It’s an overcast 60-degree spring day at Anduril Industries’ test range in the drought-parched hills of Southern California, but while the weather is a tad chilly for humans, it’s perfect for surveillance. “The seeing” is good, explains Palmer Luckey, Anduril’s billionaire founder, who made his first fortune selling his virtual reality startup, Oculus VR, to Facebook for $2 billion in 2014. Cool temperatures mean little thermal distortion, which makes it easier for Anduril’s sentry towers to spot immigrants on the U.S.-Mexico border.

The self-taught tech prodigy—who will turn 30 in September and transition from whiz kid to “whiz man,” he jokes—cues his engineers to demonstrate why investors are set to pour another $1 billion into Anduril, for a total of $1.8 billion raised since 2017. (Luckey owns at least 11% of the company, which, when added to his Facebook windfall, brings his current net worth to an estimated $1.4 billion. That will rise after the latest round closes in June, which is expected to swell Anduril’s valuation by 70%, to $8 billion.)

A ground-based infrared sensor spots a pickup truck on a road in the test area, triggering a camera on a mast to focus on it. An AI program called Lattice, which Anduril considers its core technology, highlights the truck and identifies it as a vehicle. With the truck set to disappear behind a hill, a stealthy black helicopter drone called Ghost shoots out to keep it in view. Onscreen, the truck stops—a man exits and launches a drone. A radio-frequency sensor picks up a signal from the drone, revealing it’s a Chinese-made DJI P4. Lattice instantly labels the man and the drone as “suspicious.” To neutralize the drone, a metal box pops open and a burly quadcopter called Anvil takes off at startling speed. It has one job: to crash into intruding drones, knocking them out of the sky.

“We can torch the batteries, torch the airframe and just go ridiculously fast,” says Luckey, who despite the weather is wearing his trademark uniform of flip-flops, cargo shorts and a Hawaiian shirt. “There’s definitely a homefield advantage when it comes to drone-versus-drone combat.”

It’s a brute, direct approach, in keeping with Luckey’s current incarnation. Eight years ago he won reams of adoring press (including a Forbes cover story) as a puppy-like teen wunderkind pioneering virtual reality. But three years after he sold out to Mark Zuckerberg, he was fired by Facebook amid furor over his support for Donald Trump in the 2016 presidential election.

His swift move to found a defense startup—partnering with libertarian billionaire Peter Thiel’s Founders Fund and executives drawn from Thiel’s spooky spy software maker Palantir—completed Luckey’s spiritual exit from left-leaning Silicon Valley. He departed physically, too. Anduril is headquartered in Costa Mesa, closer to San Diego’s military bases than the center of the metaverse in Menlo Park.

Anvil, seen here exiting its launch box, autonomously stalks intruding drones and bludgeons them out of the sky. Luckey doubts the high-tech approaches competitors are taking to drone interdiction: Copper foil can protect sensitive electronics from microwave weapons; comm links can be protected by encryption; spinning a drone fast can prevent a laser from torching it. “You can't really make a drone survive getting hit by another drone going 200 mph."

After losing friends who criticized him as a warmonger, Luckey is suddenly feeling vindicated. In the wake of Russia’s invasion of Ukraine, where Anduril has systems on the ground (he won’t say what precisely), some people are reaching out to apologize. They now realize “it is actually really important for the U.S. to have better weapons,” he says.

How he goes about building those weapons is key to understanding what sets his company apart. Anduril, which generated an estimated $150 million in revenue last year, develops much of its technology on its own dime, a high-stakes wager that inverts how military contractors normally operate. Rather than waiting for the Department of Defense to launch a multiyear process of defining tech requirements and inviting bids to develop prototypes, Anduril plows ahead and makes weapon and surveillance systems that it believes the government will want—as long as they work.

“We want to be the company that when the DoD needs something, we're the first people they think of,” says Luckey.

Two previously undisclosed projects: a fast armed drone that Luckey says is intended, in some cases, to replace manned fighters in the job of scrambling to intercept air-zone violators. Then there’s a large surveillance drone—shown on the condition that it not be described in detail—designed to launch and land vertically (making it runway-independent) and to fly long distances autonomously, making it seemingly suited for the vast reaches of the Pacific. There’s no formal Pentagon request for either.

Marrying Luckey’s past and present, there’s also a complex simulation tool that melds Lattice with a video game engine from Carbon Games, a studio Anduril acquired in 2019. It’s intended to let DoD run thousands of rapid “what if” scenarios on how conflicts could play out, viewable both with VR goggles and on regular screens. It also helps Anduril decide what hardware to build next.

