THE LARGEST single contract ever awarded to a venture-backed defense firm dropped on a Friday evening, as such things tend to do. On March 13th the United States Army announced a deal worth as much as $20 billion with Anduril Industries — a ten-year agreement covering software, hardware, and services spanning everything from autonomous drones to command-and-control platforms. The timing was quiet; the implications are not.
Anduril, founded in 2017 by Palmer Luckey (the erstwhile wunderkind behind the Oculus VR headset), has assembled a remarkable run of Pentagon wins in short order. The company assumed Microsoft's troubled $22 billion IVAS augmented-reality contract in early 2025. It won a spot competing for Golden Dome space-based interceptors alongside Lockheed Martin and Northrop Grumman. It opened a solid-rocket-motor factory in Mississippi, becoming the nation's third supplier. Revenue reportedly doubled to roughly $2 billion in 2025, and a new funding round led by Andreessen Horowitz and Thrive Capital is expected to value the company at around $60 billion — up from $30.5 billion just nine months prior. For a firm that ranked 74th on the top-100 defense contractors list in fiscal year 2024, the trajectory is, to borrow the Pentagon's own preferred term, formidable.
The new contract establishes what the Army calls an "enterprise" procurement vehicle: a five-year base period with a five-year option, designed to eliminate the thicket of subcontractor markups and compress delivery timelines. Gabe Chiulli, the Defense Department's chief technology officer, framed the deal in language that would sound more natural in a Silicon Valley pitch deck than a Pentagon briefing room: the modern battlefield, he said, is "increasingly defined by software."
Full metal packet
Yet a $20 billion ceiling is not $20 billion in revenue. Enterprise contracts of this kind set an upper bound; actual spending depends on task orders that may or may not materialize over the decade. The Pentagon has a long history of splurging on ambitious framework agreements only to see budgets shift, administrations change, and priorities drift. Lockheed Martin's F-35 program, for context, carries a lifetime cost estimated in the trillions — but its procurement timeline has stretched across more than two decades of congressional wrangling. Anduril's contract is an invitation to sell, not a purchase order.
There is also the question of whether Anduril can scale to match its ambitions. The company is building Arsenal-1, a five-million-square-foot autonomous weapons manufacturing complex in Columbus, Ohio, with production slated to begin in July 2026 — a facility it describes as a defense-sector gigafactory. It is simultaneously ramping solid-rocket-motor output to 6,000 units per year, integrating acquisitions (radar firm Numerica, infrared camera maker AIRS, engineering outfit Klas), and staffing up across multiple continents. Revenue of $2 billion is impressive growth; Lockheed Martin generates roughly $75 billion annually. The gap between hype cycle and production cycle in defense is measured not in quarters but in administrations.
The incumbents, for their part, are not standing still. Lockheed recently announced plans to triple PAC-3 missile production. Northrop Grumman continues to win satellite contracts worth billions. Defense Secretary Pete Hegseth has promised longer, larger contracts to firms that invest in capacity — a signal aimed at both upstarts and stalwarts. The Pentagon reckons it needs both the speed of venture-backed innovation and the scale of the traditional primes; whether a single company can embody both remains an open question.
What the Anduril contract does clarify is the Pentagon's willingness to treat software-defined systems as a procurement category unto themselves — not a bolted-on afterthought to hardware programs. For the primes, that represents a structural challenge: their business models were honed over decades of building exquisite, expensive platforms on long timelines, not shipping code updates in 18 hours (as Anduril claims to do with IVAS). For Anduril, the challenge is the inverse — proving it can deliver at institutional scale without becoming the kind of lumbering bureaucracy it set out to replace.
The contract's ten-year span will carry through at least one more presidential administration, several budget cycles, and whatever geopolitical crises 2026-2036 may produce. Palmer Luckey built his first fortune selling the world a virtual reality headset. His second depends on making the Pentagon's reality a good deal more efficient — which is to say, the harder trick by far. ■
For more, join 75,000 subscribers getting tech's favorite brief here
