Anduril Industries is an American defense technology startup
founded in 2017 by Palmer Luckey (the creator of Oculus VR) and former
executives from Palantir. Named after the legendary sword from The Lord of the
Rings, the company builds autonomous drones, air vehicles, and missile defense
software designed to take humans out of the direct line of fire.
The company isn't just producing hardware; it's redefining
the modern battlefield. Its technology ecosystem acts like a real-life defense
network:
Anduril has rapidly grown into a $61 billion behemoth,
backed by heavyweights like Andreessen Horowitz. Moving away from the slow,
costly production of traditional defense contractors, they operate massive
hyperscale facilities, like their Arsenal-1 mega-factory in Ohio, to
mass-produce intelligent, software-defined weapons at consumer-electronics
speeds.
“Iran has shown us that war as we’ve known it is over.
The future of high-tech warfare has arrived. Just look to
the conflict in Iran to see how much drones and robots have remade the modern
battlefield. Is the U.S. positioned to win wars in this new era? What are the
ethical constraints of waging autonomous warfare?
My guest
this week is Christian Brose, the president and chief strategy officer of
Anduril, a defense technology company building a slate of autonomous weapons
and defense systems for the American military.
Below is an edited transcript of an episode of “Interesting
Times.” We recommend listening to it in its original form for the full effect.
You can do so using the player above or on the NYTimes app, Apple, Spotify,
Amazon Music, YouTube, iHeartRadio or wherever you get your podcasts.
Ross Douthat: Chris Brose, welcome to “Interesting Times.”
Chris Brose: Thank you. Great to be here.
Douthat: So
it seems to me like the future of high-tech warfare has arrived, that we are
living through a revolution in war fighting unlike any — at least in my own
lifetime — in which drones and robots and autonomous weapons are remaking
battlefields.
And your professional work puts you at the center of this
shift.
You are the president and chief strategy officer of Anduril,
which is a defense technology company that’s trying to be the hub — or a hub,
at least — for autonomous warfare. But you’re also someone with a deeper
background in national security and American government. You worked as a policy
adviser to Condoleezza Rice, to John McCain, and you’re the author of a book
about the high-tech military future.
So I want to start by asking you to describe where we are
now, generally, to someone, let’s say, who fell asleep at the end of the Iraq
war and just woke up.
Brose: In order to talk about the future, we probably also
have to talk about the past and present. So if you look at, I’d say, the
assumptions that we have been operating under for the past 30, 40 years, I
think that’s what’s driven the kind of military that we have.
We have
assumed that if America is ever going to have to fight a war, we are going to
enter the battlefield with technological superiority against any rival; that we
have military primacy in the world and dominance over any potential competitor;
and that if our military is called to fight, the war’s not going to last very
long. We’re not going to shoot a lot of weapons. We’re not going to lose a lot
of ships and planes and other types of big military platforms.
So we have
built and sized and shaped our military around exactly the kinds of systems
that you would expect to flow from that assumption: Very expensive, very
exquisite, very hard-to-produce military systems and weapons.
When you
look at the future, I would argue that the assumptions that are now very
evident to us in the present are almost the opposite of what we’ve built our
military around. I don’t think that we have the kind of military dominance that
many of us in the 1990s and early 2000s just took for granted. We have peer
competitors and rivals in the world who are adapting to and really disrupting
the American way of war.
I think that
we are going to find a much more contested battlefield, where we’re going to
lose a lot of planes, ships, satellites and other things. We’re going to shoot
a lot of weapons, and we’re going to have to replace that as an act of
production over a long period of time. I think that is not a future that we’re
really ready for.
All of this
points in the direction of autonomous systems, lower-cost systems — things that
are much more like consumer technology or commercial capabilities than they are
legacy military capabilities.
Douthat: And this isn’t just the future you’re describing.
This is the present of one major ongoing war right now, the Iran-America war.
Brose:
Present and recent past. I think this has been apparent going all the way back
to, frankly, the Middle East in the past six or seven years. I think all of the
technology that everybody is talking about, in terms of one-way attack drones
and other things, were evident on the battlefields in Iraq and Yemen and Syria,
going back to 2017, 2018, 2019. Then, obviously, the war in Iran puts this all
in high relief.
It’s a way of saying that we tend to have this belief in the
United States that the future of war is something that’s going to happen to us
in 10 years, and we have a long time to get ready for it. I think it’s been
unfolding for years and is very much right now a present problem.
Douthat: So let’s just use Ukraine as a template or a case
study, because it’s the biggest conflict and it’s the one I think that
Americans have followed the most closely.
Brose: Yeah.
Douthat: The Ukraine conflict starts out with Russia trying
to do basically a sprint to the capital, Kyiv. Basically their equivalent,
maybe, of the U.S. sprint to Baghdad.
Brose: Yeah. “Shock and awe.”
Douthat: Right. And that doesn’t work. And very quickly, the
conflict becomes a grinding stalemate.
But how quickly do drones and autonomous weapons change the
nature of that conflict? How would you describe the role that they play, for
Ukraine itself especially?
Brose: I think that it is not something that happens
immediately.
It was only, I think, once the battlefield lines hardened
and you began to see both sides struggling to advance and gain ground. It
basically becomes a hider/finder problem, and it became very difficult to hide
on that battlefield. So things like tube artillery became increasingly risky
bets to make. That’s where I think you started to see attack drones really
taking the lion’s share of the burden in terms of the killing that they were
doing and being critical to military operations, which they are today.
Douthat: For someone who hasn’t watched a video of an attack
drone in action, which you can in fact watch on the internet, what does a
one-way attack drone do? Describe a typical mission for one on the
Ukrainian-Russian front.
Brose: There’s going to be different kinds, for sure. These
are small hand-carry drones that you can fly either autonomously or
human-piloted. They have quite capable sensors on them, in terms of being able
to fly over an area and identify people or military systems that you want to
strike.
