The late Pentagon strategist Andy Marshall once remarked that technology makes a military revolution possible, but it is not a sufficient condition to initiate that revolution on its own. Recent developments in warfare and defense reform alike suggest we are now on the cusp of a military revolution, focused on attritable mass. The future of military power will not be determined by the most exquisite hardware, but by the most effectively scaled and networked family of capabilities.
This is not an altogether new observation. But far less discussed are the implications of this fact. As the United States begins a major shift toward employing systems that are attritable by design—deliberately engineered to be used fully, deployed widely, and replaced quickly—procurement, force design, and industrial policy must also be reorganized and reconsidered to an extent not experienced in decades.
Below, we’ve outlined some institutional principles for what a revolution in attritable mass will mean—for investors, Department of War officials, industry, and others.
First, define attritable. It is critical to acknowledge that ‘attritable’ does not necessarily mean ‘cheap.’ A more precise definition deals in economies of scale. The right heuristic to understand attritable is a trade-off: a given system should be roughly 90 percent as effective as its hard-to-replace counterpart, but come with an order-of-magnitude price reduction. This trade creates freedom of action for military planners to take bigger risks, scale more quickly, and fight in ways previously thought impossible when solely relying on exquisite weapons systems.
Since this concept is focused on comparative costs, ‘attritable’ as a category varies in price point depending on the platform or weapons system and its mission set. For example, an attritable drone may cost $50,000 dollars, rather than $500,000; an attritable submarine might cost $10 million, not $2 billion; an attritable loitering munition might be $100,000, not $3 million. Attritable is not cheap, rather it affords the operator new ways of achieving and exceeding traditional mission outcomes.
Then, transform the industrial base with software at its center. Buying large quantities of attritable systems requires a fundamentally different kind of industrial strategy. Bespoke performance goals take a distant backseat to speedy production schedules, lower costs, and high-rate manufacturability. As such, every component should be assessed not by a maximalist wish list of capabilities, but instead by pragmatic benchmarks like rapid scalability and supply chain resilience. It’s about building the most of a given thing—not building the most capable and impressive single thing. Therefore, bills of materials for platform development should be redesigned to eliminate the usage of highly specialized, low-yield parts.
Importantly, none of this means surrendering exquisite hardware for suboptimal replacements. Instead, the exchange is exquisite hardware for superior software. Effects that have traditionally been delivered through costly and slow precision engineering can be done increasingly through high-fidelity autonomy, AI-driven targeting, mesh networking, and swarming. This shift in operational concepts necessitates a connected shift in the character of the defense industrial base, which must be software first, cycle time driven, horizontally integrated, and globally integrated.
From stultified bureaucracy to “compete, then commit.” If the U.S. military is serious about adopting an attritable force structure, it must fundamentally rewire how it procures capability. Continuous competition, compressed timelines, and front-loaded investment are integral qualities to a better process, and none of them are manifest in the extant system. But competition alone isn’t enough to create scale in new programs.
The future is best thought of as a careful dance between persistent competitive pressure and periods of concentrated investment. We might call this model “compete, then commit.”
Imagine a timeline for a given program: Year 1 initiates a 12-month competition among several vendors. Year 3, even while scaling the incumbent’s solution, launches the next round of competition for a follow-on tranche. This structure would force defense companies to behave more like technology companies, moving profit into internal research and development (IRAD) and versioning products at software speed.
This process would be a massive force multiplier for government purchasing power. Think of it as a virtuous cycle: companies would have to make significant expenditures to compete for an original award. Those that lose could turn to private equity, venture capital, and other forms of capital collection for internal research in the interim, creating a pathway for a potentially successful recompete for the follow-on award a few years later.
It's hard to overstate what this would do to streamline government efficiency. Currently, the government bears the heavy load on interim research, development, testing, and evaluation (RDT&E), and must split taxpayer funds between procurement and RDT&E. Under our new model, the government could reduce the RDT&E burden by supplementing it with industry IRAD, and could plow savings into procurement. This would significantly increase the effective buying power of the military, possibly two- or three-fold.
To maintain and steward this powerful model, Program Executive Officers (PEOs) must evolve into portfolio managers, not shepherds of single platforms over decades. Their job—properly understood—is to declare performance and cost outcomes, then manage risk across a slate of options. This means embedding personnel with vendors, promoting internal bureaucratic champions, and constantly recalibrating based on who is delivering.
Yet even these competitive changes demand a steady focus on front-loaded investment. Signaling seriousness, attracting top talent, and allowing performers to plan against real scale comes with a non-trivial price tag. The Department therefore must from the outset signal not just its desired capability, but how many units, and at what cadence: 2,000 missiles, 10,000 drones, 50 submarines—numbers that make firms pay attention.
