SOMEWHERE NEAR THE CALIFORNIA-NEVADA BORDER— in the early morning of October 8, 2005, a great race began as the sun rose over the Mojave Desert at Primm, Nevada. Twenty-three vehicles with no drivers set out on a 132-mile course, on jagged roads through desert mountain passes. In the ninth mile of the course, long before the finale over Beer Bottle Pass, the entry from Cornell University hit a concrete wall. The “Spider” was a modified Singaporean Spider Light Strike Vehicle, outfitted with multiple LiDAR rigs, a server in the back seat, and an AC system ducted through the roof to cool the computers. The Spider, after hitting the wall and failing to recover, placed 20th.
It was a disappointing result for seniors Brian Schimpf and Matt Grimm, who had originated the DARPA Grand Challenge team at Cornell. The team raised $450,000, a combination of donated equipment and $116,000 in cash. The donors included AMD, Northrop Grumman, General Motors, SpaceX, Texas Instruments, and in particular Singapore Technologies Kinetics, which had donated the Spider Light Strike Vehicle (“the Singaporean Humvee” as Grimm called it) manufactured for the Singapore Army and sent it all the way to Ithaca.
The after-action report, published by Brian and others in the Journal of Field Robotics, explained the two major problems. First: “it chose to turn into the wall… The only explanation for such behavior is that the Spider detected an obstacle in front of it, though postrace investigation revealed no obstacles in the overpass.” And second: the Spider had no reverse driving capability! “Reverse driving was not implemented due to time constraints, so the Spider could not recover from its collision,” the paper reads. “Instead, it was disabled as it struggled to turn free from the wall.” No turning back was an early principle for Brian, you see.
Today, Brian and Matt are still working together on autonomous systems as two of the co-founders of Anduril Industries, the defense business that has raised billions to upend military technology for America and its allies. As fate would have it, they made it back to that same swatch of desert years later. In August 2022, Anduril made the Nevada National Security Site, a testing range not a hundred miles from Beer Bottle Pass, a testing area for a number of products, including its counter-drone systems. (They’ve since moved on to another site)
And this time, they brought a lot more than $450,000 worth of equipment and personnel to the desert. Anduril has raised billions with Brian at the helm. I went to the Costa Mesa, California headquarters of Anduril to talk to him and others about how to build an engineering enterprise ready to go to war.
“Maybe it was foreshadowing, maybe it was destiny, maybe it was just a natural evolution, but we fell into very similar roles on the DARPA challenge compared to what played out over our careers,” Matt Grimm told me when I interviewed him for this piece about his lifelong friend Brian. “We had a mechanical engineer, an electrical engineer, a computer scientist, and a pointy haired business guy. I’m the pointy haired business guy, and Brian is the computer scientist.”
At Anduril, Matt is the COO, and Brian is the CEO. Maybe you knew that, maybe you didn’t. Anduril’s most prominent leader is, without question, Palmer Luckey, who invented virtual reality as we know it, got acquired by Facebook, got unceremoniously fired by Facebook after endorsing Donald Trump in 2016, and who hasn’t been seen in a suit in years, opting instead for Hawaiian shirts. Palmer Luckey doesn’t need a title with a “C” in it. He’s not CEO, not CTO, not Chairman. He’s Founder.
Anduril is arguably the most ambitious enterprise for American defense since the formation of DARPA itself. And at the center of it is Brian Schimpf, lifelong engineering leader, and a son of Rochester, New York, a city shaped by the work of a bold industrialist. To hear some of the others at Anduril tell it, like founder and executive chairman Trae Stephens, Brian was always the CEO – even before the company existed.
✺
The Eastman Kodak Company of Rochester, New York was formed in May 1892 by one George Eastman. In 1888, Eastman had patented a camera shutter and trademarked the name “Kodak.” Under the Kodak name, Eastman made photography available to the masses. He patented camera film rolls and a number of chemical processes for producing camera film.
The work made Eastman a very wealthy man. He amassed a wealth that in today’s terms would make him a multibillionaire. And in good form for an early American industrialist, Eastman put the money to work in the city of Rochester. Among the things that bear his name is the Eastman School of Music at the University of Rochester. Eastman endowed the Rochester Philharmonic Orchestra, and built them a theater, the Eastman Theater. The Eastman Museum, the mansion which George Eastman built for himself, is the world’s oldest museum of photography.
