Inside Anduril: The $30.5 Billion AI Weapons Startup

In 2017, Palmer Luckey was 24 years old, had just sold his virtual reality company to Facebook for $2 billion, and had been publicly fired in what many interpreted as a politically motivated termination. Eighteen months later, he co-founded Anduril Industries. By 2024, the company had been valued at $14 billion. By 2025, that figure had more than doubled to $30.5 billion. The defense industry has not seen a disruption like this in decades.

The Founding Moment: From VR to Warfare

Palmer Luckey founded Oculus VR in 2012, at age 19, out of a garage in Long Beach, California. Facebook acquired Oculus in 2014 for approximately $2 billion, making Luckey one of the youngest technology millionaires in Silicon Valley. His departure from Facebook in March 2017 — following reports that he had secretly funded a pro-Donald Trump internet "shitposting" group called Nimble America — became a Silicon Valley controversy. Facebook denied the departure was politically motivated. Most observers concluded otherwise.

Luckey's response was to double down on confrontation. He co-founded Anduril in 2018 alongside Trae Stephens, Brian Schimpf, Matt Grimm, and Joe Chen — a team drawn heavily from Palantir and other technology firms with defense experience. The name was drawn from J.R.R. Tolkien's fictional sword, a choice that said something about Luckey's taste for operatic branding. The company's first product was not a weapons system but a border surveillance technology: autonomous sensor towers deployed on the U.S.-Mexico border under a Customs and Border Protection contract.

That origin story is important context. Anduril was not born as a weapons manufacturer. It was born as an autonomous sensing and software company that happened to work in a national security context. The weapons came later — and they came faster than anyone outside the company expected.

Lattice OS: The Brain of the Machine

Every Anduril product connects to Lattice, the company's proprietary AI operating system and the true technological core of its business. Lattice is not simply software that runs on individual hardware platforms. It is a network-centric operating environment that ingests sensor data from any connected platform — towers, aircraft, submarines, ground vehicles — fuses it into a common operational picture, and enables autonomous decision-making and human command across a distributed force.

Lattice can ingest radar tracks, optical sensor feeds, acoustic data, and signals intelligence from multiple sources simultaneously and present that integrated picture to operators in real time. It can classify and track objects of interest autonomously, assign threat scores, and in certain configurations trigger or recommend autonomous action. It supports what Anduril describes as "human-machine teaming" — a doctrinal concept in which autonomous systems handle the sensor, decision, and engagement cycle at machine speed while human operators set rules of engagement and retain override authority.

The Lattice architecture is Anduril's primary competitive advantage. When the company adds a new hardware platform — a new drone, sensor, or munition — it is not building a standalone product. It is adding a new node to a network that already exists, already has paying customers, and already has integration relationships with U.S. military command systems. The compounding effect of this architecture means that each new product Anduril ships makes every other product more capable.

The Product Suite

Roadrunner-M: The Autonomous Interceptor

Roadrunner-M is a twin-jet autonomous aircraft designed to intercept aerial threats — cruise missiles, drones, and low-altitude targets that traditional air defense systems are not optimized to engage cost-effectively. The system takes off and lands vertically, making it deployable from austere locations without prepared runways. Its "M" designation indicates the munitions variant, which carries a fragmentation warhead and is designed for terminal engagement against incoming threats.

The economics of Roadrunner are a central part of Anduril's pitch to the U.S. military. Intercepting a $200,000 commercial drone with a $3 million Patriot missile creates an unsustainable cost exchange ratio. Roadrunner-M, if produced at scale, aims to shift that calculus — an autonomous interceptor that costs less than the threats it defeats. This was a core argument in Anduril's early discussions with the Army and with DARPA about the future of counter-drone and missile defense.

Ghost: Autonomous VTOL ISR

Ghost is an electric vertical takeoff and landing (VTOL) drone designed for intelligence, surveillance, and reconnaissance missions. It is quiet, compact, and designed to operate autonomously within the Lattice network — collecting and transmitting sensor data with minimal operator workload. Ghost-X, a larger variant, carries heavier sensor payloads and has greater endurance. The platform has been deployed by U.S. Special Operations Command (USSOCOM) and with partner nation forces in multiple operational environments.