And most of those systems have been weaponized — they’re
carrying small amounts of explosives, so you can then literally just fly them
into the target. You see these horrible videos on YouTube of Ukrainian soldiers
running away from these drones that are chasing them down, horror movie style.
And then I think you see the larger, more complex operations
that both sides have also innovated in conducting, where they have larger
drones that look more like missiles, and other drones that fly out in advance
of them and do a kind of targeting, spotting of targets, feeding that
information back. These larger drones are operating more like precision-strike
weapons.
So the ability to have these systems that are out there on
their own looking for targets, identifying systems or personnel and then are
able to fly out and strike those targets with precision, and do it all at a
price that’s affordable — I mean, that is how Ukraine has stayed in the fight
for over four years. They would not have been able to do this otherwise.
Douthat: Is there a near future where infantry itself starts
to be obsolete and you literally just have drones and robots maneuvering
against each other? Or is that still further out into the hypothetical?
Brose: I think that’s further out, if it’s ever something
that becomes feasible, simply because, so long as human beings continue to live
on and inhabit the Earth — which I’m pretty sure we’re going to do for the
indefinite future — I think it becomes very difficult for these types of
robotic systems to entirely go in, take and then hold ground. We’ve seen plenty
in the conflict in Ukraine that militaries can be, at various different times
in the battle, adept at taking ground. It’s the holding of it that becomes very
difficult.
The question then becomes: Can those gains be solidified?
Can those gains be held entirely through nonhuman means? That’s not a bet that
I would make at the moment.
Douthat: Let’s talk about Iran, because this is a war that
the United States is directly involved in.
Brose: Yes.
Douthat: We’re not just funding and observing as we are in
Ukraine. How much of the specifics of the Iranian stalemate are connected to
technological change?
Brose: I would argue that Iran is still in the fight in
large part because of the sort of technological systems that we’re talking
about. If you believe public reporting, we’ve done an enormous amount of
military damage. The claims of sinking their navy and destroying their air
force and destroying their air defense systems, of going after their
military-industrial capacity — a lot of that’s happened.
At the same
time, the war is continuing because they’re still capable of building, fielding
and using one-way attack drones. These kinds of robotic drone boats that are
quite effective in threatening the Strait of Hormuz and keeping that area of
the world closed — these systems are largely the reason I think that they’re
still able to project power and still able to hazard the United States, our
allies and partners in the region.
This could all change in a week because of the nature of how
quickly these things change.
Douthat: I know you’re not inside government, but do you
think that the Pentagon was prepared for the kind of responses that Iran has
offered to us? Do you think that this has gone as the military expected, and
maybe it’s just the political side that didn’t anticipate it? What’s your take?
Brose: I
have a hard time imagining that if the premise of this was “we’re going to very
quickly decapitate the leadership, strike, and they’re going to sue for peace”
— it might happen, but I think the whole nature of a military is to plan for
the worst-case scenario. So I have to imagine that that kind of planning was
done.
The reality is we are still in the fight. We are still
striking targets. We are still conducting the military operations that the
military’s been focused on. But when you look very closely at the statements of
the chairman of the Joint Chiefs of Staff, about munitions inventory, he’ll
say:
Archival clip of Dan Caine: Well, sir, we have sufficient
munitions for what we’re tasked to do right now, that’s what I hear from the
COCOMs, but what I will say is that we always want more. So I appreciate the
effort of this committee and the Congress, we’re always gonna want more
munitions.
The premise or assumption being that if somehow that
changes, or if the military objectives change, or if the conflict lengthens,
that might be a different situation.
I think that
there are certain things that Iran has been doing that were pretty easy to
forecast. Closing the Strait of Hormuz was something that the U.S. military has
been worried about for a very long time. The ability to project power through
one-way attack drones, etc. — this is something that they’ve been doing, again,
for a very long time.
I think you
can take issue with how we’ve responded to that, which suggests maybe we
weren’t as ready as we needed to be, and that maybe there are things that we
needed to be doing — learning the lessons of Ukraine ourselves and changing the
way that we build our military to be ready for these kinds of disruptions. But
it is clearly not creating the political outcome or the military outcome that
at least, again, through public statements from the leadership of our country,
was the intent going into this.
Douthat: Let’s talk about American military readiness, in
general. There’s a lot of talk about how the U.S. is burning through its
stockpile of missiles and munitions.
Brose: Yeah.
Douthat: War
with Iran is not a major great power war.
Brose: No.
Douthat:
It’s not war with Russia. It’s not war with the People’s Republic of China.
Looking at what’s played out in the Persian Gulf over the last month, and
looking at trends generally, is the U.S. prepared for a major war?
Brose: If
you look at it narrowly in the question of munition stockpiles, which is a
pretty important indicator of military preparedness, I would say no.
And this has been known to us for a very long time. I think
the deeper question is: Why is that, and how do we fix it?
In the
opening days and weeks of Operation Epic Fury in Iran, what I’ve read publicly
is that we fired something like eight years’ worth of Tomahawk missile
production. That’s an exquisite weapon. It does remarkable things. The problem
is that it takes a very long time to build, and once you shoot it, it takes
time to replace. And we don’t have an infinite supply of them.
So if you
look at why we are not ready for this, it goes back to the comment I made at
the beginning, which is the assumptions that we’ve made about warfare. Our
assumptions are that we would not have to fight protracted conflicts. We have
assumed that we would get into a war, we would enter the battlefield with
dominance — with all of this exquisite military capability — and the war would
be over very quickly. The fact that we don’t have deep inventories of
munitions, for example, is not something that we’d have to worry about.
And again, this has been known for a long time, so ——
Douthat: I guess my sense had been, and you can tell me why
this is wrong, that the goal of the U.S. military was supposed to be to fight
at least one protracted war. And there would be controversy back and forth
about whether we are capable of fighting more than one. Like, are we capable of
fighting Russia and China simultaneously?
But it seems like ——
Brose: The goal was one major regional or theater war. There
was nothing said about the duration of that war.
Douthat: OK.
