as the deterrent. It was always the ability to produce the stockpile. So years of trying to get things down to the minimum lines and affordability and shutting production down because you're like, look what I have in the stockpile, that was a mistake, a huge mistake. That would then lead to a kind of sense of like, well, do I need to design these weapons so they have a 30-year shelf life? If I recognize that the deterrent is my ability

to produce them, what if they had a three-year? or five-year shelf life, and I just constantly use them because I'm going to constantly be making them. That would lead to cheaper munitions. So the second one, the second major conclusion I have from reflecting on Ukraine is drone obsolescence life cycle is like two to six weeks. So you make a new drone, it's going to be jammed. So the only question is not what does your weapon do today? It's how adaptable is your weapon. What is your ability to change how it behaves and functions? So you could have the most exquisite thing in the world and it'll be irrelevant in a few months. And that seems highly problematic as we start thinking about the conflicts that we're going to be in. How much of our

exercises and practice is not about what does the weapon do today, but can our uniform service members, can our industrial base upgrade our weapon to meet a changing environment and condition? And the third one is a real argument for new platforms, why you need more competition. People love saying that the Ukrainians sunk half the Russian Black Sea fleet, even though they don't have a Navy, you know, despite not having a Navy. And I think, no, no, you got it wrong. because they didn't have a Navy that they sunk half the Black Sea fleet. And so, you know, how do we motivate ourselves to experiment with new force employment concepts with entirely new systems where we're not so wedded to the legacy that we can't actually go experiment? And that's

where I think you need competition. Because it's very hard to ask someone to do both things simultaneously. You know, you need to leverage the fundamental human incentives. And really this idea, my lived experience with this, I was in a major competition on a platform with a big It was a team effort. The two of us were going at it. We had one program manager, and we had all these soldier touch points. In the midst of this competition, we had the chat GPT moment. Wow, look at this. This miracle of technology. It's come out. It's obviously not in any requirements document. No one's conceived of it before. But I'm thinking to myself, man, when my son's in the back of this thing, I want him chatting with a chat bot about how to maintain this asset, not flipping through a 10,000 page PDF manual. And I offered the PM a free upgrade, no research. source, no risk to schedule, that would deliver this chat bot. There was no interest because there was no upside for that. So it's just a clear, and they're doing their job. They're not actually, they are following the incentives that we are giving them, right? So it's not a critique of this human. It's a critique of the system. They do not wake up thinking my job is to deter she. It's my job

is to deliver this program with the least amount of risk, and I don't see how this does anything here. You know, even though it's more capability, it's, you know, maybe the same risk or why do this? But I tell you a counterfactual. If you had two program managers, one that was assigned to us, one that was assigned to our competition, they would have said yes. They would have said yes, let's go murder that other team at the next shoulder touch point. We're going to show with something better, something that's more awesome. And I think that there's a fundamental reality in that, that we as Americans will wake up to murder the bureaucrat one corridor down every single day. And that is going to force us to get better every single day. And I say that as a proxy for

the U.S. commercial market. Why are we so innovative? That's what our companies are doing to each other every single day. When you sit there with Roger Wicker, who's been there better here on the show, or, you know, Lloyd Austin or their staffs, and you run this idea of why don't we have multiple program managers by them? What do they say? What's their response? What's the response from the establishment as it is? Well, so I'd say that, you know, there's probably maybe slightly less than half, but it's close to half of people who think this is a great idea. And, but I'd let's steal man the pushback. The pushback is often, That sounds duplicative.

We can't afford that. It sounds like it's going to require a lot more resource. And I think this comes down to your fundamental view of technology. Is technology something that is deflationary? Is it going to make you better, faster, cheaper, or not? And I think implicitly the folks who are coming with the pushback, their lived experience, because we've had this exotic Galapagos Island industrial base, is that costs rise faster than inflation.

I can't really imagine a world where competition brings down prices in a meaningful way more than just a few percentile here or there, but transformational. And that's where I like to point to SpaceX. You know, when I was a kid watching shuttle launches, that was $50,000 a kilo to orbit. You know, with Starship Heavy reuse, it's going to be $10 to $20.