Anvil: The Counter-Drone Kinetic Kill Vehicle

Anvil is a kinetic interceptor designed specifically to defeat small unmanned aerial systems through physical collision rather than explosive warhead. It is autonomous: Anvil receives a target track from Lattice, classifies the threat, and executes an intercept without requiring individual operator inputs for each engagement. This speed is critical against drone swarms, where the volume of simultaneous threats can overwhelm human-in-the-loop engagement rates.

Sentry Tower: Autonomous Border and Perimeter Surveillance

The Sentry Tower was Anduril's original product and remains one of its most widely deployed. A modular autonomous surveillance system integrating radar, optical sensors, and AI-powered classification, Sentry Towers form sensor networks across borders, military installations, and critical infrastructure. The system can detect, identify, and track ground and aerial targets autonomously, alerting human operators to threats of interest. Thousands of Sentry units have been deployed through CBP contracts along the U.S.-Mexico border and at military installations across the United States and allied nations.

Dive-LD: Autonomous Underwater Vehicle

Dive-LD is a large-diameter autonomous undersea vehicle (AUV) designed for long-endurance underwater operations. It won a contract from the U.S. Navy as part of the Overlord program for large unmanned underwater vehicles — a competition Anduril won against Boeing, among others. Dive-LD can conduct intelligence, surveillance, and reconnaissance missions, mine countermeasures, and payload delivery operations over extended durations without human operator involvement. Its selection over Boeing for a major Navy program was one of the clearest signals that the traditional defense primes were vulnerable to new entrants in autonomous systems categories.

The Funding Architecture: Silicon Valley Bets on Defense

Anduril's investor roster reads like a Silicon Valley power registry. Founders Fund — Peter Thiel's venture capital firm — led early rounds. Andreessen Horowitz (a16z) has participated in multiple funding rounds. Valor Equity Partners and other strategic investors round out the cap table. As of Anduril's 2025 Series G round, which valued the company at $30.5 billion, the total equity raised exceeded $4.5 billion.

The willingness of these investors to back a defense company reflects a genuine ideological shift in Silicon Valley that Luckey himself claims credit for accelerating. For much of the 2010s, major technology firms — including Google — faced internal employee revolts over defense contracts. Google's Project Maven, an AI initiative for the Pentagon that involved analyzing drone surveillance footage, was cancelled in 2018 after more than 3,000 employees signed a petition. That era appears to have ended. The war in Ukraine, the rise of China's military technology ambitions, and a generational shift in Silicon Valley's political outlook have made defense investment broadly acceptable — even fashionable — in a way it was not a decade ago.

Luckey has been vocal about his role in this shift, arguing publicly and in congressional testimony that American technology talent and capital have an obligation to apply themselves to national defense, and that the previous generation of Silicon Valley's near-total avoidance of defense work represented a strategic mistake. Whether or not one agrees with the political framing, the capital flows confirm the trend: Shield AI raised at a $5.6 billion valuation; Palantir crossed a $160 billion market capitalization; Joby Aviation, Joby, and a dozen other dual-use technology firms are now in defense contracting in ways that would have been unthinkable in 2015.

The Neoprime Strategy: Disrupting Lockheed and Raytheon

The traditional defense prime contractor model — Lockheed Martin, Raytheon, Boeing, Northrop Grumman, General Dynamics — is built around large, complex, long-cycle programs. A fighter jet program runs 20 years from contract award to initial operational capability. Cost overruns are structurally embedded. Software updates move on hardware timescales. The primes have immense lobbying infrastructure, deep government relationships, and massive physical manufacturing capacity. They also move very slowly, charge very high margins, and have limited ability to iterate software.

Anduril has explicitly positioned itself as a "neoprime" — a next-generation prime contractor that competes on software iteration speed, autonomous systems capability, and cost-per-effect economics rather than on traditional platform scale. This means competing for contracts that the primes would prefer to avoid (complex software-intensive autonomous systems with uncertain requirements) and offering capabilities the primes cannot replicate quickly (Lattice integration, AI autonomy, rapid deployment).

The strategy has shown real results. The Dive-LD Navy UUV contract win over Boeing was a neoprime moment. Anduril's selection for participation in the Collaborative Combat Aircraft (CCA) program — a U.S. Air Force initiative to develop AI-enabled autonomous wingmen for crewed fighter aircraft — alongside General Atomics over several traditional primes was another. The company also won a significant contract in 2024 to supply counter-drone systems for the U.S. Army's Short Range Air Defense (SHORAD) program.

