The Arsenal Dilemma: The Case Against Centralized Manufacturing
Why Anduril Could be Taking a Strategic Misstep for US Defense—and its Investors
Last week, Anduril Industries, America’s leading DefenseTech scale-up, announced it had closed a $1.5Bn Series F fund raise.
Before I go any further, I want to highlight that I’m a massive fan of Anduril, consider myself to have quite a few friends working there, and wish to highlight that the raise and associated $14Bn valuation are well-deserved!
Anduril has absolutely transformed the conversation around Aerospace & Defense and has been a major force in driving both the Defense Technology sector and in the conversation about attritable systems.
All of that is to say that I love the company, love the people in the company, love their impact. Yet, I have reservations about their announced use of the most recently raised funds, and specifically the intent to build a large (five million square foot), vertically-integrated manufacturing facility, that they’ve dubbed Arsenal-1. Further, the company has announced their intent for a follow-on Arsenal-2 facility.
While I hold deep respect for Anduril and their accomplishments, it’s precisely because of their potential and influence that I feel compelled to critically assess their recent decision
To be sure, there’s much to like about what Anduril is building with Arsenal-1. Using mass-manufacturing as a first principle is critical to being able to produce the necessary volumes to fight and win future wars. Software-first approaches will allow for greater speed and higher quality in design and production.
But, the overly-centralized, vertically-integrated design of the effort has several specific drawbacks that must be addressed. First and foremost, the effort smacks of exquisite systems. Wired likens it to Tesla and Apple manufacturing programs—notoriously focused on higher-end, exquisite systems compared to their competition. This in and of itself isn’t an issue. We need high-end, exquisite systems. But it is inconsistent with Anduril’s stated goal of high-volume production of low-cost systems. This is further borne out by Anduril’s role in the high-cost / exquisite collaborative combat aircraft (CCA) program, which carries an estimated $25-30M price tag per aircraft.
As I’ve said before, we certainly need exquisite systems—and someone in the Pentagon really ought to run some numbers and determine what the right mix of exquisite and attritable is.
Now, let’s get into the bigger problems with the Arsenal model.
First, centralization of manufacturing creates a critical vulnerability. During the Second World War, the Allies conduct a campaign of strategic bombing, which varied from the firebombing of entire cities seeking to hurt morale (Tokyo, Dresden, Hamburg being prime examples) to a more targeted series of attacks against factories producing war materiel for the Axis. For example, air power hero Curtis Lemay led a daring raid against Regensburg1 that destroyed up to 1,000 Messerschmitt Me-109s and crippled the effort to build Me-262 jets. Several campaigns were similarly conducted against ball-bearing plants in Schweinfurt concomitant and subsequent to the Regensburg raid.
Just as WWII-era bombing campaigns targeted centralized production facilities, today’s adversaries, with cyber capabilities and precision strikes, could devastate a single-point manufacturing hub like Arsenal-1, bringing U.S. defense production to a standstill. Centralizing production capabilities for military systems is tantamount to painting a large target on the roof of the building. We have witnessed in Ukraine how Russia has targeted large factories, forcing the Ukrainian industrial base to disperse and disaggregate. While America has for centuries enjoyed a certain geographical security provided by oceans on either side of us, advances in military technology including cyber capabilities, space-based systems, and hypersonics, have largely negated that advantage. In other words, in a shooting war with, let’s say, China, our adversaries will be able to target industrial centers in the United States and its folly to think they won’t do it.
In a very similar manner, this highly centralized and vertically integrated approach is susceptible to other supply chain interruptions. The U.S. military increasingly relies on continuous, resilient supply chains to maintain its operational edge. A single, centralized facility introduces a significant point of failure. Should Arsenal-1 experience a disruption—whether from an adversarial attack, a natural disaster, or even a logistical issue—the ripple effects could be felt across multiple defense programs. This vulnerability is antithetical to the principles of distributed, resilient manufacturing that the U.S. government and military need to prioritize, especially as geopolitical tensions rise.
