A group of roughly forty engineers are developing a drone swarm system in an unremarkable office park outside of Austin, Texas, which the U.S. Air Force will assess in less than a year. There is no fleet of black SUVs in the parking lot, no marble lobby, and no Pentagon contractor credential showing over the elevator. With a kitchen full of energy drinks, a few foosball tables that are never used, and whiteboards covered with system architectural diagrams, the office resembles a Series B software firm.
However, the timeframe is different from what Lockheed Martin or Boeing normally follow, and the contract on the wall is true. From concept to testable prototype, it takes twelve months. That’s the situation. If you miss it, the government will move on.
From the inside, the defense contracting boom appears to be a true structural change in how the Pentagon purchases capacity rather than a reorganization of the same old primes. Over the past few years, the Defense Innovation Unit, which operates out of offices in Mountain View and Washington, has worked to create a direct pipeline between commercial technology companies and Department of Defense contracts, purposefully avoiding the procurement bureaucracy that has traditionally prevented smaller businesses from entering the market.
The Air Force’s innovation division, AFWERX, funds dual-use technology from local entrepreneurs. Innovation in Small Businesses Engineering teams that might otherwise be creating consumer goods are seeded by research mechanisms to create autonomous surveillance systems or AI-assisted targeting algorithms for military use. The infrastructure for a new type of defense contractor has been purposefully built, and it is attracting precisely the talent that the traditional primes have had difficulty luring.
Five years ago, software engineers and machine learning researchers would have gone to Google or Meta, but today they end up at Anduril, Palantir, and Shield AI, or at tiny regional engineering firms that are unknown outside of a certain procurement office. Patriotism is part of the pitch, but it is not the main attraction. It’s because the work is technically fascinating in ways that most commercial IT jobs aren’t, the security clearance pays well, and the drive to meet deadlines fosters the kind of problem-solving atmosphere that draws engineers who want to design products that are actually deployed.
Anduril has developed counter-drone technology and autonomous sentry systems that are currently in operation. Military troops use Palantir’s defense data systems to process information in the field. Autonomous AI pilots for aircraft are being tested by Shield AI. On a roadmap, these are not hypothetical capabilities. These are fielded goods.
It would be a mistake to undervalue the legacy primes because they are not going away quietly. The F-35 is still manufactured by Lockheed Martin. The submarines are still manufactured by General Dynamics. These are not issues with software. Traditional contractors have spent centuries accumulating the physical engineering needed to build platforms that function in harsh conditions, at scale, across decades of service life.
Rather of displacing the primes, it’s feasible that the next generation of defense tech companies will find its niche—agile, software-defined, quick to deploy. What happens when the line between software and hardware becomes increasingly hazy and autonomous systems begin to need the kind of physical manufacturing infrastructure that local engineering firms do not yet possess is the more intriguing topic.
Observing this change from the outside, there is a sense that the Pentagon’s present excitement for non-traditional contractors is sincere, but there is also a short institutional memory when it comes to procurement cycles. SBIR financing and DIU partnerships are motivated by a genuine sense of urgency stemming from specific worries about adversary capabilities in autonomous systems and AI-enabled warfare.
The next five years will determine whether that urgency endures budget cycles, administrative changes, and the unavoidable friction of turning good prototypes into mass-produced military weapons. The engineers in the office park in Austin appear self-assured. Generally speaking, they do, at least up until the contract review.