‘Colors of money’ and thinking about incumbency. Competing and committing also mean changing the ‘color of money’ that creates firewalls for iteration. Dollars that go toward Operations and Maintenance (O&M) treat programs like a 20-year tenure track, not the urgently needed five-year commercial product cycle. It’s worth acknowledging that vendors are the most expensive they will ever be during early-stage competition. But by concentrating awards on a temporary basis, cost curves can sharply drop while future recompete prospects keep incumbents focused on performance.
Critics might contend that incumbents will have inherent advantages. But that isn’t necessarily bad as a matter of principle. Incumbency is most dangerous if the incumbent knows they will win. A healthy system ensures that’s never a guarantee. Performance and cost should improve over time—not by accident, but by design.
Software as infrastructure, not government-authored property. Software is the foundation of any attritable architecture—it’s what enables mass to behave with the precision and coordination previously reserved for exquisite systems. But the right approach to the software stack isn’t one-size-fits-all, nor is it the government developing its own software. Roughly 85 percent of needed software capabilities are already available commercially and improving rapidly; modules like autonomy frameworks, mesh networking, and onboard processing can be bought, integrated, and iterated through market competition.
The remaining 15 percent presents an undoubtedly harder problem. Such components generally lack commercial incentive for development (such as visual navigation in GPS-denied environments), or they require constant testing in real-world conditions for optimal performance—conditions vendors cannot easily access. None of this, however, should mean that the government needs to build these capabilities itself. Instead, the government should anchor their development. It should create the conditions, data environments, and access pathways that allow vendors to iterate and mature these capabilities over time. The government’s job is to identify which software modules demand theater-level feedback loops and ensure those loops exist—through access, not authorship.
This is analogous to how we treat hardware reuse. No one re-competes wings or battery packs every cycle. Some software—especially where real-world validation is scarce—should be treated like semi-fixed infrastructure. The key policy question isn’t who writes the code, but who ensures it can be tested. The government can ensure innovation where it's possible and stability where it's necessary. These propositions should guide ownership, contracting, and investment structure.
Toward a “Operations and Maintenance-light” military. What if the military stopped treating sustainment as a default and started designing for disposal? In an O&M-less force, innovation, rapid fielding, and replacement—not refurbishment—become the operating logic. Legacy sustainment models, which dominate today’s defense budgeting, would be left behind.
Numbers reflect the continual cost of the old approach. Since the post–Cold War consolidation of 51 defense firms into just five primes, O&M spending has surged from ~30 percent of the DoD budget in 1990 to over 40 percent today. This bloat isn’t simply caused by aging systems, rather it’s the logical outcome of structural incentives.
President Truman observed that, at the height of the Arsenal of Democracy during World War II, cost-plus contracts “were offered by the government in much the same way that Santa Claus passes out gifts at a Christmas party.” Little has changed in the many decades since. Cost-plus contracts that reward lifetime extension and regulations that trap major maintenance with original equipment manufacturers exclude third-party innovation.
We finally have an opportunity to change this. Most affordable mass systems don’t have long-tail obsolescence risks because the parts most likely to break or fall behind are modular by design. Instead of chasing exquisite standards for 30-year survivability, systems should be built to perform brilliantly for five. Seekers and communications and guidance modules can be upgraded independently, and networked warehouse infrastructure enables software to be continuously integrated, deployed, and patched—even at the edge.
When we treat hardware like a consumable, software remains evergreen. This would improve efficiency, but even more importantly, it would amount to a quantum leap in strategic agility. The O&M spiral would end, and the industrial base could move at the speed of relevance.
Employing the principles of the attritable mass revolution. As simple as some of these principles may sound, they would be utterly transformative on a remarkably short time horizon.
Measuring the success of these principles would be simple, too. Ask these basic questions:
Are programs buying thousands of units instead of dozens?
Are they iterating every 12–24 months instead of every 12–24 years?
Are prototypes transitioning into fielded systems at scale?
Attritable systems should not be rare, exquisite unicorns. They should be ubiquitous, affordable, and constantly evolving. If implemented, the attritable model will unleash a flourishing defense industrial base with more entrants competing, faster iteration cycles, and a broader range of investors participating. We will see systems with comparable or superior performance delivered at a fraction of the cost. Faster iteration will drive greater adaptability and resilience, while scale will unlock expanded mission sets.
Adherence to these principles would be the full realization of Marshall’s concept of military revolution, applied to attritable mass. The true measure of deterrence in the 21st century will be the volume, velocity, and adaptability of systems in the field, not the persistent presence of legacy systems on tarmacs and in munitions depots.
It’s time to break out of the domain of slide decks and pilot programs and end the long dominance of a maintenance-centric defense model. The revolution is here. Let’s seize its momentum.
Matthew Steckman is President and Chief Business Officer at Anduril Industries.
Eliot Pence is Founder of Dominion Dynamics.
 | A guest post by
|
 | A guest post by
|