I'm not missing any zeros there. So we're talking about a thousand-x price performance improvement in roughly two decades. That's the power of transit. I just had a senior government person telling me, you know, the problem is now SpaceX has a monopoly on launch. I'm not sure it's a literal monopoly. I think they're so damn good at it that for that price,

it's very hard for other people to compete. But let's not forget that they earned it by dropping the price by a factor of a thousand. Think about what that implication is for our national security as a consequence. Like, this is actually the market working. This is what we want to see. And sure, let's encourage more competition in the next generation. Why stop here? Let's get a factor of 10,000 and keep going.

But we have to recognize how much we're getting from the commercial innovation that is possible. You made reference a few minutes ago to the Chinese defense sector and made the point that perhaps ironically, given that they're literally a communist country. They embody more of the old pre-consolidation American system with companies that have multiple purposes, et cetera. Say more, if you will, I don't know how much time you spend thinking about our counterparts there across the Pacific, but about how they go about these things, given that there quite literally is a Chinese Communist Party, there is central planning. How does their acquisition system work? How does their acquisition system work, just at a little bit greater length than you alluded to before.

So I'm not the world's expert on it, but I can tell you what I'm observing and what I see. So my clip about this is everyone has given up on communism, including the Chinese and the Russians, except for Cuba and the DoD. Yeah. And so if you look at even the Russians, they basically had two ministries of defense, right?

They had the Wagner group and the MOD. That's competing programs right there. And so my push on that is not that you never eventually consolidate them, as they clearly did, but that you need to be able to engage That's one way to put it. Yeah, they consolidated them. Sorry. You need to be able to engage in periods of strategic competition that drive innovation. And so what I see with the Chinese is they know the weaknesses of their central planning,

and they're planning around that. So they will have multiple primes competing against each other. They are essentially having multiple competing efforts as well. It's part of their system and how they're going. And it's what I think we could be doing as a monopsony, you know, where there is going to be an aspect of central planning. The question is, do you lean into it or do you lean away from it and try to approximate market forces wherever possible? Yeah. You have this interesting section of your commentary online at firstbreakfast.com where you talk about heroic figures in the defense sector over the years. Some of these names will be familiar. You know, we've talked about John Boyd on the show a couple of times.

Hyman Rickover is a pretty well-known name. Then there are names that I will confess are pretty new to me. Abraham Karam was somebody whose story I did not know while I was preparing for the episode, Pierre Spray. Pick one or two. I mean, what are you trying to illustrate through this effort? What do you think is important about people like this? And then pick one or two and tell us why they're models for how you think about what needs to be done. So the series is called Heretics and Heroes, and the Heretic Coming First is very intentional. It's because, you know, I'm trying to illustrate that we're going to create these capabilities because of great Americans, of great, unique,

singularly talented people that we empower to do that messy, chaotic process of innovation. Admiral Rayburn, I'll tell you one story about this, which I think is very funny. So Admiral Rayburn understood this. He was building the submarine-launched ballistic missiles. Congress was breathing down his neck. The Navy was breathing down his neck. They wanted, you know, the perfect program plan. They wanted the Gant chart of Gant charts. And so he hired contractors to build P-E-R-T, which many people have heard of. It's, you know, he used the latest in decision sciences, Monte Carlo simulation, to give you the bounds of how long things would take. The reality is it was all a very clever smokescreen.

He was basically throwing BS data in front of these lawmakers and policymakers so that he could create a shield of protection around the engineers who were doing the messy, chaotic, completely unplantable and unpredictable work, so they wouldn't be bothered. Unfortunately, his program went so well in the end that people thought, wow, Pert is fantastic. Look at the results it delivered, and it was institutionalized. Henceforth, all programs must use Perth. I think Rayburn was probably too embarrassed to admit that it was a smokescreen in the immediate

aftermath. So we institutionalized mediocrity as a consequence of this. Maybe we should praise him and blame him at the same time. But I think it highlights something special. We call it the Apollo program, but it would not exist without Gene Krantz. Kelly Johnson was a singularly talented aerospace engineer as a kind of, we would recognize him in the Valley today as a. thousand X, 100 X engineer. As a 22-year-old, he saw a 2D drawing of an aircraft, and he went to

