Fear as Architecture: The Pentagon's Real Acquisition Problem
The Department of War's acquisition system processes hundreds of billions of dollars annually through roughly 200,000 contracting officers, program managers, and acquisition specialists. This edition of Defense Tech and Acquisition, filed from the Creative Disruptors in the Desert conference in Palm Springs, assembles multiple voices arguing that the system's dysfunction is not regulatory but behavioral. The contributors disagree on specifics but converge on a single diagnosis: the people executing defense procurement are rationally choosing caution over speed, and no memo from the top will change that without structural reform.
Bonnie Evangelista frames the problem in the starkest terms. Her argument is not that the Federal Acquisition Regulation (FAR) works well, or that Congress provides adequate oversight, but that those obstacles are secondary to something deeper.
The greatest threat to acquisition transformation is not the FAR. It is not congressional oversight, statutory constraints, insufficient training, or an absence of strategy. The greatest threat is fear. Not fear as emotion, but fear as architecture.
Evangelista points to a pattern that has repeated for three decades. Senior leaders announce reform. Innovation teams stand up, deliver results, then dissolve when their champions rotate out. The system reverts. She names specific examples: the Defense Innovation Unit trailblazing Commercial Solutions Openings, AFWERX leading its open topic model, Kessel Run pioneering continuous Authority to Operate, and the Joint Artificial Intelligence Center's Tradewinds Marketplace. Each succeeded because individuals at low levels chose to act against inertia.
Leaders publish memos, give flowery speeches, and issue calls to arms. However, the operational reality shows leaders demand speed while rewarding caution.
Timothy Cooke picks up this thread with a more granular lens. He describes the daily calculus of a contracting officer who might choose a slower, more expensive option from an established contractor, not because the innovative alternative fails to comply, but because choosing it requires justification, which invites scrutiny, which creates career risk.
Procurement professionals follow patterns of risk avoidance that look coordinated but emerge from individual self-preservation.
This is a critical insight. The system looks like deliberate obstruction from the outside. From the inside, it is thousands of rational actors independently choosing the safest path. Cooke's proposed remedy is to change the default: mandate that software acquisitions use Commercial Solutions Openings paired with Other Transactions, making innovation the path of least resistance rather than the career-threatening exception.
Commercial First, Development Last
Jamie Garcia, drawing on decades in federal procurement, reinforces the cultural dimension with a concrete example. When her team needed to award an 85-million-dollar support services contract, traditional timelines projected fourteen months. They delivered it in six, without protest.
The secret wasn't cutting corners -- it was eliminating unnecessary process steps that added no value to the procurement outcome.
Garcia cites the Army's Acquisition Career Management Office, which recently published four reform imperatives: maintain a bias for action, put commercial solutions first, prioritize speed, and focus on outcomes. The shift from "development first to development last" is, in her assessment, the most dramatic philosophical change in recent guidance. The Army now explicitly instructs that if a capability cannot be adopted outright, it should be adapted, and development should be a last resort.
There is a tension worth noting here. The articles collectively celebrate the idea that commercial industry innovates faster than government can develop. That is often true for software, sensors, and drones. But it is less obviously true for nuclear-powered submarines, hypersonic propulsion, or intercontinental ballistic missiles, all of which appear elsewhere in the same newsletter. The "commercial first" mantra risks becoming doctrine in domains where it does not apply, and none of the contributors draw that boundary explicitly.
Lessons from Ukraine and the End of Per-Seat Software
Michael Brown and Matt Kaplan argue that Ukraine has shattered assumptions held since the 1991 Gulf War. American military planners had concluded that future wars would be short and decided by expensive, exquisite platforms. Ukraine proved otherwise.
There has never before been a war with such broad and effective use of satellites, autonomous systems, AI software, and other commercial gear.
They argue that American defense officials must learn to adapt systems at the speed of software development, meaning hours and days rather than years, and must design cheap weapons that manufacturers can produce at scale. The emphasis is on mass over exquisiteness.
Ben Van Roo extends this logic into software procurement. He contrasts how the Air Force buys readiness for C-17 transport aircraft under a 23.8-billion-dollar performance-based logistics contract with Boeing -- where the metric is mission capable rate, not spare parts purchased -- against how the same government buys software: per seat, per license, per module.
SaaS companies sold nouns -- platforms, dashboards, seats. AI sells verbs -- analyzing, deciding, executing. When the verb gets commoditized, the noun that enabled it stops being worth what it used to be. The government is still buying nouns. Expensive ones. With annual renewals.
Van Roo points to the Fiscal Year 2024 National Defense Authorization Act, which created an "Anything-as-a-Service" pilot where the Department of War pays for capability metered by actual usage rather than by license. The initial rollout targets Special Operations Command, Cyber Command, and Transportation Command. He also highlights the SWIFT program, launched in May 2025, which uses artificial intelligence and large language models to grant rapid provisional Authorities to Operate for software.