Brose: So
when you look at national defense strategies going all the way back decades or
so — for instance, how big do we have to build the military? What’s the shape
of the military we have to build? — it was all built around this idea that we
had to be able to fight two major regional conflicts at once, and then that got
downgraded to one. But the assumption was that those conflicts were going to be
over very quickly.
Back to your
question on munitions, over the past ten years, as a country, we’ve actually
tripled the amount of spending that we’re putting into Patriot missiles and
Tomahawk and these kinds of weapons that now have household names. The
challenge is that even as that spending has gone up 200 percent, 300 percent,
production has not moved in a commensurate fashion. It’s gone 14 percent, 23
percent up. And you can go critical munition by critical munition and see that
we’re putting a significant amount of resources in, and we’re not getting
significant or commensurate amounts of production out.
To me, the
problem is they are remarkable pieces of technology, and we need them to do
what they are uniquely built to do, but they’re effectively artisanal products.
They’re luxury goods.
Over the past 30 or 40 years, we have had a predominantly
high-end military — a very exquisite military — and its systems that we all are
very familiar with. It’s all of these weapons that we talk about. It’s F-35s,
it’s aircraft carriers, it’s submarines — and we need that. I’m not here
suggesting that we can just do away with all of that. And I don’t think that
the conflict in Ukraine has rendered that stuff obsolete.
Douthat: Can we just pause? Because you’ve used the word
“exquisite” now a number of times. And you don’t mean, I think, “exquisitely
beautiful” like a Ming vase. What is an exquisite weapon?
Brose: The way I use “exquisite” — and others may define it
differently — it means it is something that is very scarce. It is very hard to
produce. It is something that you’re never going to have a lot of. It is going
to be very technologically sophisticated or difficult to make.
The problem
is we can’t do only that. And I think you’re now starting to see, in the
Pentagon, the leadership recognizing that you need what’s referred to as a
“high-low mix.” You need all the high-end stuff, but you also need this
lower-end of capabilities that are going to be more producible, more
affordable, oftentimes more autonomous.
I think the
lesson of history, and really, the lesson of these recent and ongoing
conflicts, is that these technologies are going to change warfare constantly.
And you’re now dealing with adversaries that are quite capable, quite
sophisticated, quite high-tech in their own right, very disruptive — and that
what we are going to field is inherently going to give us a fleeting advantage.
That, I think, is the real lesson that we’re going to have
to take out of this. It’s not over-indexing on a particular piece of technology
or a particular way of fighting an old system or a new system — it’s how do we
consistently stay at the cutting edge, learn and field?
That is, I think, what has made the Iranians so effective,
and where the United States military, frankly, still has a lot to learn. And
the U.S. bureaucracy that supports that military definitely has a lot to learn.
Douthat: All right. So, Anduril. It’s a Tolkien reference.
To what? I know the answer, but some people might not.
Brose: So the name is a conscious nerd detector. We have
identified you as such. The reference, Andúril, is Aragorn’s sword in “The Lord
of the Rings.” “The Flame of the West” is, I think, the Elven translation.
Douthat: Narsil reforged.
Brose: Exactly.
Douthat: It’s a broken blade that’s been reforged.
Brose: Correct. And there is an active debate about the
proper pronunciation, of ANN-dur-ril or AHN-dur-ril.
Douthat: How do you say it?
Brose: Look, I’m a pretty conventional guy.
Douthat: You’re not a native Elvish speaker?
Brose: Suburban Philadelphia, so it’s ANN-dur-ril to me. I
think others might insist that the proper Elven pronunciation is AHN-dur-ril.
These are the fights we have.
Douthat: I’m glad we haven’t clarified that question. So
having a military-industrial company named for something in “The Lord of the
Rings” is something that you obviously have in common with Palantir.
Brose: Yeah.
Douthat: Another cutting-edge military technology company,
one of whose representatives has been a guest on this very show.
You’re both contractors doing work for the Pentagon.
Sometimes you work on the same project. How are you different?
Brose: It’s a great question, and I think it’s very easy to
think that we’re the same. Putting aside the name, we have common DNA in the
company, common investors. We’re both fundamentally software-centric companies.
We’re just not working with a lot of the customers that
Palantir is working with. We don’t work with the I.R.S. or Health and Human
Services, because what we do is just not relevant to them.
I think the
obvious difference is that Anduril is also building hardware. We’re building
sensors, drones, autonomous systems, weapons. We’re actually doing that
manufacturing.
And the software that we’re building is a software system
that we call Lattice, which is actually very tightly integrated and coupled
with that hardware in the sense that it is fundamentally focused on the
autonomous operation of machines and robotic systems. How an individual drone,
for example, is perceiving its environment, maneuvering through space,
identifying objects of interest, moving information and collaborating with
other robots, with all that software having to run inside of that robot.
To get concrete about it, we do a lot of work in
counterdrone and air defense. And what Lattice has to do as a software system —
what we have to do as a company — is actually be able to take the information
that’s coming out of those sensors — radar feeds and imagery from cameras and
signals intelligence from electronic warfare systems — and the software has to
be able to build an understanding of the world. It has to build objects of
interest, transform data into objects and targets, to differentiate a bird from
a drone or an airplane from a missile or a civilian airplane from a military
airplane.
Then, having done that, it needs to be able to communicate
with those machines and tell them to keep custody of that target, to keep
looking at it, to be able to track it through space and time so that the system
— the software system — can then task a weapon to go shoot it or defeat it, if
that’s the intent of the human operator. So it’s just to say that this kind of
system is what the military would refer to as a fire-control system, because
you are literally controlling acts of violence through software.
And this is
why you look at the OpenAIs and the Googles and others — I mean, we have
natural relationships and partnerships with all these companies. They’re
incredibly eager to work with us because we have a treasure trove of unique
military data that you’re not going to find in a public setting. You’re not
going to scrape it off the Internet.
Douthat: But you’re also building the weapons themselves.
Brose: Correct.