his boss, his first job, and he said, this thing won't fly. And if you're so smart, why don't you fix it? And he did. And this guy built 41 airframes in his lifetime. He built the SR 71, the U2, which we still fly. You know, he was not a fan of the F-35 at the end of his career. He said, you know, it doesn't make sense to have a Swiss Army night. And you can contrast that to many of the programs we have today. They have no founder. They have no father. They're kind of like a test tube baby. made up through committee, a series of requirements. Hyman Rickover worked on nuclear reactor for 30 years. I don't think there's a counterfactual where you swap someone in and out of that place

every two or three years, and you would have achieved the goal. And there's a reason that naval reactors is an eight-year gig. You know, we even understand today that continuity and knowledge is really important to the leadership of these asymmetric capabilities that we have. You know, let's switch from business to strategy for a second. And I want to ask you just a really broad, question because you, you know, obviously you're, you're helping to steer this company now, but your background is as an engineer and as a software engineer, which is a world that I,

I can't begin to understand how the things actually function. But my really big picture question for you is when you think about the battlefield, let's say 10 years out, maybe 20 years out at most, I don't want to ask you how should we be picturing that battlefield. Maybe that's a little unfair though maybe you have some tentative answers. I think I want to ask you, how should we begin the process of thinking about that? Because presumably that's part of your responsibility at Palantir is thinking about how to position the sorts of things you'll be providing that have impact 10 to 20 years out. And you have this technical background. How is war changing? How does your particular specialty in software engineering? How is that increasingly relevant? And how does that

relevance work? Like, how do we even begin picturing these things? I think about the battlefield that I experienced some 15 years ago now and how just different things that I've seen in Israel are today are things I read about in Ukraine are today. And I just, I could use some help thinking about how to think about the battlefield 15 years from now. Well, I think maybe to answer this slightly orthogonally, you know, technology is a tool for our commanders. And our humans, we could say, but like, we could focus in on the commander for a second,

which doctrinally believe is the most important actor on the battlefield. And so you can think about this as like an Ironman suit for them, physical software, you know, all this stuff is a lever. And so I think first you have to start with this deep understanding, which I think is still heterodox, that technology, people think somehow technology will make humans less relevant. Or they think technology will raise the capability of the median human. And it will. But it will make the most important human substantially more important.

So small differences in the abilities of your humans are going to be magnified. And so then I think reorienting around this concept of this is an Iron Man suit for our commanders, How do we get into the business? As a technologist, I hold myself accountable to the first derivative. What is my rate of improvement? It's not the capability I'm delivering today. How responsive can I be to what the commander is asking for tomorrow? And that's borne out of the forward-deployed engineering methodology.

Like, where did it start? When we were in Balad or in Bagram, people would go out on a mission. They'd come back. They'd have feedback. They'd say, this, this, this, and this. And while they were sleeping, we were coding. And they'd go out on the next mission with this, this, this, and they'd come back with more feedback. And so Boyd would recognize this is like, oh, this is the Oudaloup. You guys are just,

we're just tightening it up. And so I think in some way, the future battles field is not different. We're in character from the present one, which is whoever has the best Oudaloup is going to win. Now, how are we going to achieve that? And yes, will the platforms be different? Will the technology be different? Yes. But you're not going to think of those things in D.C., you know, doing just blue sky R&D projects. You know, there's a role for that. You need to to do that stuff, or where the rubber is going to meet the road is the folks who figure out to employ those things with the greatest amount of efficacy. And that employment, you know, when the world was moving slower, we could really silo

these things. Hey, there's a group of people over here doing research. When that's, like, let's have a linearized technology readiness life cycle thing. Let's, you know, TRL 1 to 9. And as it graduates, we'll move it through this process. The digestion of that is way too slow. And the assumptions of how long it takes to mature things is not right. And so we need to be co-locating, just like the SpaceX engineers are co-located on the factory floor.

We need to co-locate the people who produce our weapons and our technology with the people who employ them and create the sort of partnership between them that you're the pit crew, you know, understand our role. We're like, we're not telling you how to fight the war. Our job is actually understanding how you want to fight the war and give you that capability. You know, it's funny. You laying that out actually makes me think that one thing that's then critical to understand and act on is not a particularly cutting-edge insight at all, but if the role of technology is to make more powerful

and to give more powerful tools, and thus to make more powerful commanders, and as you sort of put it, like a smaller group of people, but there will still be people at the center of it, then the education of those people and the training of those people and the preparation of them to operate in a world where all the traditional realities of war still apply, but in this fast-shifting technological environment,