Brian Schimpf grew up in the world left by George Eastman. Brian’s mother and stepfather both worked as engineers at Eastman Kodak, and it was Brian’s first job, as well – as a 9th grader, he wrote code at Kodak in the summertime at the end of the 1990s.
“I’ve always liked to build stuff,” Brian told me. It began with building carnival games out of wood with other kids in his neighborhood. Brian’s mother, the industrial engineer, made him do expected-value computations on the wooden carnival games, in order to figure out the correct payout schemes. He got into computers, and then robotics. In high school, he worked on engineering solar cars and stage lighting for theater, among other things.
Kodak has seen better days. The population of Rochester peaked in the 1930 census, two years before Eastman’s death. In 1986, Japanese company Fujifilm introduced the first true disposable camera. Kodak followed suit, but from that point it was a steady decline for the company. More than 50 years after Eastman died, the new generation of company leaders were struggling to keep up against Japanese innovators. The film market gave way to digital cameras, a market in which Kodak was the first player (the digital camera was invented in 1975 by a Kodak employee) but never the leading one. Kodak’s business was film, and the declining interest in film was the downfall of the company. By 2012, Kodak went bankrupt. It emerged as a smaller version of itself in 2013. Its camera business, its soul, was licensed away.
“Kodak’s still around, kind of,” Brian said. At a certain point, his mother and stepfather stopped working there. “I saw that they were selling air filters recently,” he laughed to me. “It was very strange. They had some covid production scam they were doing, too. They had gotten some handout.”
That handout? An ill-fated July 2020 plan to turn Eastman-Kodak into a pharmaceutical component manufacturer, backed with a proposed $765 million loan from Uncle Sam. On just a single day in July 2020 after news of the deal was reported, shares of Kodak were halted twenty times at the New York Stock Exchange. The stock, which had been trading around $2, rallied to as high as $60. It was its best day ever. A few days later, the Development Finance Corporation put the deal on hold. The stock came back to Earth.
What were people thinking? Pouring government money into broken systems and broken cultures doesn’t unbreak them. It’s a lesson that has become central to the new companies in this lane: Palantir, SpaceX, Anduril among them. Cultural excellence matters, and in the race between culture and capital, culture will eventually win. “The erosion of these big companies is that they kind of forget how to do the basics,” said Brian.
✺
From Rochester, Brian went to Cornell University, 90 minutes away through the Finger Lakes of New York. When the Defense Advanced Research Projects Agency created its Grand Challenge – and put up prize money – in the hope of spurring innovation in autonomous vehicle research for military applications, Brian and Matt, who’d worked on other car projects at Cornell, answered the call and formed the team.
He studied operations research, the area of applied mathematics focused on informing decision-making in the business and industrial context.
“I trust his decision making more than I trust my own,” said Trae Stephens, another co-founder of Anduril, whose official title is Executive Chairman. “So even when there are situations in Anduril where I think that something is dumb and I wish we didn’t do it, if Brian says that we’re going to do it, and gives his reasons why, I can’t think of a single occasion where I walked away feeling that I was right.
Brian and Trae met in the Washington office of Palantir Technologies, where Brian had gone in 2007. He was set to graduate Cornell in 2006, but stayed on an extra year to participate in the DARPA Urban Challenge, a version of the Grand Challenge that took place on a California military base, and had requirements like obeying California traffic laws. Brian’s Cornell team modified a 2007 Chevrolet Tahoe, codename “Skynet.” And Skynet finished the course.
Matt Grimm had graduated in 2006, and “went to Booz Allen Hamilton to waste his life,” as Brian put it. “It was terrible.” But the terrible didn’t last, and within a year of Brian joining Palantir, Matt had, too. Brian, Matt, and Trae were all forward deployed engineers, a job position that is synonymous with the culture of that company.
Maybe the most important person that Brian met in the Washington Palantir office is his wife, Kori, who was also an engineer. Their 10th wedding anniversary is later this year.