Douthat: So you’re building the model through which soldiers
and operators interface with drones and counterdrone technology.
Brose: Yes.
Douthat: But you’re also building drones and counterdrone
technology.
Brose: We’re also building hardware. Right.
Douthat: So, describe the hardware. Give me a couple
examples, offensive and defensive, of the hardware and what a typical product
looks like.
Brose: One
of the bigger systems that we’re building right now is an Air Force program
that we competed on and won called the Collaborative Combat Aircraft, which is
a fancy way of basically saying a robotic fighter jet. So, fully autonomous,
launching entirely without human control, maneuvering through the battle space,
conducting operations under the supervision of human beings, firing weapons,
and again, doing all of this over very long ranges, carrying large amounts of
payload, sensors, weapons and other things.
That is the
kind of system that we’re building that is both a very unique and
differentiated piece of hardware. But what makes it special is everything that
is smart and intelligent that’s going on inside of it, which is foundational,
and that’s what the Lattice software system is doing.
Douthat: Is that cheaper than the alternative? Is that an
example of the kind of lower-cost warfare that you were just talking about?
Brose: Yeah. That’s what it means.
Douthat: Because that sounds very high-end to me.
Brose: It is high-end, but I think this is where ——
Douthat: Even exquisite.
Brose: There
are many things that we’re doing that I would say are exquisite, but that also
doesn’t mean that they’re exorbitant in terms of their price or in terms of the
time it takes to produce them or change them or modify them. This is an
incredibly capable system, but it is a fraction of the cost of an F-35, for
example.
Douthat: OK.
Brose: And the whole notion is that it is an unmanned or an
uncrewed or autonomous system, so that you can take risks with it that you
would never take with a $100-million airplane that has a human soul aboard.
This isn’t a robot that is going to necessarily replace human beings and
exquisite military systems. It’s going to make those systems more capable and
more survivable because you can collaborate with them. You can not just send a
single pilot in a single fighter jet out to conduct an operation — you can have
that pilot flying with three or four or five of these robotic wingmen who are
now capable of operating almost as, you know, wingmen.
Douthat: Why do you need the pilot at all, though? We can
get into the question of whether you need any humans in the system in a minute,
but if this works, aren’t you replacing, at the very least, Tom Cruise?
Brose: I mean, maybe someday. I don’t think that’s where we
are now. And I would argue it’s probably not going to be the case that we’re
going to replace humans entirely in the future. But why do you need humans now?
Because the systems are not advanced enough that they can just operate
completely without human supervision. You need a human being that is going to
be in command of those robotic systems in the way that a human in a fighter jet
is going to be collaborating with those robotic aircraft.
Douthat: But is there an advantage specifically to having
the human be in the fighter jet with his wingmen?
Brose: I would argue there is.
Douthat: Or could you have them just in the command center
piloting from afar?
Brose: I think that there’s an advantage that comes from the
human being in that same operational environment and being closer. In a world
where we’re seeing jamming and denial of communications, the ability to have an
aircraft that is physically in communications with those airplanes, even as
they are operating in highly autonomous ways, I do think is a necessity in
terms of how we are going to think about the ethical use of this technology. I
think that’s important.
I also think the military advantage is that you look at this
more — I look at this more — as the augmentation of human beings with
intelligent systems or robotic systems. Ultimately, it’s not a question of
whether the human is better or the machine is better, but is the human-machine
team actually more effective than either the robot or the human by itself?
And I think a lot of the analysis that we’ve run, and a lot
of the analysis that we’ve seen from the government, is that that is absolutely
the case.
Douthat: What about on defense? In terms of both drone and
missile defense, give me an example of work you’re doing.
Brose: So air defense is an inherently integrated problem,
which is why it’s referred to as integrated air and missile defense. You are
basically needing to do what I was describing a minute ago, which is you have
to complete actual kill chains. So the ability to understand what’s happening
in the environment, be able to identify and target the systems that you want to
defeat, and then be able to task weapons or military effects, kinetic or
nonkinetic, to defeat those threats and do all of that in a matter of seconds,
to operate it at a speed that a human being is not going to be able to keep up
with.
One of the systems that has done remarkably well in
protecting American bases and American aircraft in Operation Epic Fury is the
electronic warfare system that we built called Pulsar.
Douthat: Take that example. You’re trying to defend, let’s
say, a U.S. base in the Middle East. So that is trying to shoot down both
missiles and drone attacks?
Brose: So this is doing this in what the military would
refer to as nonkinetically.
Douthat: How would you or I refer to it?
Brose: It is preventing a weapon from being able to strike
its target by using energy, using the electromagnetic spectrum, to disrupt its
operations. At the most basic level, that is what it is. And to get more
technical, it begins to sound like witchcraft because it mostly is.
Douthat: So you’re not firing interceptors?
Brose: We’re also doing that. Basically, this is a military
interceptor that is going to launch from a base, that is going to fly out in
much the same way that an air defense weapon will. It is going to be able to
identify a target, to physically run into it and explode and knock it out of
the air.
Douthat: Say a little more about the witchcraft, though.
Let’s say I’m running a U.S. military base in the Middle East, and I have a set
of drones — Iranian drones — heading towards me. On the one hand, I have as
countermeasures my set of maybe ideally reusable rockets that will intercept
and take them down. And so some subset of the incoming fire is taken out by
that, and then some subset is disrupted through electromagnetic fields?
Brose: Yeah. It’s basically what the military would refer to
as a layered defense. I think that traditionally we’ve thought of air and
missile defense as very point-specific — I’m trying to defend a very small
piece of real estate, a military base, and the volume of threats that I have to
defend it against is not very large.
And I think what we’ve seen is that with the proliferation
of these one-way attack drones in Ukraine and Iran, air defense is shifting
into more area defense. You have to be able to protect very large areas of
territory — whole cities, whole regions — and it just requires a much larger
volume of sensors, a far larger volume of weapons, and it basically requires
you to integrate everything.