Whereas you put it even today is just a few weeks in a cycle of drone technology on the battlefield in Ukraine. Identifying those people, educating them, training them, that's just even more important somehow. And I don't even know how you go about, for the technology piece in particular, I don't even know how you go about starting to think about that process of things for people who are on track to become, you know, senior commanders in the military. I'm not sure if it's in your purview to think about those things, too, but that's my immediate takeaway from you for you saying that. Well, structurally, I'm not, you know, I wouldn't say I'm the best at systematizing things, but I think it starts with having, you know, identification. You mentioned the word identification. I think that's really important because it starts with, you know, there is a kind of like an intrinsic capability that you're building a kernel around in training. And I'll even say, my own experience,

you know, there are commanders to look at technology and say, oh, that's where the nerds in the corner. And then their commanders who look at technology and grab it and pull it in very close to them. They'll rearrange their jock floor so that, you know, they essentially have the tailors of the Iron Man suit in the right positions to give them the capabilities they need to win. And that, you know, maybe that's generational. Maybe it's kind of add to non. I'm not quite sure. But I do think we should expect that technology will change our selection criteria for who are the right commanders going forward. And another thing, I want to be respectful for your time here, so only a couple more questions. But another thing that this inspires me to ask is, if the cycles are really this

fast, and if anything, we kind of expect them to get faster, and then we have to think about the problem of how do you have commanders who can handle that and think about that. You just can't produce new hardware on that cycle, at least in the world we live in today or in any reasonable world I can picture in 10 years. Like hardware, you know, new ships, even fancy small drone ships or whatever, like, it's just going to take some time to put that thing together. But the software and things you can plug in and out of the hardware, smaller things. Presumably, that can all, the software can be changed at the speed of which the humans and the software tools can change it.

And the smaller things that can be plugged in and plugged out, presumably can operate on a little bit of a faster turnaround than the larger platforms. So is this, like, is this future Battlefield one that is, like, quite modular and quite software-intensive? Yeah, I think the future is software-defined. Like, you know, you're not going to, you're not going to shoot bits at your adversary. At the end of the day, you know, we are talking about bending metal. But then what parts of the, you want the bending metal to be as unexquisite as possible,

because it can be done at mass, at scale, in a distributed way, and you want the software to be as exquisite as possible, because it's what's adaptable, and you can compound on that. And I think it also plays to a deep and profound American advantage, one that America actually tends to underestimate. You know, we are the best at software by a yawning gap. The second best country is Israel, and their second best by quite a distance, actually.

Think about the fact that there's zero Indian or Chinese enterprise software company, that are competitive on the world stage. So why is that? It's obviously not IQ, otherwise this would be more evenly distributed, maybe pro rata to population or something. It's culture. And culture is the hardest thing to replicate.

And that culture, people look at the valley and they sometimes think maybe we imported it from Israel or India. We imported it from Iowa. It came from Bob Noyce, the co-inventor of the transistor, the co-founder of Intel, the semiconductor company.

And it's deeply rooted in Midwestern values. It's a willingness to play positive some game. Bob Noyes is the, he coined the term open door policy. We don't recognize the degree to which actually all Silicon Valley culture is descendant from Nois. And that is actually the hardest thing to replicate. You know, if you're the Singaporeans and you come to Stanford or come to Silicon Valley, you're like, maybe I need a research university. Maybe I need venture capital like Sand Hill Road. These are all symptoms.

Yes, they're enabling. You need those things. But the one thing you actually need is the one thing you're not going to want to copy. You like your culture. You don't want to be Midwestern. And that's why it's so profoundly sticky for us. Now, we shouldn't rest on our laurels on it, and I think one of the ways it allows us to build things of immense complexity. Like, if you look at the Israelis, culturally, they can build can't build aircraft carriers.

And that's a culture I have an immense respect for. So I don't mean that as taking a pot shot. But we can build these things at tremendous scale and complexity in this country. It's like 87% of all tech market cap in the world is American. That's not by coincidence. Shom Sankar, chief technology officer and executive vice president of Palantir. you can find his commentary at firstbreakfast.com, amongst other places. This has been a truly thought-provoking conversation. I have learned a lot, and I think I'll learn a lot just as I process

things that you have said, and I really appreciate you making the time. Thank you, Aaron. This is a nebulous media production. Find us wherever you get your podcasts.