Different companies though they are, the influence of Palantir on Anduril can’t be overstated. “One of the principles that worked really well at Palantir that we’ve tried to replicate here is to give younger people who are ambitious a lot of autonomy and ownership of what they do,” Brian said. Eventually, Brian became the Director of Engineering at Palantir, leading its technical teams as they built Foundry, its commercial software platform.
In Alex Karp’s recent book, The Technological Republic he wrote, “The technology companies that have come to dominate our lives were in many cases small nations built around a set of ideals that many young people craved: freedom to build, ownership of their success, and a commitment above all to results.”
One of the core ideas in that book, “Engineering Culture” as a superior culture of organizations, is something that is easy to see at Anduril. “What I find really motivates most good engineers is that degree of ownership over the outcome,” said Brian. “And the more you can create that, the more they feel like it is theirs to deliver and they’re on the hook for it.”
An engineering culture can turn individuals into the most empowered versions of themselves, able to accomplish things that on their own they might never be motivated to do. When building a company as ambitious as Anduril from scratch, that really matters.
“It’s shitty work, but it’s your shitty work,” said Brian. “There are all these hard things like showing up late at night, upgrading things, racking servers, and if you described that job to anyone, they would be like, ‘this is a terrible low-level job.’ But if you say, ‘no, no, no, you own this, do whatever you’ve got to do to succeed,’ it’s actually extremely rewarding.”
Just shy of a decade at Palantir, Brian left to become CEO of Anduril. “Karp was very, very gracious with the whole thing,” he said of the CEO of Palantir.
✺
“Trae probably remembers this differently,” Brian told me about the sequence that led to him becoming CEO of the new company. That would be putting it mildly. You see, the only person who wasn’t sure from the beginning that Brian Schimpf would become CEO of Anduril was… Brian Schimpf.
“I was absolutely convinced from day one that Brian needed to be the CEO of the company,” said executive chairman Trae Stephens. “It was clear that he was the guy… there was no alternative.”
Trae had left Palantir in 2014 and became a partner at Founders Fund, the venture capital firm founded by Peter Thiel, another Palantir founder, who is still chairman. Matt Grimm had gone to work at another Thiel investment outfit, Mithril Capital. At Founders Fund, Peter encouraged Trae to find the next generation of national security technology companies. Peter had co-founded Palantir, and been an early investor in Elon Musk’s SpaceX.
As it happened, Founders Fund had been the first institutional backer of Oculus, the VR company founded by Palmer and acquired by Facebook in 2014 for $2 billion. Through the firm’s “F50” gatherings—an unconference of top tech talent—Trae got to know Palmer. They bonded over a shared interest in national security. Before founding Oculus, Palmer had worked on treating PTSD with VR at the University of Southern California.
Trae had been mulling a new type of defense company: one that was software-defined and hardware-enabled, inverting the traditional model of the defense primes. In 2015, he pitched the idea to Palmer. Palmer, true to form, said that he was actually testing ramjets in a drained swimming pool at his Woodside house. He was still committed to working on VR, though, and not ready to leave Facebook. Around this time, Trae also shared the idea with his two closest friends from Palantir—Matt and Brian. Again, neither was ready to jump.
The event that set Anduril in motion more than a year later was Palmer’s political ouster from Facebook in early 2017 (for more on that event, shrouded in confusion for years, I can do no better than Jeremy Stern’s excellent profile of Palmer in Tablet). He called Trae and asked about that defense company idea he had pitched him on. Trae went to Matt, who was ready.
They presented their barebones deck to a handpicked group of top talent from Palantir, SpaceX, Tesla, and Oculus. Nearly all of the company’s first dozen employees were in that room—except Brian, who was absent. The co-founding team at that point was Trae, Palmer, Matt, and Joe Chen (Oculus’s first hire and Palmer’s close friend). After that meeting, Trae recalls saying to Brian, “you are the CEO of this company. You don’t know it yet, but you need to be the CEO of this company.” Over the course of the next week, over cigars at Trae’s house in San Francisco late at night, meals with Palmer up and down the Peninsula, the group made progress.