So the first thing that you’re going to do, once you see an
air defense threat, once you see an inbound missile or a one-way attack drone,
is hopefully not shoot a Patriot missile at it — you’re going to deploy
something like what I’m talking about in the Pulsar system, which is an
electronic warfare system. It’s reusable. It is essentially trying to defeat
inbound threats with jamming, with energy.
Douthat: Is it shooting something up?
Brose: No, it’s a directed energy weapon.
Douthat: It’s creating fields around …?
Brose: It’s a force field, Ross. It’s effectively a force
field.
Douthat: That’s what I want you to say. There you go. Yes.
Brose: So for the listeners, it is a force field. We have
done it. The next up will be cloaking devices.
Douthat: But it functions like there is a zone of
electromagnetic interference and missiles or drones enter this?
Brose: It’s more targeted than that. It’s not trying to
basically just pump huge volumes of energy out into the environment and do kind
of broad-spectrum jamming.
Douthat: So then, the second layer would be things that come
in, and then you start ——
Brose: If that didn’t work for some reason, then you start
shooting.
Douthat: But this is what’s being used now?
Brose: This is being used now, across the world.
Douthat: For instance, does Israel’s Iron Dome use the
directed energy?
Brose: We are not currently a part of their program, but I
have to imagine that Israel has similar capabilities that they’re fielding as
part of Iron Dome.
Douthat: And long run, be speculative — this is obviously
designed for base defense. In Israel’s case, you have missile defense geared
around cities and regions. Do you imagine this long-term as a version of or an
upgrade of missile defense for the American homeland? Is there a world where
this ends up being sort of a means of derailing a nuclear attack, or is that
just in a completely different realm of threat?
Brose: Yeah, I think the nuclear realm is just a wholly
different animal. But in terms of being able to protect military bases or
critical infrastructure from drone attacks here in the United States, we’re
doing a lot of that work as a company. The electronic warfare system that I
mentioned is part of that work, but it is a much larger problem that again gets
back to the question of how we do this kind of air defense mission at scale.
Douthat: How hard is it to get this to actually work? And
here, I’m just going to throw at you — and you can respond to any of them —
some examples of publicly reported Anduril failures.
So there’s reporting that Ukraine used some of your drones
and then stopped using them after a certain period of time because there were
battlefield failures. There are specific case studies, like one of your
counterdrone system tests in Oregon reportedly sparked a 22-acre wildfire.
There was an engine-related failure to one of your autonomous fighter
prototypes, Fury.
Those would be examples. How well does this stuff work? And
how hard is it to get it to actually work?
Brose: Yeah, without relitigating the individual incidents,
because the engine fire just didn’t happen, and we corrected the misimpression
that a reporter had because it had been fed wrong information ——
Douthat: Well, how about ——
Brose: The bigger point is we fail every single day. Right?
I mean, that’s the tip of the iceberg in terms of the failures that we
experience as a company. But that’s, I think, the whole point here.
The answer to your question of how hard it is to make this
stuff work — it is incredibly hard. It is far harder than people realize. And
for that reason, we have invested as a company. This is our own money. This is
not money that the taxpayer is funding us to do. At this point, I think we have
about 330,000 acres of test sites and ranges across the United States that ——
Douthat: Where are those?
Brose: Undisclosed locations.
Douthat: OK. Mostly in the West?
Brose: In places where you can afford to have large amounts
of space.
Douthat: Not in Connecticut? Just for my own ——
Brose: Um, no.
Douthat: OK.
Brose: But if you want to open a test site in Connecticut,
Ross, we should talk. It could be your side hustle.
Douthat: OK.
Brose: It’s to say that we go out every single day and break
these systems. We figure out what their capabilities and limitations are. We
develop them to the point where they are going to be trusted by an operator
that’s going to use them, but even then, the expectation cannot be that when a
system shows up in Ukraine, it’s going to work immediately.
And I think that is an experience that any company that has
been operating in Ukraine since the beginning of the conflict, which we have —
and the military systems that we’ve been fielding are still in operations in
Ukraine — no one has had an experience where they fielded something that has
just been perfect and worked out of the box. And to the extent that it did, it
didn’t work a week later or a month later or a couple of months after that.
Again, the whole lesson of this is not whether we can build
a perfect piece of technology. It’s whether we actually have the right
institutional culture or organizational culture where we are learning, we are
testing, we are training these systems, we are building trust in them, we are
enabling operators to build trust in them, and when those systems don’t work —
because they won’t — because of things that new technology makes possible or
responses that our adversaries engage in, we have to be able to modernize those
systems and improve them and change them and rebuild them.
And that’s what we do every single day. We’re integrating
new sensors and new payloads onto these systems. We’re rebuilding entire
systems altogether at a hardware level. Entirely new versions of aircraft are
another thing.
Douthat: And actual war is an accelerant.
Brose: It is.
Douthat: The Iran war is accelerant of this technology’s
development.
Brose: Yes. At least it should be. At a technical level,
it’s an accelerant because you are in a nonfail environment where you have to
perform and you have to get better, and it’s all mission-critical.
That kind of
incentive is exactly the kind of pressure that we want to be under because, at
the end of the day, we are building autonomous systems. And autonomous systems
are by nature very finicky. You take them into one environment, and they
perceive the world differently. You take them into a different environment, and
you have to do a certain amount of retraining.
All of this
ultimately comes down to the trust that human beings are going to place in
them, so if those systems are not predictable and reliable, if they don’t do
things in a repeated and high-trust way, operators are just not going to use
them.
There’s no points for being autonomous. There’s points for
being effective and for being useful. Autonomy is a means to that end in our
mind.
Douthat: What about the capacity to build the kind of things
you’re building at scale? You’re one company.
Brose: Yeah.
Douthat: You said earlier, basically, that under current
conditions, we can spend a lot of money, and with the existing legacy weapon
systems we have, we just don’t get that large a number of missiles, and so on.
