The Pentagon's Munitions Acceleration Council has a list. Fourteen weapons — PAC-3 interceptors, Tomahawks, long-range anti-ship missiles, hypersonic strike weapons — identified as critical for 2027. The MAC has inked framework deals with manufacturers to ramp production. It has threatened penalties for companies that miss agreed ramp rates. Jules "Jay" Hurst, performing the duties of Pentagon comptroller, put it plainly: "We're making them put skin in the game."
That's the right instinct. The problem is that skin in the game doesn't conjure factory floor space, skilled machinists, or specialty materials out of thin air.
The Bottleneck Isn't Ambition — It's Physics
The defense tech conversation in 2026 is dominated by software: autonomous systems, AI-enabled ISR, Hivemind autonomy stacks running on V-BAT drones. That's where the venture capital is flowing, where Emil Michael is meeting with startups and dreaming of rivals to Lockheed and Northrop. And the software story is real — Shield AI's recent $800 million Navy ISR contract is evidence that new entrants can win serious programs.
But software scales at the speed of a deployment pipeline. Hardware scales at the speed of a factory.
A March analysis in Breaking Defense by John Ferrari and Dillon Prochnicki made the structural case bluntly: the Pentagon has spent decades designing weapons systems without asking whether the industrial base can actually manufacture them at the required scale. Requirements writers are rewarded for performance, not manufacturability. The result is systems that are exquisite, expensive, and slow to produce — fine for a post-Cold War world of short, limited conflicts, catastrophic for a world where Ukraine and the Middle East have demonstrated that wars consume weapons at rates that outpace American production timelines.
The strategic implication Ferrari and Prochnicki draw is uncomfortable: when adversaries like Iran watch American stockpile depletion rates and production timelines, the message they receive is that the United States can be outlasted. That's not a software problem. That's a factory problem.
The New Requirements Architecture Still Has the Old Blind Spot
The Hegseth-Feinberg memo last November tried to fix the front end of this — disbanding JCIDS, devolving requirements validation to the services, creating new structures to move faster. The MAC's 14-weapon priority list is the back-end complement: identify what matters, fund it, hold manufacturers accountable.
What neither reform fully addresses is the middle: the industrial base capacity question that has to be answered before a requirement is written, not after a contract is signed. You can streamline acquisition all you want, but if the supply chain for solid rocket motors or specialty electronics is a single-point-of-failure, the ramp rate ceiling is set by geology, not policy.
Lockheed Martin's answer to this is its new Rapid Fielding Center, opened in late March — a facility designed to compress the path from prototype to low-rate production. The pitch is that integrating design and manufacturing floor proximity reduces iteration time and cost. It's a real improvement over the traditional model. It's also a legacy prime adapting at the margins of a structural problem, not solving it.
What the Startup Moment Misses
The Emil Michael vision — incubate a fleet of new companies, remake the whole architecture — is genuinely exciting if you believe, as I do, that the old primes have been insulated from competitive pressure for too long. But the startups winning contracts right now are mostly winning on software and autonomy. The manufacturing constraint doesn't disappear because a new company is writing better code.
The defense tech companies that will actually matter at scale are the ones figuring out how to build hardware faster and cheaper — not just design it smarter. That means vertical integration, domestic supply chain investment, and manufacturing processes that can surge. It means asking, before the requirement is finalized, whether the factory exists to fulfill it.
The MAC's penalty clauses for missed ramp rates are a start. Watch whether those penalties actually get enforced when a prime misses a 2027 production target — that's the real test of whether this administration's procurement reform has teeth, or whether it's just a better-organized version of the same structural avoidance.