Brian’s friends were skeptical. He had a great job at Palantir, the company was on the rise. Why leave? He answered them, and me: “It is one of the most ambitious ideas you can have, and with one of the best founding teams that you’re going to ever have with the best funding story you’re ever going to have, it’s like, what better version could you ever concoct to go do this? There’s no bigger swing.”
✺
A few miles from Orange County’s John Wayne Airport, just past an Ikea store and a Hooters restaurant, where Harbor Blvd meets Anduril (Anduril, that’s the name of the street), there’s a gated entrance to the campus of Anduril Industries. Its R&D building, clad in dark concrete and secrecy, looms over the intersection. The company has constructed a five-story parking garage, in which I parked. Nearly 3,000 people work out of the Costa Mesa campus nowadays, out of over 4,500 Anduril employees. For a sense of the company’s growth, that figure was under 500 just four years ago. Anduril plans for its Columbus, Ohio factory Arsenal-1 to employ 4,000 or more. Last year, the company raised $1.5 billion at a $14 billion valuation. In February, just a few days before my visit, it signed terms to raise $2.5 billion at a post-money valuation of $30.5 billion.
When Jason Levin, Senior Vice President of Engineering, joined Anduril, Sentry was the only hardware product. “Brian was effectively the chief engineer of that as the CEO,” he told me. In fact, Brian wrote much of the software for both Sentry, Anduril’s autonomous surveillance tower, and Ghost, its unmanned helicopter. As of September, over 300 Sentry towers have been deployed on the US Southern Border to assist the Border Patrol with operational awareness. They’ve since worked on maritime, mobile, extended-range, and cold-weather versions of Sentry for applications far beyond the border.
In his first week on the job, he and Brian drove to Apple Valley, in the high desert north of Los Angeles, for Sentry testing, which Brian led himself.
“I remember Brian and me trying to get these ATVs up the hill in the mud and it being a pain,” he reported of his first week at Anduril. “But I think in a way it set the culture that the CEO was not above driving an ATV in the mud to get to the tower to do the tests. We hold to that same idea today. Nothing’s above or below anybody.”
“Brian is very much a product person, and he instilled that in the culture,” Levin explained. “I come from the traditional defense world. It’s just not the culture there. To build a product, you have to really know the problem.”
“One of the things Brian’s taught me is the concept of product led growth,” said Gokul Subramanian, SVP of Space and Engineering, who along with Levin is responsible for a sizable portion of the engineering enterprise. “A traditional defense prime designs and builds to order. What I mean by that is, they go sell a program. They’re very sales focused. They sell a program, and then they just turn around and tell the engineering team to build what they sold.”
Subramanian leads the “Software Platform Organization” – the Lattice group, basically. Over three-hundred engineers work on Lattice, which is the system of APIs that enables autonomy across every Anduril product. The product-first philosophy manifests most clearly in Lattice. More than just software, it represents Schimpf’s vision of how modern military systems should work.
To deal with the difficult communication environment of the Southern Border, Brian engineered and patented a “mesh networking” system. The result is that if even one Sentry tower remains connected to outside Internet, every Sentry can remain connected via the mesh – to monitor them and deliver software updates. What started as a solution for border surveillance has become a cornerstone technology, now deployed in multiple Anduril products, including its airplane Fury and its submarines.
By building autonomy into disparate products via Lattice, Brian wagered that the software platform, not a sales process, could be the primary driver of what gets built. In the process, they’d invert the sales process, too. “Instead of a customer telling you what to do, you’re telling the customer what to do because you’re putting software in front of them that totally changes their opinion of where the hardware should go, or how the airplane should be built,” said Subramanian.
At Anduril, everything hinges around Lattice and autonomy, and Brian never lets anyone forget it. Imagine, for a moment, a graph of cost and quantity. The typical defense program has become high-cost, low quantity. “What Anduril is trying to do is to move the entire Defense Department to the other corner of that graph,” said Subramanian, “to get a high number of things for an extremely low cost. And if you understand that, you will understand that is what connects every one of our businesses together. The only way to get a high number of things for low cost is if you have autonomy. Without autonomy, you have to have huge numbers of people manning these systems. So that’s why Brian is so focused on autonomy, because it’s the only way to push you to that other corner of the graph.”