Does the U.S. have the existing industrial capacity to build
out autonomous weapons — drones and other things — on the scale that we need in
an environment where you’ve just said that we aren’t maybe well positioned to
win a major war?
Brose: I think the answer is yes and no. I think the yes
answer is we absolutely have the ability to stand up industrial infrastructure
and bring the work force in. These are not problems that we’re concerned about.
We just closed another fund-raising round, so we’re taking in $5 billion of
private capital.
Douthat: By the way, you mentioned earlier the venture
capital role here — this is a different model from the traditional defense
contractor, right?
Brose: Yes.
Douthat: You are effectively raising money, building
weapons, and presenting them to the Pentagon before a contract is signed?
Brose: In part. There are plenty of instances where we have
conviction in a solution that needs to be built to solve a military or national
security problem, and we go out and do exactly that. We put our own money at
risk, we spin up a team, we build a system, and we take that system to the
government and say: You probably wouldn’t have come up with this on your own,
but we think that it is an answer to your problem.
As we’ve become a larger company, we’ve also had to change
the business model and engage in different bets, which I would argue makes it
start to look much more like the commercial economy that we’re all familiar
with, which is: If you give me a contract to build weapons the traditional way
of doing it, the government is going to pay the provider of that traditional
weapon all of the costs to build their facilities, to hire their people, to
cover the overruns when the program goes over budget and off schedule. The
taxpayer’s on the hook for all of that, and the industry partner really isn’t
bearing a lot of that risk.
I think what is changing now, and I think Anduril has done a
lot to change it — but I also give a lot of credit to the current leadership of
the Pentagon that’s trying to change this business model as well — is let’s try
to make this more like a commercial transaction, where I sell weapons and you
want weapons. You give me a contract to build a lot of weapons, I’m going to go
off and build those weapons. If I am off schedule or if I underperform or if
I’m behind, that’s all on me, and I’m the one that has to eat the cost of that
failure in order to meet the obligations of my contract. The same way that you
would put a roof on your house or go buy a piece of electronics technology.
Douthat: Well, except that, unless you start selling to
other countries, it’s a market with only one buyer.
Brose: It is ——
Douthat: Which means that you are, as a company, in a
position where you are sort of at the mercy of political changes, political
decisions.
I want to ask about this because Palantir, again, has a
higher profile than you do, and has a specific political profile. There’s lots
of people who hate and fear Palantir, who associate it with general fears about
the surveillance state, but also specific fears or concerns about its alignment
with Republican politics — the Trump administration, and so on.
I think this is less of an issue for Anduril — see? I did
the Tolkien pronunciation there. But your prominent founder, Palmer Luckey, is
a prominent Republican. Obviously, there is overlap with Palantir — you
yourself have mentioned it. One of the funds investing in you is connected to
Donald Trump Jr.
Is there a world where your business model just goes away in
a Democratic administration? Like, “Anduril — those were the Trump guys or the
Pete Hegseth guys.” How vulnerable are you to politics?
Brose: No, I personally don’t think that we are that
vulnerable. I think that from Day 1, we recognized that Anduril is a defense
company, and defense is a long game. You have to be able to play that long game
not knowing what the future politics of the country are going to hold.
For us, politics is something that individuals can engage
in. Palmer certainly has his political profile. Our CEO is a very public and
avowed Democrat. I have no idea what I am anymore. But it’s to say that it
doesn’t matter.
At the end of the day, what we’re focused on as a company,
we are actually engaged in a bipartisan way across both houses of Congress,
both parties, to provide an enduring capacity in this country, to build the
kinds of defense technologies and capabilities that I think, on both sides of
the aisle, people want.
Douthat: So let’s say that vision is right, and you become
an enduring part of the defense industrial base, and the next president is a
Democrat. It doesn’t matter — you’re doing the work you want to do. And we get
into the early 2030s, and we get into a war with China over Taiwan. Give me a
brief vision of what Anduril’s success looks like in that environment. What
actually happens?
Brose: First and foremost, the actual success that we
measure ourselves by as a company is whether we prevent that war from ever
happening.
Douthat: Sure. OK.
Brose: I know you get it — table stakes. But it needs to be
said. Because, look, we are a defense company. We do build military technology.
But we’re not excited for it to be ——
Douthat: Then put it this way: Describe the scenario that
China would fear that would induce them not to go to war.
Brose: So some specific things, because I think the
specifics are what is interesting. Ultimately, we want to contribute to
America’s ability to defeat their strategy. And their strategy is, I think —
not reading their internal emails or what have you — they want to prevent us
from being able to project power. They want to prevent us from being able to
come to the aid of our allies and partners in the region, whether it’s Taiwan,
Japan or someone else. And they want to be able to endure through what could be
a protracted conflict by keeping us out of it.
Douthat: That means, concretely, they want to, initially,
destroy our bases and drive our aircraft carriers out of the region.
Brose: They say it very clearly. They say it is winning
without fighting. They want to prevent us from being able to fight, and that is
the sort of deep disruption that has been happening to the United States
military for the past 30 to 35 years. China went to school on us in terms of
how we fight and with what we fight, and they have been building up and
modernizing a military to hold all of that at risk.
So, specifically, we want to be able to break that
advantage. We want to make the bases and ships and aircraft carriers that are
going to be under threat in that region defensible, having larger numbers of
robotic systems that are purpose-built for the Indo-Pacific region.
I think a lot of people look at the conflict in Ukraine and
say: On the one hand, we just need to buy everything that the Ukrainians are
building, quadcopters and others. Or on the other hand ——
Douthat: Or have the Taiwanese buy them.
Brose: Well, I’ll get to the Taiwan question and the allied
question in a second. But the criticism of people who over-index on the conflict
in Ukraine is: Oh, none of this stuff is relevant in the Indo-Pacific region.
The distances are too vast. The geography is so large. The threat from the
Chinese is so much greater.