Readiness Lost to a Two-Dollar Washer
Two stories in this edition illustrate the same absurdity at different scales. Colonel Michael Mai describes how a significant portion of the Army's Abrams tank fleet sits non-mission capable, waiting for parts that could be manufactured using additive manufacturing in hours. The tanks are sidelined not by complex components but by hoses, brackets, and washers produced by sole-source vendors optimized for large batch production.
Mai calculates that if advanced manufacturing restored forty tanks, the readiness impact would be immediate: forty tanks multiplied by fifty training days yields two thousand additional Abrams training days per year. The total non-recurring cost to engineer and qualify the entire catalog is less than one hundred million dollars. Yet the Defense Logistics Agency's pricing model cannot justify purchasing a higher-cost part today, even when doing so would immediately restore vehicle availability.
At Marine Corps Base Camp Lejeune, Marines in the 2nd Maintenance Battalion faced a similar problem when a shortage of single-use suspension washers took multiple Amphibious Combat Vehicles out of service. Warrant Officers Anthony Juedes and Matthew Pine tasked their fabrication platoon to reverse-engineer the part. It took less than thirty minutes to create a working prototype and less than ninety seconds to cut each subsequent part. Fifteen non-operational vehicles returned to service, each for a part costing less than two dollars to manufacture.
These stories are the newsletter's strongest evidence. They make the abstract arguments about fear and risk aversion viscerally concrete. The system is not failing at the frontier of technology. It is failing at washers.
The Bigger Bets: Drones, Directed Energy, and Golden Dome
The newsletter covers an enormous range of programs across the services. The Department of War began testing for its Drone Dominance Program at Fort Benning, evaluating twenty-five competing vendors in a process called "the Gauntlet." By program's end, the department aims to acquire some 300,000 drones and name five primary vendors. The Army held its first Best Drone Warfighter Competition, drawing over 800 attendees and signaling that drone operation is becoming a baseline soldier skill rather than a specialist function.
Chief of Naval Operations Admiral Daryl Caudle floated the idea of creating a Robotic Autonomous Systems commander to coordinate unmanned capabilities across domains, comparing the role to a joint task force commander dedicated to autonomous systems alone. The Navy also approved Raytheon's StormBreaker smart weapon for the F/A-18E/F Super Hornet and is pushing aggressively on directed energy weapons to counter drone and missile threats.
Ryan Frigm makes the case that the Golden Dome for America initiative -- the homeland missile defense program -- will fail without a software-defined architecture from the outset.
At its core, Golden Dome is a systems integration problem. The only way to solve it on the required timeline is through digital alchemy: using software to transform existing materiel into a capability greater than the sum of its parts.
Frigm argues that the technologies required -- massive compute, edge artificial intelligence, and autonomous orchestration -- are no longer exotic but industrial-grade and deployed commercially at scale. The first step, he writes, is high-fidelity modeling and simulation, not as visualization but as a decision-making engine.
Industrial Base and Capital Formation
The Export-Import Bank of the United States (EXIM) is emerging as an unexpected player in defense industrial policy. The bank authorized 185 million dollars for Cesium Astro's new 270,000-square-foot production facility in Texas, its largest deal to date under the "Make More in America" initiative. This mechanism, which did not exist five years ago, allows EXIM to fund domestic factory construction provided fifteen to twenty-five percent of output serves export markets. EXIM operates like a bank -- lending and charging interest -- and has historically turned a profit for the Treasury.
The newsletter also covers Project Vault, a twelve-billion-dollar public-private initiative to stockpile all sixty critical minerals on the United States Geological Survey list, backed by the largest loan in EXIM's ninety-two-year history. Three commodity trading houses -- Hartree Partners, Mercuria Energy Americas, and Traxys -- provide the logistics infrastructure.
Think of it less as a government program and more as a structured insurance product for American industry.
One legitimate question the newsletter does not address is whether EXIM's expansion into domestic factory financing and critical mineral stockpiling stretches the agency beyond its competence. Trade finance and strategic industrial policy are different disciplines, and the speed of expansion invites the same kind of oversight gaps the acquisition reform articles warn against.
Bottom Line
This edition of Defense Tech and Acquisition is a snapshot of a system simultaneously moving fast and standing still. The Department of War is testing 300,000 drones, fielding autonomous ships, accelerating B-21 Raider production, and deploying nuclear reactors that fit inside a C-17. At the same time, Abrams tanks sit idle for want of a washer, the Sentinel intercontinental ballistic missile program slides further behind schedule, and contracting officers still default to the safest procurement path because the incentive structure rewards exactly that behavior. The contributors to this newsletter -- Evangelista, Cooke, Garcia, Brown, Kaplan, Van Roo, Mai, Frigm, and others -- are largely preaching to the converted. The harder audience is the mid-level acquisition professional whose career depends on never being the person who justified the wrong contract. Until the system makes it safer to choose speed than to choose precedent, the memos will keep flowing and the culture will keep resisting.