The next greatest risk that this vertically-integrated facility presents is one of obsolescence. Massive CAPEX takes time. Creating a 5 million square foot facility, buying equipment, hiring 1,500 people all take time. So do setting up production runs. And technology continues to accelerate creating a potential vulnerability. In an industry where technological advancements can render existing systems obsolete almost overnight, the rigidity of a large-scale, vertically-integrated facility like Arsenal-1 becomes a liability. The need for constant retooling and updates not only incurs massive additional costs but also slows down the deployment of cutting-edge technologies that could be decisive in future conflicts
It seems that Anduril has acknowledged this risk and they seek to somewhat mitigate it through a ‘software first’ approach. Nevertheless, the risk persists. If a new technology emerges, it may require retooling, layoffs, retraining, or otherwise additional massive CAPEX costs. It will slow Anduril down in their responsiveness—potentially to the speed of the ‘primes.’ The same primes that won’t do anything until they have a contract from the government, the sort of mindset anathema to innovation.2 This slow responsiveness is part of what has driven the primes to the high-end, exquisite systems that they favor.
Closely related to the obsolescence risk is the potential for delayed fielding of required capabilities. The military depends on speed and innovation, particularly in conflict scenarios where the ability to quickly produce and deploy new systems can be the difference between victory and defeat. A decentralized, networked production approach, in contrast, would allow for faster adaptation and quicker delivery of critical systems to the battlefield. Assistant Secretary of Defense for Sustainment Chris Lowman has acknowledged this requirement at least for maintenance, repair, and overhaul capabilities creating the Regional Sustainment Framework to help address this concern.
Finally, I want to address the business risks and associated economic impacts of Anduril’s centralized approach. Anduril will grow its headcount by 60% or 1,500 new employees to staff Arsenal-1. Fortunately, manufacturing is coming back into vogue and is becoming considered fun, cool, & sexy (h/t to Christine Holly Ngo), but how quickly will the company be able find and hire 1,500 employees? The challenge of scaling the workforce by 60% to staff Arsenal-1 cannot be underestimated. Failing to attract and retain the necessary talent could lead to production delays, missed deadlines, and ultimately, a loss of trust with the U.S. government. Furthermore, if Anduril were to face economic headwinds, such as losing a key contract, the company might be forced to downsize, leading to layoffs and underutilization of the facility, exacerbating the financial strain.
In the end, I am wishing Anduril all the success in the world. We need them to be successful. And I’m confident that they will be. At the same time, I think their model is not the right one for rebuilding the defense industrial base and it’s not the right one for mass-producing truly low-cost, scalable, and attritable systems.
So, what is?
Well, 3 of the 4 tenants that Anduril is espousing are ideal:
Design for simplicity and scale
Resilient supply chains
Software-defined production
The only one that they’ve gotten wrong is the centralized facility and manufacturing. Instead, we need a tech-enabled, distributed, and federated ecosystem of manufacturers to enable cost-effective autonomous systems.
A distributed, federated ecosystem also smooths out perturbations in requirements by enabling rapid scaling up or down of production across multiple sites, which is particularly valuable in responding to shifting defense needs. This approach helps keep overhead costs lower by avoiding the significant capital expenditures (CAPEX) associated with building and maintaining a massive, centralized facility. Instead, resources can be allocated more efficiently across a network of smaller, agile manufacturers. This not only ensures cost-effective production but also reduces the financial risks for both the government and investors, making it a more sustainable and adaptable model for rebuilding the defense industrial base.
As we navigate an increasingly complex and unpredictable global security environment, it’s imperative that we rethink our approach to defense manufacturing. A distributed, federated model not only aligns with the future of warfare but also ensures the resilience, adaptability, and cost-effectiveness on which our national defense—and our economic security—depend.
Stay tuned. Big news is coming on this front and, as always,
Keep building,
Andrew
Regensburg is a lovely city and home to many in my family and in the Building our Defense family.
This may signal that as any start-up or scale-up grows to sizes nearing the rimes, that they necessarily must look more like a prime. I sincerely hope that’s not the case, but there isn’t enough data out there yet.
I am the owner of a small startup company, over the past two years we have been developing High Altitude Heavy Lift Quadcopters.
We have developed two models both have the newest Flight Control technology and have RTK Zed GPS. Both use open source ground control stations.