And there’s a degree to which that’s true. I think the point
is that what we actually need to be doing is what we’re doing as a company,
which is not just carbon copying what the Ukrainians are doing, but learning
the lessons of that battlefield and the attributes of those technologies and
building different kinds of military systems that are purpose-built for the
United States and its allies, for different geographies, and for a far higher
bar of threat that we would be facing from the Chinese Communist Party.
And specifically to your question about Taiwan, because
these are the places where you have the flash points of conflict where a
conflict might emerge, I think that we have to do our best to support the
Taiwanese government in making themselves more defensible, making themselves
into the kind of adversary that is so distasteful that the Chinese military and
the Chinese political leadership just doesn’t want to take that on.
So looking at this as ——
Douthat: And that means, like, drone swarms when you have an
attempted amphibious landing?
Brose: Yeah, making them more defensible from the types of
inbound missiles, cruise missiles, ballistic missiles, one-way attack drones
that they’re going to be faced with, giving them the ability to defend against
and endure that sort of onslaught. The ability to project power in a way that
keeps them in the fight.
That is work that Anduril is doing. We’re providing that
capability — obviously with the U.S. government’s blessing — directly to the
Taiwanese armed forces. And it’s something that we intend to do a lot more,
provided that they have the will and the means, and the U.S. government
continues to back us in doing that.
Douthat: Let’s end with the ethics of this strange new
world. You talked earlier about emphasizing the idea of keeping humans in the
loop, keeping humans involved. What are your actual core principles as a
company when it comes to what is allowed to happen autonomously in a conflict
area?
Brose: My general frustration with the ethical debate right
now is that it feels very unsophisticated and stale. On one side, you have
folks who basically are looking at the reality of the security dilemma that we
face, which is that our adversaries are going to do all of this. They’re going
to build these autonomous systems, they’re going to take human beings out of
the loop, and they’re going to gain a military advantage. And if we don’t do
that, we’ll be on our back foot. And there’s a degree to which that’s real.
Douthat: By the way, in the case of Ukraine right now, you
have drones killing people without a human in the loop already, right?
Brose: Again, I don’t know everything that’s going on there.
I actually think a lot of what you’ve seen more recently is you don’t have
that. You have drones that are being piloted by very capable drone pilots that
are the ones flying them into their targets, whether those targets are people,
soldiers, or military systems. I don’t think that you are seeing a large amount
of automation yet on the battlefield, but I think that, too, could change.
If the
Ukrainians start building up more of an internal capacity to shoot longer-range
one-way attack drones in volume deeper into Russia, you’re probably going to
have to automate more of the operations of those systems.
Douthat:
Then you would be just sending them out and saying: Go until you see something
to kill and kill it.
Brose: That is a version of it. And again, to go back to
your question about the ethics, I think on the other side are folks who look at
this and say: This technology is so new, it’s so unprecedented, that we just
have to ban it outright. We just have to never go there — and I think that
that’s also not a sophisticated position.
To answer your question directly, what I think we try to
stay focused on as a company is the reality that over the past decades and, I
would argue, centuries, Western civilization — the United States in particular
— has built up a body of laws and policies and doctrine that enables us to
bring brand-new technology into military use, to weaponize it and to use it in
a way that is still ethically constrained and thought about in a serious and
ethical way.
What we need to be doing a better job of as a country is not
throwing out this remarkable body of ethical thinking and action and doctrine
that we’ve built up, but actually applying it to govern how these technologies
are being built and introduced to military operations. There’s nothing to me
that is so new and so unique and so unprecedented about these technologies that
the ethical frameworks that we have brought to bear to solve past problems
can’t enable us to solve these future problems.
Douthat: OK.
That’s very general, though. I want you to make this concrete. What’s your
understanding of the constraints on what the Pentagon allows a drone or an
autonomous weapon to do without a human deciding, “Kill this person, shoot this
person”?
Brose: So if you look at the actual policy, it’s a very
serious document, but it’s also quite broad, and it would account for a lot of
things. I actually think that’s right and well and good.
I think that people also need to realize ——
Douthat: Wait, wait, wait. But what does it say? I know it’s
very broad, but is there a specific——
Brose: The
more important thing is what it doesn’t say. It doesn’t say you’re not allowed
to automate the kill chain. It doesn’t say that you’re not allowed to build a
military system that is capable of basically being a lethal autonomous weapon.
Douthat: So
you’re allowed to do that?
Brose:
You’re not not allowed to do that.
Douthat: Right.
Brose: I think the point is that people also have to
appreciate that bureaucracies by nature, and military bureaucracies in
particular, are inherently conservative. And I think that it’s a fundamental
misunderstanding — and I’m not saying that you’re guilty of this, but I think
that many are ——
Douthat: I could be guilty of it. That’s OK.
Brose: It’s a misunderstanding of our defense institution
that they’re just going to take a bunch of unproven technology and then
willy-nilly throw it onto the battlefield and see what happens. Because at the
end of the day, what they are accountable for and the consequences of their
work is life and death. Not just life and death with respect to enemy
combatants and civilians, but also for our own personnel.
You don’t
want a weapon system that malfunctions. You don’t want a drone or an autonomous
system that hallucinates. So there is a process that all new technology has to
go through.
And I think that it is going to be no different in the case
of these kinds of more highly intelligent, highly autonomous robotic systems,
where you are going to rigorously train them to do very specific things. You’re
going to test them to determine that they actually perform those tasks the
right way, effectively, repeatedly, predictably. And in the process of that
training and testing, you’re going to build trust that those systems are safe
to use and effective to use.
Douthat: And where in that process do you determine where
the moral line is for letting robots kill people?
Brose: So to take a specific point of differentiation, how I
would answer your question would be very different in a defensive application
of this technology than an offensive application.
If I’m going to take a highly intelligent machine and send
it downrange to go hunt targets and basically make its own decisions about what
to do, what to shoot, etc., there’s going to be a far higher bar applied to
letting that system go off and do those things than a similar system would be
if it were employed in a defensive setting, like an air defense application we
talked about, where the risk of not doing that is that the human beings that
are under threat from those inbound missiles will be too slow and too incapable
of being able to defend themselves.