The case for Anduril is that systems of American defense production are broken, and that the unlimited spigot of money is part of the problem, because it has failed to incent companies correctly. Big, bumbling programs that cost a fortune have been rewarded, at great cost to the country.
Anduril is designed to be a conglomerate of many working parts and programs. Only, in contrast to the defense primes whose programs are often silos, Anduril wants to connect every program together via Lattice.
Gokul recalled walking into Brian’s office a few weeks before our chat. “There was a stack of papers on his desk – not financial projections, that are typical for a CEO to be looking at. It was technical papers across like ten different disciplines. He had a paper on the Link 16 standard that DOD uses for data networks. He had a paper on autonomy that he was reading. I was thumbing through those papers, being like, ‘damn it, the CEO has enough time to read this. Why the hell am I not digging into this stuff?’”
As for what Brian says about his sources of information: “Dude, my number one website is Hacker News still,” he told me. “I love it. It’s my most visited website. It’s crazy. I just addictively check it.”
That addiction to information is one of the reasons he is able to command the respect of thousands of engineers at once. “Brian is not an aerospace engineer who’s going to design the airframe of the Fury,” said Matt Grimm. “And he’s not an explosives engineer who’s going to design the frag pattern on a weapon. But he dives in, and he learns enough to deeply understand the program structure. And that matters for how the program gets shaped and how the sales operation works. And he clearly delegates out to the actual individual engineering leadership and then the engineers below them to actually go execute it.”
✺
“He’s literally zero drama,” said Trae of Brian. “He’s been one of my best friends for sixteen years. He has never once in sixteen years introduced drama into my life. He’s received a lot of drama, but he’s never introduced drama.”
When Trae and Brian lived in Washington, their families moved into the same apartment building. “We wanted someone that lived a few doors down from us that we knew, if anything happened, they would be the first people that would be at the hospital next to us, or take care of our dog, or whatever it was.”
“It’s hard to put into words” – he managed to find the words – “but he’s the type of person that you would want to manage your estate, or to be the father to your kids if something happened. I’ve never met a person, professionally or personally, that is more reliable.
Matt, Brian, and Trae were all seniors in high school when the 9/11 attacks happened. Trae said that he thinks that as a result, that micro-generation is “probably the most civic minded of any age group in America.” And among the first people at Anduril – with the exception of Palmer, who is a decade younger – a majority are in that range.
“As you’re probably aware, the founder situation at Oculus did not go well for Palmer at the end of the story,” Trae said to me. “So, I think for him it was really important to establish trust and loyalty. And he’s been awesome.”
I didn’t interview Palmer Luckey for this story, but he did tell me this about Brian: “Brian takes insane complexity and just makes it work — at scale and speed. Most people have no idea how hard this job is. Brian makes it look easy.”
“Brian takes insane complexity and just makes it work — at scale and speed. Most people have no idea how hard this job is. Brian makes it look easy.” – Palmer Luckey
Here’s Trae again: “The thing that is least well understood publicly right now is that there is literally zero tension at the founder level. It just works incredibly well.”
Palmer Luckey has no direct reports at Anduril, it turns out. And that’s, in large part, so that he can focus on the cutting edge of hardware and technology design.
“It’s a great deal. We’re great,” Brian says of the relationship with Palmer. “I’m building the organization that can make all of these things get produced, get the right engineers in the door, get the right talent, get the right strategies in place, and activate all these really talented humans against the problems in the right way. That is what I’m building right now. That’s my project.”
“Brian is the final decision maker on all things. And I think his personality is incredibly well suited to being that executive decision maker, given the way that he processes information and the breadth across the business that he has exposure to.” – Trae Stephens
Matt Grimm put some more color on what he and Brian do at Anduril, compared to Palmer: “Say Palmer wants to train wolves to be soldiers or something. It’s like, ‘okay, cool. Do we want to do this? How much should we budget for it? How big is the team? Where are we going to run this? Where are we going to test this?’”
“Anduril is like an ‘it’s so over, we’re so back’ roller coaster, with very high amplitude on those curves. Brian is very calm in his ability to take the raft of issues and compartmentalize them into ‘not worried about this’, ‘going to focus here,’ and ‘not worried about that’ and maintain an even keel.” – Matt Grimm, COO
“Organizations are an engineering problem,” he says. And one particular organization is his engineering problem.
