Anduril Fury And The Death Of Cost-Plus Defense

In April 2024 the Air Force picked Anduril and General Atomics to build the first increment of the Collaborative Combat Aircraft program — autonomous drone wingmen designed to fly alongside F-35s and the still-vaporware NGAD. Boeing was eliminated. Lockheed was eliminated. Northrop was eliminated. The two finalists were a company founded in 2017 by the Oculus guy and a company that mostly builds Reapers in the desert.

Two years later, Anduril's CCA — designated YFQ-44A Fury — has flown. Increment 2 selection is now in front of the Air Force, the FY2026 budget request asks for production funding, and the program is, by Pentagon standards, on schedule. The schedule itself is the news. The traditional primes are the people who taught the Pentagon what "on schedule" means, and that definition has been a euphemism for two decades.

Here is the structural part. The CCA contract was awarded as a fixed-price development effort with company capital on the line, not the cost-plus framework that has defined every major aircraft program from the F-22 to the KC-46 to the B-21. Cost-plus means the government reimburses the contractor's costs and adds a fee. The contractor's incentive, encoded directly into the math, is to have more costs. The F-35 program cost over $400B to develop and is projected to cost $2T to operate over its lifetime. The KC-46 tanker is eight years late and Boeing has taken over $7B in losses on it because Boeing, exceptionally, agreed to fixed-price and discovered what fixed-price actually means when your supply chain is the same supply chain that produced the F-35's helmet-mounted display.

Anduril's pitch was not patriotism. It was that they would build the airplane the way Tesla builds cars and SpaceX builds rockets — vertically integrated, software-first, manufacturing-aware from day one, and priced as a product instead of a procurement. Their Arsenal-1 factory in Columbus, Ohio is being built to produce 5,000 autonomous systems per year. That is not an aerospace number. That is a consumer electronics number applied to a domain that has, for forty years, treated single-digit production runs as normal.

The fair counter-argument: Anduril has not yet delivered combat-capable aircraft at scale. Fury has flown. Fury has not gone to war. The primes have actually-existing F-35s in actually-existing squadrons, and the YFQ-44A is, today, a flight-test article with a thirty-month timeline to initial operational capability. The bet could still go wrong. Software-defined warfare has a habit of meeting the cratered runway of reality.

But the bet has already done something. It has forced the existing primes to write white papers about "digital engineering" and "model-based systems engineering" and other phrases that are now uttered with the desperation of a tenured professor learning what TikTok is. Lockheed's NGAD bid was reportedly priced at a level that suggested the company finally understood it was no longer the only adult in the room. Boeing has restructured its defense unit twice in the last 18 months.

The lesson is not that Anduril is good. The lesson is that an industry which spent forty years arguing that aerospace is fundamentally different from every other engineering discipline turned out to be wrong about that, and the proof was a 2017 startup, two governors of West Virginia and Ohio fighting over the factory, and one airplane that flew on schedule.

Cost-plus is not a contracting mechanism. It is a worldview. The worldview is dying. The replacement is product.

Until tomorrow.