At defense tech meetups in Arlington or El Segundo, you come across a certain type of founder: sleeves rolled up, a little preoccupied, three coffees deep. Usually, they’ve created something truly remarkable, like a drone autonomy stack, an inexpensive interceptor, or an edge-AI module that can be thrown. Additionally, they have typically been “almost there” with the Pentagon for four years or so. Almost all of the story lies in that qualifier.
It’s not that the Department of Defense doesn’t want their technology; rather, it’s the defense startup paradox. It frequently does. The paradox is that, by the time a contract passes the final review, the company that pitched it may no longer exist because the system intended to purchase it operates on a timeline that is so disconnected from how startups actually operate. In a few months, China will be able to field a new military capability. It usually takes ten years in the United States. Reps. In their joint article earlier this year, Mike Rogers and Adam Smith stated unequivocally that the acquisition process “incentivizes compliance over capability, process over speed and certitude over innovation.”
Much of this stems from a 1962 law that most Americans are unaware of. The purpose of the Truth in Negotiations Act, or TINA, was to prevent defense contractors from raising prices. Reasonable enough. However, more than 60 years later, the requirement that contractors spend more than $2 million provide certified cost and pricing data has turned into a silent deterrent for the very companies that the Pentagon now most needs. The cost-certification process is harsh for software companies, AI labs, and advanced materials startups, which are businesses whose value is found in R&D and intellectual property rather than physical components. Some just glance and move on. Some dedicate a full year to simply learning the vocabulary.
The structural image is depressing. During the Cold War, there were 51 prime defense contractors; today, there are only about five or six. That consolidation was the inevitable result of a winner-take-all bidding system that prioritizes survival and scale over agility. Primes outsource parts to networks of subcontractors that are so interconnected that supply chains are unable to expand when necessary. From 2014 to 2024, there were twenty-four cost overruns that needed to be reported to Congress. Even after receiving initial funding ten years ago, the F-47 fighter is still not in service. Both the Sentinel ICBM and the Columbia-class submarine, which are meant to support America’s nuclear deterrent, are overbudget and delayed.
The phrase “the Valley of Death,” which has become somewhat grim shorthand, can be heard in the lobby of any Andreessen Horowitz portfolio meeting these days. It’s the difference between an actual program of record and an early Pentagon prototype contract. After a startup receives a small SBIR or OTA award, develops a functional version, and shows it to an enthusiastic colonel, nothing happens for eighteen months as the program waits for a budget cycle, a rewrite of the requirements, or an OSD representative to determine whose portfolio it belongs in. The engineers have already departed, the runway has burned, and a prime that knew how to wait has silently absorbed the technology by the time the response comes.
Then November 7, 2025, arrived. Speaking at the National War College, Defense Secretary Pete Hegseth unveiled what the administration refers to as the Acquisition Transformation Strategy. For the Pentagon, the wording was remarkably straightforward. He referred to the bureaucracy as “the adversary.” He suggested appointing Portfolio Acquisition Executives with actual budgetary authority and a four-year term. He advocated that a “85% solution in the hands of our warfighters today is infinitely better than a 100% solution years from now.” Reading the memo gives me the impression that what defense reformers have been whispering about since the Obama administration has finally been voiced.
It remains to be seen if it is effective. The House Armed Services Committee is working on the SPEED Act, which aims to shorten the requirements process from almost three years to as little as ninety days. For munitions and “low-end” systems, multi-year contracts are being extended to provide industrial-base subcontractors with a predictable base against which to make plans. Eleven of the Pentagon’s fourteen identified critical technology areas—AI, biotech, space launch, quantum, and advanced materials—are already led by commercial companies rather than defense ones, which is one reason why commercial-first contracting is being pushed harder than at any other time in recent memory. In other words, instead of politely inviting the innovators to come knock, the Pentagon is finally pursuing them.
Even though the cynicism in any defense-watcher should be loud, it’s difficult to avoid feeling a glimmer of cautious optimism when reading the new strategy. Reform has previously been pledged. The Section 809 panel, the Better Buying Power initiatives, and the various “Valley of Death” task forces all used language that sounded just as serious. However, the F-47 is still ten years away from a hangar, and a founder working on autonomy software in a loft in Long Beach will still have to spend the next eighteen months battling FAR clauses that were drafted before her parents were even born. The next two budget cycles will subtly address whether this round of reform ultimately closes the gap between what startups can produce and what the Pentagon can purchase.