So you’re going to be far more willing to put that kind of
advanced technology into a defensive use case because you’re literally
protecting human life, as opposed to an offensive case where you’re sending
that system out to take human life.
Douthat: Is
there a world where the military, for offensive warfare, has a policy that
human beings have to be in the loop, but in effect, this technology is so fast
and you’re so caught up in the operations of artificial intelligence and
autonomous weapons that the human being ends up being just sort of a rubber
stamp? Because that seems like a possible future, too, that you technically
maintain humans in the kill chain, but in practice, you never want to be the
midlevel soldier who says no to what the A.I. is telling you to do.
Brose: Yeah,
I think here, too, it’s also highly contextual. I absolutely can imagine a
world where we build lethal autonomous weapons and we use them exactly as you
were describing.
Again, back
to this ethical framework — when the United States military goes to war, we
will declare what are called areas of active hostilities. And what you’re
saying to the world is: This is a war zone. Do not go there. If you’re a
commercial fisherman or a commercial mariner, do not take your ship through
there, do not fly your aircraft through there, because this is going to be an
area where we are lowering the bar for how we are going to use violence, or we
are going to delegate more of our military operations to intelligent machines
in order to gain that advantage or not lose that advantage.
But you are doing so in a context that you’ve created that
gives you much higher assurance that you’re not going to be making mistakes,
killing civilians, killing your own people.
I apologize for continuing to come back to the context and
the specifics, but I think that’s where all the devils lie in these kinds of
ethical discussions of these new technologies.
Douthat: So then, two last questions on that. Just from your
perspective, as someone who is making weapons for the military but who does not
work for the military, does the private company that does this kind of work
with the military just have to accept as the basis for doing business that you
are trusting the military in these specific circumstances?
Because this is obviously something that’s come up a lot
with the big A.I. companies, which are not defense contractors, but now kind of
are defense contractors — and famously, Anthropic maybe had problems with how
the Pentagon might use its A.I., and this led to conflict between the Pentagon
and Anthropic — what’s your view of the place of the private company doing this
work in thinking about how the weapons are actually used?
Brose: Yeah. I think working with or working for the
government is kind of an all-or-nothing venture. And I think that’s how it
should be. I don’t think that we ever want the builders of technology in
America to basically enact a veto over how our government is using that
technology. I think that is a decision that needs to remain in the hands of the
American people and their elected representatives.
And I think that for companies that want to work with the
government, that’s what you’re signing up for. You are signing up for the
belief that working for our government is inherently good, it is something that
is necessary and right, and that you are believing that the government is going
to be following the laws of the land. It’s going to be constrained by the
checks and balances of the Constitution and our other institutions of
government, but you can’t show up and try to veto individual use cases.
As a builder of weapons in the United States, I can’t show
up to the Pentagon and say: You can buy my weapons, but you can’t give them to
Israel, you can’t shoot them in Yemen, you can’t provide them to the American
military in Operation Epic Fury or share them with the Taiwanese or the
Europeans.
You basically have to say: Look, I am here to provide a
capability or a service, and I am trusting that the government is going to make
policy, they’re going to follow the law, they’re going to be checked and
balanced and constrained by the other institutions of our government.
That’s, I think, what you’re signing up for. If a private
actor, a company or a person doesn’t believe that they can support their
government in that way, providing them that technology or service, the right
thing to do is just to walk away and not do that work.
And there are plenty of companies in America that choose not
to engage in defense or choose not to provide capability to the government. And
that is, I think, an absolutely ethically supportable decision. And I think
it’s a good thing that in America we have that choice, we have that luxury.
So I think, yeah, where Anthropic went wrong — and there’s
plenty of blame to go around — I don’t believe we want these kinds of
technology companies, including us at Anduril, showing up and trying to dictate
to our elected representatives, our leaders, our Senate-confirmed officials and
ultimately the American people what they can and can’t do with the technologies
that we’re building.
Douthat: Stipulating that that’s true and that that’s just
inevitably going to be the necessary approach that you take, I’m just curious,
as a last question, whether there’s anything concrete that you fear with this
technology.
And I’ll
just say, as context, the pattern in major wars in most of human history,
especially where technology is involved, is that you get in a major war, it
goes on a long time, and you escalate the use of new technology to a threshold
of moral danger. This could be poison gas in World War I. It could be the
strategic bombing of cities and the use of nuclear weapons in World War II.
There are other examples. But it seems very imaginable to me that you get in a
major war with autonomous weapons, and a system of moral constraint that works
right now starts to go out the window.
I’m just
curious: Are you ever afraid of where the things you’re building could be
taken?
Brose: I think you have to be. It would be irresponsible if
you weren’t — for me or for anybody who is working in this area of technology —
to fear the future use cases that these systems could be put to or the ways in
which they would develop.
But I think, to your question about those future scenarios,
at the end of the day, the only thing we can focus on is building the best
technology that we’re capable of building, to support the United States, our
government, our allies and partners, recognizing that the future use cases, the
context — all of these things — are going to change and be different.
And it’s hard to answer your question in the abstract
because a few years prior to World War II, if anyone could even contemplate an
atomic bomb, it would be hard to imagine that someone would support that as an
ethical use. The point being, these things remain highly contextual and what
you are willing to condone and what you’re willing to do is going to change
based on the circumstances that you find yourselves in.
I think the only thing that you can do as a builder of
technology is, again, provide the best capability that you can, keep the United
States and our allies and partners on the cutting edge of technology to try to
ensure that those wars never happen. But in the event that, God forbid, we find
ourselves in them, recognize that those are going to be the decisions that the
elected representatives of the American people are going to have to make under
very difficult circumstances, and we pray they get them right.
Douthat: All right. Chris Brose, thanks for joining me.
Brose: Thank you. Great to be here.” [1]
1. Our Military Is Built for the Wrong Century: interesting
times. Douthat, Ross. New York Times (Online) New York Times Company. May 28,
2026
