Annotated transcription · 18 min read
How Modern Defense Manufacturing Learned to Stop Worrying and Ship Fast
A century of Pentagon bureaucracy created the opportunity for a new kind of prime contractor—one that treats deterrence as a software problem.
The Self-Inflicted Wound
Understanding the current defense landscape requires grappling with a paradox: the United States emerged from World War II as the world's manufacturing superhouse, commanding over 40% of global production capacity and funding more than a third of all research and development worldwide through the Department of Defense alone. Yet today, war games consistently show American forces losing hypothetical conflicts with peer adversaries. The nation that once built bomber aircraft at a rate of one per day now struggles to replace munitions that would be exhausted in under two weeks of large-scale combat.
This decline did not result from external conquest or technological surprise. It emerged from decades of compounding policy decisions, each individually defensible, that collectively strangled the industrial base that once made American military dominance inevitable. The roots trace back to the 1960s, when Secretary of Defense Robert McNamara introduced sweeping bureaucratic controls designed to rein in what he perceived as runaway defense spending. His Planning, Programming, Budgeting, and Execution system blanketed the defense industrial base in layers of auditors, process requirements, and vendor evaluations—friction intended to prevent waste.
The immediate effect was a temporary reduction in spending. The long-term consequence was the opposite: every program became dramatically delayed and significantly over budget. High-friction processes made experimentation prohibitively expensive. Development timelines stretched from years to decades. Engineers who once worked on twenty or thirty different platforms over their careers found themselves spending entire decades on a single program. The ecosystem that had rapidly iterated its way to technological superiority retrained itself to build enormous, exquisite assets—aircraft carriers, nuclear submarines, stealth bombers—at costs measured in the tens of billions and timelines stretching across generations.
This might have remained sustainable during the Cold War, when deterrence primarily meant deploying the biggest, most threatening capabilities on day one of any hypothetical conflict. But the world moved on. The brief unipolar moment following the Soviet collapse created a dangerous assumption: that American supremacy was permanent, that no peer competitor would arise, that the rules of warfare had fundamentally changed in favor of the established superpower. Meanwhile, potential adversaries studied American vulnerabilities and developed asymmetric responses.
The Cost Imbalance of Modern Conflict
The new arithmetic of warfare became starkly visible in Ukraine, where Operation Spiderweb demonstrated the brutal math that now governs military effectiveness. Ukrainian forces smuggled roughly one million dollars' worth of autonomous drones into Russian territory, concealed beneath false floorboards. Once deployed, these systems autonomously identified high-value targets and engaged, ultimately destroying approximately seven billion dollars in enemy assets—including a third of Russia's long-range bomber fleet.
This million-to-seven-billion exchange ratio represents the new normal in modern conflict: high-volume, low-cost, attritable assets overwhelming expensive legacy platforms. China has systematically developed what it calls 'carrier killers'—missiles costing perhaps twenty million dollars that can destroy multi-billion-dollar naval vessels. This forces a grim calculation: American aircraft carriers, which take decades to build and represent enormous concentrations of capability, become assets too valuable to risk in the western Pacific.
The Pentagon currently maintains munitions stockpiles sufficient for roughly eleven days of high-intensity conflict. Rebuilding those stocks, given current manufacturing capacity, would take years. Replacing naval assets lost in combat would take decades—if the shipbuilding capacity even exists, which in many categories it does not. Meanwhile, China operates under a military-civil fusion doctrine that treats all domestic manufacturing as dual-use: civilian factories can be rapidly retasked to military production when needed.
This is not a technological problem. The United States possesses cutting-edge capabilities across domains. It is an industrial problem, a supply chain problem, and ultimately a systems integration problem. The question is not whether American engineers can build superior hardware—they demonstrably can—but whether the acquisition system can iterate fast enough to stay relevant, manufacture at sufficient volume to matter in conflict, and integrate capabilities across domains to create overwhelming advantage.