Luckey is confident his tech is superior to that of the defense giants, but Anduril isn’t taking any chances. According to federal disclosures, 39 lobbyists worked to push the scales in Washington for Anduril in 2021. It hired a strong team of D.C. and DoD veterans from the start, headlined by Christian Brose, a former chief staffer on the Senate Armed Services Committee. They see their cause as virtuous, pushing to speed up the sclerotic acquisitions process and to do away with proposal-based competitions in favor of more trials and bake-offs.

Luckey personally spends a lot of time in Washington. “People want to believe that if you build the best thing, then you’ll win. That’s not the way that the real world works,” he says.

Home-schooled by his mother in Long Beach, California, Luckey got his first lessons in engineering working on cars with his dad. Eventually, he took over half the garage, advancing to building things like high-powered lasers and coil guns, which fire high-speed projectiles using electromagnets. In his mid-teens he took to upgrading old game consoles with miniaturized electronics to make them portable. Games led to virtual reality: He started collecting bulky old VR headsets (the underlying technology dates to the 1960s) and tinkering with them. His breakthrough was realizing that he could replace their expensive, heavy optics with cheap, light ones if he used software to manipulate the images. Thus was born Oculus Rift, the VR headset Luckey created when he was just 16—and the one that attracted the attention of Mark Zuckerberg.

Luckey sits in a building in a complex that used to house the Los Angeles Times’ Orange County offices and printing plant that’s being renovated to serve as Anduril’s headquarters at a total cost of $180 million. The new HQ will accommodate around 2,400 employees; Anduril has just under 1,000 now, up sharply from 230 two years ago.

While Luckey was at Facebook, he tinkered with even wilder things—building a ramjet engine in his swimming pool, (unsuccessfully) trying to make rocket boots—and he started talking to Trae Stephens, a partner at Founders Fund, about ideas for defense startups.

Peter Thiel had tasked Stephens with finding the next Palantir or SpaceX to tap the government’s deep coffers. Drawing a blank, Stephens, a Palantir veteran, was encouraged to build the type of company he thought would work from scratch. Stephens and Luckey agreed that DoD’s greatest weakness was software—the brass still treated it as just an add-on to big weapons systems. But Luckey wasn’t interested in doing anything about it himself—until Facebook dumped him.

Stephens recruited his best friends from Palantir: Brian Schimpf to helm software and serve as CEO, and Matt Grimm to run operations. But after rocky experiences selling software to the Pentagon while at Palantir, they made a plan to slip AI into DoD in a hardware pill: futuristic weapons based on cutting-edge software. Luckey, Stephens hoped, would command respect in D.C.—and be their hardware guy too.

“Smart people, especially engineers, want to be around Palmer because he’s electric,” says Stephens, who’s chairman of the company. Luckey’s creativity is chaotic, he adds, but his fellow founders are there as bumper guards. “He is, when channeled appropriately, unstoppable.”

“Smart people, especially engineers, want to be around Palmer because he’s electric."

In 2017, the newly formed Anduril sold Customs & Border Patrol on trying out its first proof of concept: sentry towers that automatically detect people and vehicles illegally crossing the border, freeing agents from many routine patrols. In 2020, the agency gave Anduril a contract worth up to $250 million; by February, CBP had 176 towers deployed on the Mexican border.

In January, Anduril scored its biggest validation yet: a contract to take charge of U.S. Special Operations Command’s drone defenses that could be worth almost $1 billion over 10 years. An even bigger opportunity is on the horizon. The Pentagon is eager to knit together all its surveillance and weapons systems to create a unified view of the battlefield and orchestrate them from afar, all the while resisting hacking and jamming. The program is called Joint All Domain Command and Control, or JADC2, and Anduril and others—including Palantir and Redwood City, California–based C3 AI—are jockeying for tens of billions of dollars in potential spending.

Anduril is hopeful that its Lattice software system can pull it off. At a 2020 Air Force trial, Anduril fused radar with its sensor towers to detect incoming cruise missiles and automatically route targeting data to multiple weapons systems, including an F-16 and a Paladin howitzer, to take them out. Remarkably, the system required the oversight of just a single airman.

“They’re definitely at the top,” says Nicolas Chaillan, who until last September was the Air Force’s chief software officer. Chaillan, who questioned the Joint Chiefs’ commitment to JADC2 in a public resignation letter, warns that the project could be doomed by siloed development efforts that might not ultimately mesh.

If it collapses like other failed modernization efforts, Luckey professes to be unconcerned. After all, in addition to its ongoing government contracts, Anduril now has a war chest fat with VC funds. The Pentagon doesn’t need to fret over Palmer Luckey, he says; it needs to fret over finding the next Palmer Luckey. “They need to be worried about how to make people like I was when I was 19—with good tech and good ideas—into successful vendors. Because right now there’s just no path at all.”