Procurement is the technology risk
The engineering is tractable. The accreditation boundary is where companies die.
Budgets have moved toward attritable systems, autonomy, and domestic supply. That shift is real, well covered, and not the hard part. The hard part is a sales cycle measured in programme years against a customer that buys on compliance as much as on capability.
Build for the boundary first
Winners design for the accreditation boundary and add capability inside it. Consumer instincts run the other way — ship capability, handle compliance later — which is why most crossover teams stall between prototype and programme of record.
Our case file is unambiguous: one team retrofitted the security and supply-chain boundary after the fact, spent eighteen months and a hardware redesign doing it, and never converted its prototype contract. The competitor that designed for the boundary first shipped less capability and won the re-compete.
Conversions, not announcements
A prototype contract is a marketing event. The number that matters is the conversion rate from prototype to programme of record, and it is low enough that treating announcements as validation will systematically overprice the sector.
We track conversions by company and by programme office, because the rate varies enormously between the two.
Cyclicality is priced badly
Political appetite moves. Primes are structurally advantaged in re-competes. Export control can remove a market inside a quarter — we watched a supplier lose roughly a third of revenue that way, with domestic demand absorbing only part of it.
We now underwrite defense revenue by jurisdiction and by component origin, and discount export share explicitly.
This post sits inside the Defense & resilience stream. The standing note holds the full thesis, bear case, and signal list.