Building the Lattice Before the Products
When Anduril's founders gathered for their first recruiting lunch in 2017, armed with mediocre slide deck and Chick-fil-A, they outlined five potential product directions. Two bookended what would become their product portfolio over the following years: 'perimeter security on a stick'—ultimately realized as the Sentry Tower—and 'VR for the warfighter,' which evolved into the Eagle Eye augmented reality system. But the through-line connecting these disparate capabilities was not any individual platform. It was Lattice, the AI-driven sensor fusion brain designed to orchestrate assets across every domain.
This architecture represents a fundamental departure from traditional defense contracting. Rather than bidding on specifications for particular platforms, Anduril builds a software-defined, hardware-enabled web of capabilities that can be continuously upgraded and rapidly integrated. Lattice serves as the nervous system; sensors, drones, munitions, and undersea vehicles become the distributed organs of a unified capability.
The product evolution followed this logic. Sentry Towers provided stationary sensing along borders and around military bases—information-gathering nodes that fed Lattice's understanding of the environment. Quadcopters and Ghost drones extended coverage into aerial domains with different endurance and range characteristics. Counter-drone systems like Altius allowed the network to not merely observe threats but engage them. Loitering munitions like Roadrunner added precision strike capability. The product suite expanded methodically into maritime domains with Seabed Sentry sensors and the Dive family of autonomous underwater vehicles.
Crucially, none of these platforms required reinventing fundamental technologies. Sentry Towers integrate commercial cameras, sensors, and batteries rather than spending years developing bespoke components. This approach—leveraging existing supply chains and proven technologies, then integrating them into a coherent operational capability—allows for rapid iteration and dramatically lower costs than traditional programs-of-record.
The Ocean as Theater
Undersea warfare represents perhaps the most challenging and consequential domain for modern conflict. The ocean floor hosts the vast majority of global data infrastructure—fiber optic cables carrying trillions of dollars in daily transactions, connecting satellite networks, energy grids, and financial systems. These cables are largely undefended, running through contested waters where hostile actors can map and potentially sever them at will.
The scale of the problem is staggering. The ocean represents the largest addressable surface area outside of space itself, yet remains the most difficult environment to monitor and understand. Traditional approaches to maritime domain awareness rely on sparse, expensive platforms: crewed submarines, surface ships, limited sonar networks. These assets are both vulnerable and insufficient to cover the necessary area.
Anduril's maritime product family tackles this through distributed autonomous systems. Seabed Sentry platforms—hardened to withstand the ocean's relentless corrosive forces—provide persistent sensing from fixed positions. The Dive line of autonomous underwater vehicles offers mobile coverage, capable of loitering, mapping, and even carrying munitions. Copperhead torpedoes add autonomous engagement capability. Most recently, a partnership with Hyundai signals entry into autonomous surface vehicles, completing coverage across every maritime layer.
This layered approach reflects a core insight about modern deterrence: coverage matters more than individual platform superiority. An adversary might be able to defeat any single sensor or vehicle. But a meshed network of affordable, replaceable systems creates persistent awareness and response capability that cannot be easily overcome. The goal is not invulnerability but resilience—the ability to absorb losses and continue functioning.
From Cost-Plus to Fixed-Price
The incentive structure of traditional defense contracting actively discourages efficiency. Under cost-plus arrangements, contractors receive a guaranteed margin—typically ten to fifteen percent—on top of whatever they spend developing and building a system. This creates a perverse dynamic: ten percent of three billion dollars is vastly more profitable than ten percent of two hundred million. Contractors are thus incentivized to maximize program complexity, extend timelines, and avoid reusing existing solutions.
This explains how a contract to build sensor towers along the southern border could balloon to a five-to-ten-billion-dollar program expected to take fifteen years, with the prime contractor ultimately concluding the requirement defied the laws of physics. There was no incentive to find a simple, rapid solution. The incentive was to make the problem as expensive and time-consuming as possible.
Anduril's fixed-price model inverts this dynamic entirely. The company funds its own research and development, putting its capital at risk rather than asking the government to pay for experimentation. Products are built to satisfy mission requirements—not detailed specifications—and sold as proven capabilities. The government pays for outcomes, not effort. This aligns incentives: Anduril profits by delivering effective solutions as quickly and cheaply as possible, then selling at scale.