✺
To finish our talk, I asked Brian what he’d be working on today if he weren’t in charge of the whole organization. “Autonomy is a really hard, really awesome problem and I know I could crush it, and it’d be so much fun,” he said. “It’d be more entertaining if I got to work on the actual technology, but that wouldn’t work today,” he lamented to me.
The Anduril campus was buzzing with life on the Friday afternoon I visited. Machines whirred in the halls of its hardware areas. In the principal campus building, which used to be a newspaper printing plant, the sound of keyboards clacking could be heard as you weave from room to room. In a large atrium, an American flag hangs. Below is the Anduril showroom, where models of its public, announced products are displayed.
For a company as young as Anduril, it’s a lot of stuff. A tower, a helicopter, a drone interceptor, multiple drones, multiple submarines, a fighter jet, and more
“He’s something of a zen master and he’s very calm, which is very important when you run a big company. We have 16 or 17 divisions now, and on any given day some of those divisions are working well and some are not.”
“I think one of the best decisions we’ve made was not being in Silicon Valley,” Brian says. “The thing I hated about Silicon Valley was that everyone was a mercenary. If you worked at a job for more than three years, people started to ask questions of why you couldn’t get another job. There was no real attachment to what you were creating for the long-term. It was very much just, ‘how do I advance myself and get on the right opportunities and jump on the right bandwagon chasing the IPO?’ It felt bad, and weirdly competitive on the wrong things.”
But as his senior engineering leaders learned from him, the Silicon Valley model of product development really works. Brian himself put it this way: “No one is smart enough to figure out what these products need to be. But I think a lot of what Silicon Valley got right is that the pace of iteration and correction need to be frenetic. Every day you are constantly assessing, what information do I have that changes my strategy, what I need to do?”
We writers have a saying: kill all your darlings. In defense, it might be spun as kill all your programs. Brian: “There should be very little attachment to what you’ve done in the past. The people I’ve seen who are very successful are happy to throw everything out that they’ve already done.”
What’s in a Name?
The name “Kodak” was George Eastman’s invention. “I devised the name myself,” he told a magazine in 1920. “The letter ‘K’ had been a favorite with me—it seems a strong, incisive sort of letter … It became a question of trying out a great number of combinations of letters that made words starting and ending with ‘K’. The word ‘Kodak’ is the result.”
If strong and incisive was the spirit of the word “Kodak,” the spirit of Anduril is its mission, contained in the origin. “Andúril” was also an invented word. In J.R.R. Tolkien’s Quenya language, the language of elves in Middle Earth, it means “Flame of the West.” Anduril was the name given to the sword Narsil when it was reforged and bestowed to Aragorn by the elves thousands of years after its destruction.
Then he drew Anduril and held it up glittering in the sun. ‘You shall not be sheathed again until the last battle is fought,’ he said.
The Return of the King
Narsil was the sword of his ancestor, Elendil. And with the reforged Anduril, he saved the West. As the famous Tolkien couplet prophesied:
Renewed shall be blade that was broken,
The crownless again shall be king.
The founders of and investors in Anduril have made a big bet, and it’s captured in the name: a weapon, reforged and in the right hands, can save the world. For Anduril, those “right hands” are the hands of the American warfighter.
In one instance that Brian remembers well, Anduril had pushed an update to its counter-drone system, which bricked one of the systems. The customer, part of US Special Forces, chastised the team. As Brian recalls the customer saying: “I want every engineer to imagine that there’s a drone flying at you to kill you right now. And every time you commit code, I want you to understand that is literally what is happening, so you need to take that seriously and fix your shit.”
Outfitting the warfighter with the best technology is a serious burden, of which Anduril’s leaders are regularly reminded. “This isn’t like Slack going down for four hours, and now I have to go take a long lunch break,” said Brian. System failure is not an option.
Whether Anduril Industries can reforge the broken sword of American defense is a burden that falls on its leaders, and especially its swordsmith from Rochester – until the last battle is fought.
