The vocabulary of war is changing from exquisite to expendable to attritable. Attritable systems occupy an intentionally ambiguous middle ground: platforms designed to be reusable in principle but cheap enough that their loss is tolerable in routine operations. The Department of Defense has made scaling attritable, autonomous systems a formal goal under the Replicator initiative, with explicit ambitions to field large numbers across domains. This strategic shift is not merely a procurement choice. It forces the acquisition community to reexamine how we measure cost, value, and risk across the life cycle of capabilities.

Cost modeling for attritable systems cannot be a trimmed version of existing models for tanks, fighters, or large UAS. Traditional life‑cycle cost constructs remain necessary, but they are not sufficient. Classical guidance on cost estimation emphasizes a comprehensive life‑cycle cost estimate that includes research and development, procurement, operations and support, and disposal, accompanied by rigorous sensitivity and risk analysis. These best practices must be the foundation for any attritable cost model.

Why different models are required

There are three structural reasons that attritable costing differs from conventional costing. First, attrition assumptions become a principal driver of procurement quantity and recurring cost rather than a fringe variable. The expected loss rate per mission determines how many spares and replacements must be produced, and thereby how production rates, learning curves, and supplier ramp up are valued. Second, software, sensors, and COTS components dominate both capability and obsolescence risk. For many attritable designs the hardware may be cheap, but the embedded software, autonomy stacks, and supporting cloud or edge infrastructure carry nontrivial recurring costs. Third, the industrial base and surge capacity algebra matter. A platform that is cheap when built at hobbyist scale may become expensive when hardened, certified, and produced at military scale in a constrained supply chain. Historical analogies mislead unless the model captures these differences empirically.

Concrete parameters to include

Any usable cost model for attritable systems should at minimum model these interacting elements:

  • Unit procurement cost (Pc): baseline factory price for a delivered unit.
  • Attrition rate per mission or sortie (a): probability of loss in the expected operating environment.
  • Production learning factor (L): percentage reduction in unit cost per doubling of cumulative production.
  • Operations and support per unit per year (O&S): maintenance, spares, fueling, software sustainment, training, and transport.
  • Expected useful life or sortie count (U): for reusable attritable assets, the average number of missions until retirement.
  • Software sustainment and cybersecurity amortization (S): annualized cost to keep autonomy stacks current and secure.
  • Industrial base surge cost premium (I): incremental cost to expand capacity quickly or to requalify suppliers.
  • Cost of lost capability or political cost (K): a qualitative multiplier reflecting cases where replacement is politically or operationally harder than monetary cost implies.

A simple, transparent metric is cost per effective mission (CEM). One workable formulation is:

CEM = [(Pc * a) + (Pc / U) + (O&S + S) / (annual sorties per unit)] * (1 + I) + overhead.

This formula treats attrition as a separate recurring procurement driver Pc * a, amortizes baseline procurement over expected reuse Pc / U, and layers on sustainment and industrial base premiums. The model must be run as a stochastic simulation with distributions for a, U, L, and I rather than as single point estimates. The GAO guidance on sensitivity and risk analysis is applicable here.

Putting numbers into context

The category label attritable has been legally and politically contested. The U.S. Air Force previously described a notional range for attritable aircraft of roughly two to twenty million dollars per airframe, a framing that attracted debate across services and Congress. Legislative efforts have also attempted to bracket price tiers for expendable, attritable, and exquisite platforms in the context of collaborative combat aircraft. Those numbers are not universal rules. They do, however, signal the scale at which program offices, warfighters, and lawmakers have been thinking about tradeoffs between cost and survivability.

At the more pragmatic end of the spectrum, some service budget documents show procurement asks that imply much lower per‑unit prices for particular classes of launched effects. For example, an Army FY procurement presentation requested roughly $20.04 million to procure 40 Launched Effects systems, which is on the order of half a million dollars per item. Such divergent figures demonstrate why a single dollar threshold will not capture the operational nuance. Mission context, payload, and expected attrition shape what counts as affordable.

Policy and accounting pitfalls to avoid

1) Treating attritability as a sticker price. A platform labeled cheap may hide a high software sustainment tax or an industrial base premium that only appears under surge. 2) Omitting attrition from the life‑cycle budget. If losses are treated as single year procurement spikes rather than recurring programmatic costs, portfolio affordability will break down. 3) Transferring political costs into acquisition budgets. Replacing a lost high profile asset is more than procurement; it is also training, doctrine, and possibly diplomatic signaling. 4) Using single point estimates. Given the cataclysmic sensitivity to attrition rates and contested environment variables, scenario analysis is mandatory. The GAO cost guide provides the analytical scaffolding to avoid these errors.

Operational metrics that matter more than per unit cost

The correct operational objective is not always to minimize cost per unit. For commanders the salient metrics are cost per effect, time to regenerate capability when losses occur, and mission resiliency under attrition. A $500,000 sensor craft that provides persistent ISR for a brigade and can be produced in thousands with short lead times may be far more valuable than a $3 million platform that is irreplaceable once lost. In this sense attritable systems reframe procurement as portfolio engineering where redundancy, distribution, and rapid manufacture are themselves capabilities. Observers of recent conflicts have argued that inexpensive systems used en masse produced disproportionate operational value; the DoD explicitly cites those lessons in its all‑domain attritable ambitions.

Recommendations for an actionable cost model regime

  • Standardize definitions. The services and OSD should publish a defensible taxonomy for expendable, attritable, and exquisite categories tied to explicit life‑cycle assumptions. Congressional interest in definitional clarity is not theoretical noise; it is budgeting hygiene.

  • Require attrition scenarios in all LCCTs. Every life‑cycle cost and affordability analysis must include at minimum three attrition scenarios: permissive, contested, and degraded. These scenarios should drive procurement rate requirements and industrial base buffers.

  • Incorporate software and cybersecurity as first class line items. For many attritable systems the majority of operational cost will be software sustainment, data routing, and adversary resilience. Those costs must be amortized transparently and funded persistently.

  • Stress test industrial base surge costs. Model the premium to scale production rapidly during crises. Do not assume civilian supply chains will absorb military surge without cost.

  • Report cost per effective mission to decision makers. Present decision makers with probabilistic CEM distributions, not single point estimates. Use CEM to compare the marginal benefit of additional attritable units against alternative investments in survivability for exquisite platforms.

Concluding thought

Attritable systems force a philosophical reorientation in defense economics. Affordability is not merely a unit price. It is a condition created by doctrine, supply chains, software life cycles, and political tolerance for loss. A useful cost model must therefore be interdisciplinary and probabilistic. If the United States is to field systems at the scale the services now discuss, then it must also field cost models that are honest about recurring losses, software sustainment, and the industrial premiums of surge. To do less will be to repeat the error that exquisite acquisitions taught us: rich numbers that hide poor assumptions. The remedy is not austerity. It is rigorous accounting paired with strategic candor.