The first pitch deck explicitly framed this approach: the company would make billions by saving the government tens of billions. It was an audacious claim that appeared to defy defense industry gravity. Yet the track record increasingly validates the model. Programs delivered on time and under budget—a rarity bordering on miracle in traditional procurement—have become Anduril's calling card.
Founders and Mavericks
Throughout military history, breakthrough capabilities have emerged from a specific pairing: founder-level builders protected by institutional mavericks willing to run interference against bureaucracy. Bernard Schriever developed intercontinental ballistic missiles because President Eisenhower shielded the program from typical acquisition processes. Kelly Johnson created the U-2 spy plane and stealth bomber at Lockheed's Skunk Works because CIA leadership gave him extraordinary latitude to operate outside normal channels.
This pattern broke down over recent decades not because potential mavericks disappeared, but because no founders existed for them to champion. Traditional primes had been fully captured by the bureaucratic system—their entire business models optimized for navigating rather than circumventing regulations. Meanwhile, a generation of military leaders who understood the coming shift toward high-volume, low-cost, software-enabled warfare issued warnings that went unheeded. They would have been mavericks if anyone had been building solutions worth protecting.
The current moment reunites both sides of this dynamic. A cohort of founders—bringing Silicon Valley's velocity and commercial technology's capabilities—has entered the defense market. Simultaneously, urgency around great power competition has elevated military leaders who recognize the need for rapid acquisition reform. The missing ingredient was not political will or technical capability but the pairing of builder ambition with institutional support.
This explains why legislative changes, while helpful, are insufficient. The defense bill passes with bipartisan support year after year, reflecting broad consensus that reform is necessary. But legislative mandates cannot create the iterative feedback loops, risk-tolerant capital deployment, and engineering culture required to actually ship better capabilities faster. Those emerge from companies purpose-built around different operating principles.
The Narrative Shift
When Anduril began fundraising in 2017, working in defense was deeply unpopular in technology circles. Engineers Palmer Lucky had worked with at Meta—people who had promised to follow him to his next company—ghosted him when they learned he was building defense capabilities. Early investors faced angry messages from friends calling their backing immoral. Palmer required security at campus speaking events due to threats of physical violence from activists.
The 2022 Russian invasion of Ukraine marked an inflection point in this narrative. Suddenly, large-scale conventional conflict was not a historical artifact but a present reality broadcast globally. The comfortable assumption that great power war was obsolete crumbled. Investors and engineers who had dismissed defense as morally compromised began reconsidering whether deterrence might actually prevent worse outcomes than pacifism.
Yet even this shift proved rocky. The broader market collapse of 2022 created a brief crisis of confidence just as the strategic argument for defense investment was strengthening. Anduril found some investors backing out of commitments during this period. Only after markets stabilized and evidence from Ukraine continued mounting did the 'defense bubble' fully inflate.
This bubble is both necessary and precarious. Necessary because funding innovation requires speculative capital willing to back long development timelines and uncertain outcomes. Precarious because bubbles inevitably produce capital destruction—good-faith efforts that nonetheless fail, as well as opportunistic ventures that never had viable paths to capability. The challenge is ensuring enough genuine capability gets built and fielded before sentiment shifts again.
Beyond Prime Contractors
Not every defense technology company can or should aspire to become a new prime contractor. The skills required—building cutting-edge hardware, integrating complex software, manufacturing at scale, and navigating Pentagon acquisition processes—represent an extraordinarily rare combination. Perhaps a handful of companies will successfully assemble all these capabilities over the coming decade.
The longer tail of the ecosystem will consist of specialized capability providers: companies solving particular technical challenges in sensing, propulsion, materials science, or software that can plug into platforms built by the new primes. Anduril has demonstrated this model through acquisitions, taking companies that would have struggled to independently secure defense contracts and immediately integrating them into established programs.
Even longer tail consists of supply chain and manufacturing enablers. Military superiority has historically been preceded by industrial superiority—a lesson the United States demonstrated in World War II and China is demonstrating today. Rebuilding American industrial capacity requires thousands of companies building components, tooling, and production infrastructure. Many of these businesses will be excellent but poor fits for venture capital: steady, profitable operations growing modestly rather than exponentially.