Arsenal-1: The Manufacturing Question

One of the most significant structural differences between a software-heritage company and a traditional defense prime is manufacturing scale. Anduril addressed this directly with its announcement of Arsenal-1, a dedicated advanced manufacturing facility planned at over 5 million square feet in Columbus, Ohio. The facility, announced in 2024 with backing from a $650 million Ohio state economic development package, is designed to produce Anduril hardware at the volumes required by major military contracts.

Arsenal-1 reflects a fundamental insight about the future of defense manufacturing: the next major conflicts will require munitions, drones, and autonomous systems at volumes that existing defense supply chains cannot produce. The experience of the Ukraine conflict — where consumption rates for artillery shells, missiles, and drones repeatedly outstripped Western stockpiles — has concentrated minds in the Pentagon and in Congress on the industrial base problem. Anduril is positioning Arsenal-1 as part of the answer to that problem, which is also a way of positioning itself as infrastructure-critical to U.S. defense, rather than simply a product vendor.

Key Contracts and Revenue Trajectory

Anduril's revenue trajectory has been steep. The company reported an annual revenue run rate exceeding $1.5 billion in 2025, up from approximately $500 million in 2023. The growth is driven by a combination of expanding existing contracts — particularly the CBP Sentry Tower program, which has grown consistently — and major new program wins. USSOCOM is one of Anduril's most significant customers, using Ghost drones and Lattice across Special Operations missions. The Army SFA (Security Force Assistance) brigade programs have used Anduril systems for partner force training and deployment support.

Internationally, Anduril has secured contracts with the United Kingdom (for the Integrated Air and Missile Defense program), Australia (as part of the AUKUS defense technology partnership), and has active partnership discussions with several NATO allies. The international expansion is strategically important: as U.S. defense procurement cycles encounter political headwinds or budget pressure, allied nation contracts provide revenue diversification and, potentially, manufacturing economy of scale.

Competitive Landscape

Anduril's most direct technology-sector competitor is Shield AI, a San Diego-based autonomous systems company valued at $5.6 billion as of its 2024 Series F round. Shield AI's primary product is Hivemind, an AI pilot technology that enables unmanned aircraft to fly and fight autonomously without GPS or communications links — a critical capability in electronic warfare environments. Shield AI acquired Heron Systems in 2021, gaining an AI dogfighting capability that defeated a human F-16 pilot in a DARPA AlphaDogfight Trials simulation. The two companies occupy overlapping but not identical market positions: Anduril competes on the Lattice network ecosystem; Shield AI competes on the autonomous pilot software layer.

Palantir, valued at over $160 billion in early 2025, is a different kind of competitor — one focused primarily on data analytics and AI-assisted decision software for government and military customers rather than on hardware platforms. But Palantir's AI Platform (AIP) for defense, its Maven Smart System contract with the U.S. Army, and its Gotham platform for intelligence analysis put it in competition with Anduril in the command-and-control software space that Lattice also occupies.

The traditional primes — Lockheed, Raytheon, and Boeing — are not standing still. All three have accelerated autonomous systems development programs, acquired or invested in AI companies, and are competing for the same next-generation contracts. But their organizational structures, culture, and cost models remain significant handicaps in a market where iteration speed is increasingly the differentiating variable.

Why This Matters Beyond the Balance Sheet

Anduril's commercial success is secondary to its strategic significance. What the company represents is the integration of Silicon Valley's software methodology — rapid iteration, cloud architecture, machine learning at scale — with the hardware and procurement infrastructure of military contracting. That integration is changing the defense industrial base faster than any single procurement decision or policy initiative could.

When Anduril wins a contract against Boeing, it is not just winning revenue. It is establishing that a company with six years of history, no legacy mainframe platforms, and no decades-long government relationships can outcompete the most established defense prime in the world on a next-generation autonomous systems program. That precedent has downstream effects on every defense procurement decision that follows, on every venture capital firm deciding whether to fund the next Anduril, and on every engineering student choosing between a Silicon Valley startup and a Lockheed Martin rotation program.

The defense industry that emerges from this decade will look fundamentally different from the one that entered it. Anduril is not the only reason for that — but it may be the clearest single symbol of the transformation.