The defense bubble should ideally catalyze not just a few big winners but an entire ecosystem of manufacturers, integrators, and specialists. This requires moving beyond the standard venture capital playbook of seeking winner-take-all outcomes. It requires appreciating that a company doing twenty million in revenue at thirty percent margins, growing ten percent annually, provides essential capability even if it will never achieve unicorn valuation.
The Mission Behind the Hardware
Throughout hundreds of interviews and thousands of pages of source material, one word appears with striking frequency when Anduril's founders discuss their work: hope. This is not the cynical hope of entrepreneurs pitching inevitability to investors. It is the hope inherent in believing that action can prevent catastrophe, that deterrence can avert conflict, that building capability now protects the freedoms that make building worthwhile.
The company's name itself encodes this philosophy. Andúril, the Flame of the West from Tolkien's mythology, was a sword reforged from fragments—a weapon that represented not love of war but protection of what war threatened. The character in the novels who most resonates with Anduril's ethos says he loves not the blade for its sharpness or the arrow for its swiftness or the warrior for his ferocity, but only that which they defend.
This distinction matters because building weapons without grappling with their purpose leads either to naive optimism about technology's neutrality or cynical acceptance that profit justifies any application. Anduril's founders engage seriously with just war theory, with the ethics of autonomous systems, with the responsibility that comes from enabling lethal force. They argue that pacifism is a privilege afforded only to those protected by others willing to engage with harder questions.
The philosophical framework extends to product decisions: every platform, every integration, every manufacturing choice gets evaluated against whether it advances deterrence—the capability to make conflict so expensive and uncertain that adversaries choose not to initiate it. This is why the company self-funds research rather than asking for government R&D dollars. This is why products are built for mission rather than spec. This is why contracts are fixed-price rather than cost-plus. Each structural choice reflects the underlying belief that efficiency in deterrence ultimately reduces the need for conflict.
The Davidson Window
Admiral Philip Davidson, former commander of U.S. Indo-Pacific Command, testified before Congress that China could attempt forcible reunification with Taiwan within the current decade. This 'Davidson Window' has become a focusing concept for defense planning: a finite period in which deterrence capability must be established or conflict becomes significantly more likely.
The timeline is brutally short relative to traditional defense acquisition. Programs that take fifteen years to field are irrelevant to a threat horizon measured in years. This reality is forcing uncomfortable reckonings throughout the Pentagon: capabilities must be proven and fielded now, using technologies available now, at costs sustainable now. There is no time for another generation of cost overruns and schedule delays.
This urgency explains the current administration's aggressive push to streamline acquisition, the willingness to experiment with non-traditional contractors, and the desperation to reactivate dormant manufacturing capacity. It also explains why the defense bubble, for all its speculative excess, serves a strategic purpose: concentrating capital and attention on capability development when the window for action is rapidly closing.
The broader lesson transcends any particular company or technology. Deterrence requires industrial capacity, which requires supply chains, which requires investment in manufacturing and tooling and skills. It requires software infrastructure to integrate disparate systems. It requires iteration velocity to stay ahead of adversary adaptations. Most fundamentally, it requires confronting the reality that pacifism—however appealing philosophically—depends entirely on others willing to bear the burden of deterrence.
Key takeaways
- → American defense capability declined not through external defeat but through self-inflicted bureaucratic friction that strangled the post-WWII industrial base over decades.
- → Modern conflict economics favor high-volume, low-cost, attritable systems over exquisite platforms—a million-dollar drone destroying seven billion in assets demonstrates the new exchange ratio.
- → Anduril's core product is not any individual platform but Lattice, an AI-driven integration layer designed to orchestrate capabilities across every domain and force level.
- → Cost-plus contracting incentivizes contractors to maximize program expense and duration; fixed-price models align profit with efficiency and rapid delivery.
- → The ocean represents the largest and most critical domain outside space, hosting global data infrastructure that remains largely undefended against adversary mapping and potential interdiction.
- → Historical breakthroughs required pairing founder-level builders with institutional mavericks willing to protect innovation from bureaucracy—a dynamic recently reunited in defense technology.
- → Rebuilding deterrence capability requires not just a few prime contractors but an entire industrial ecosystem of manufacturers, integrators, and specialists operating at different scales and growth rates.