Affordable Mass and the Replicator Initiative: Lessons from Scaling Attritable Autonomy

The strategic problem is simple to state. The United States and its allies field small numbers of extremely capable systems. A potential adversary fields larger numbers of adequate systems closer to the likely theater. In a sustained conflict, exquisite quality does not compensate for a large enough quantitative disadvantage — particularly when the exquisite systems cannot be replaced within the timeframe of the war.

The Replicator initiative was the Department of Defense's attempt to solve this by fielding thousands of attritable autonomous systems on a compressed timeline. Announced on 28 August 2023 and led by the Defense Innovation Unit, its first line of effort aimed to field all-domain attritable autonomous systems by August 2025.

Its record is genuinely mixed, and that mixed record is more instructive than a clean success would have been.

What Replicator attempted

The initiative's premise was that the Department could not solve the mass problem through its normal acquisition process, which takes years to move from requirement to fielding and is structurally biased toward high-cost, low-quantity systems.

Replicator proposed a different model: identify existing or near-ready commercial capability, contract rapidly through non-traditional mechanisms, and field at scale within two years. Critically, it was framed as a process demonstration as much as a capability delivery — an attempt to prove the Department could move at commercial speed.

Industry engagement was substantial. More than 800 companies participated in Replicator-1 through various onramps including Commercial Solutions Openings, with more than 35 receiving contracts supported by over 50 major subcontractors.

A second line of effort, announced in a September 2024 memo, directed Replicator 2 toward countering small uncrewed aerial systems — a recognition that defense against the threat had become at least as urgent as fielding it.

The results

Assessment depends substantially on what is being measured.

Against the stated goal of thousands of systems fielded by August 2025, the record falls short. The exact number is classified, but credible analysis indicates the initiative fielded hundreds rather than thousands by the target date.

Departmental characterization has been more positive. Testimony in June 2026 described enormous strides toward delivering and fielding multiple thousands of uncrewed systems across multiple domains, with thousands more planned through the FY2026 defense budget.

Both statements can be substantially true if the timeline slipped rather than the objective failing. That appears to be roughly what happened: the capability is arriving, later than promised, at a scale that is meaningful but below the initial ambition.

The more useful question is why, because the reasons are structural and will recur in every similar effort.

Why scaling proved hard

Several obstacles emerged that are worth understanding in detail, because they apply well beyond Replicator.

The production base did not exist. Identifying a promising system is not the same as being able to build thousands of them. Many participating companies were small, with prototype-scale manufacturing. Scaling from tens to thousands of units requires capital, tooling, workforce, supply chain, and quality systems that take time and money to establish. A contract award does not create a factory.

Component supply chains ran through adversary territory. Motors, batteries, flight controllers, and imaging sensors are dominated by Chinese suppliers. Compliance requirements — notably restrictions on Chinese-origin drone components — forced qualification of alternative sources at higher cost and longer lead time. This single issue accounted for a substantial share of schedule slip across the portfolio.

Testing and certification capacity was constrained. Airspace, spectrum, and range access for autonomous system testing are genuinely scarce. Compressing acquisition timelines does not compress the physical availability of test infrastructure. Programs queued.

Requirements were unstable. Rapid fielding assumes a stable requirement. Operational feedback and evolving threat understanding produced changes that reset schedules. This is not unique to Replicator, but it interacts badly with compressed timelines.

Money moved slowly even when decisions moved fast. Appropriations timing, continuing resolutions, and the mechanics of transferring funds between accounts imposed delays independent of any programmatic decision. This is a structural feature of the budget process that no acquisition reform addresses.

The transition problem persisted. Getting from a Replicator-funded prototype to a service program of record with sustained funding, sustainment, and force structure proved harder than getting to the prototype. Analysts have consistently identified this as the point where innovation initiatives fail, and Replicator was not exempt.

Culture resisted. Multiple analyses concluded the initiative's central challenge was cultural rather than technical — organizations optimized for deliberate, risk-averse acquisition do not become fast by being told to. Incentive structures, career risk, and institutional habit all pull toward the traditional approach.

What the initiative got right

The criticism is warranted but incomplete. Several things worked.

It established affordable mass as a legitimate priority. Before Replicator, mass was rhetorically endorsed and structurally ignored. The initiative made it a named, funded, senior-level objective, and that framing has persisted in subsequent programs — visible in CCA Increment 2's shift toward low-cost attritable designs and in counter-UAS procurement structures.

It brought in non-traditional suppliers at scale. More than 800 companies engaging with the Department, and more than 35 receiving contracts, expanded the supplier base meaningfully. Some of those relationships will outlast the initiative.

It demonstrated Commercial Solutions Openings at volume. Proving that a contracting mechanism can move quickly at scale created precedent that subsequent acquisition reform built on.

It generated learning about scaling barriers. The obstacles above are now documented and understood. The FY2026 acquisition reforms — expanded CSO authority, the ability to move successful CSOs into production, support for startups through BOOST, and higher accounting thresholds easing compliance for non-traditional contractors — respond directly to problems Replicator surfaced.

It clarified the counter-UAS priority. Pivoting Replicator 2 to counter-small-UAS reflected accurate reading of the threat and accelerated a capability that has since seen very large procurement, including the roughly $20 billion Army counter-UAS award to Anduril in March 2026.

The affordable mass problem, restated

Stepping back from the initiative to the underlying problem clarifies what actually has to be solved.

Affordable mass requires four things simultaneously, and most efforts deliver one or two.

Low unit cost. Not "low compared to a fighter" but low enough that losing many is acceptable. This constrains sensors, propulsion, and structure in ways that conflict with engineering instincts and with operator preferences.

High production rate. The capacity to build thousands per year, sustained, with the ability to surge. This is a manufacturing capital problem, not a design problem, and it requires investment before demand is certain — which the acquisition system is poorly structured to fund.

Adequate capability. Cheap and useless is not a solution. There is a floor of sensing, range, and payload below which quantity stops compensating.

Sustainable employment concept. Systems that arrive without doctrine, training, logistics, and force structure sit in warehouses. Fielding is not the same as employing.

Replicator focused heavily on the first and third while underestimating the second and fourth. That is the central lesson.

What scaling actually requires

Several conclusions follow for anyone pursuing affordable mass.

Fund production capacity separately from units. The traditional model buys articles and assumes capacity. For attritable systems, capacity is the strategic asset. Multiyear commitments or direct capacity investment are needed to justify the capital expenditure.

Solve the component supply chain deliberately. Sovereign or allied supply for motors, batteries, controllers, and sensors carries a cost premium commonly estimated at two to five times. That premium should be treated as a strategic investment with explicit funding, not as an unfunded compliance requirement that programs absorb through schedule slip.

Design for manufacturing from the start. Systems designed for performance and then scaled encounter production problems that systems designed for production do not. This requires manufacturing engineering involvement at concept, which defense programs routinely defer.

Build the transition path before the prototype. Identify the receiving program, the sustainment plan, and the force structure before demonstrating capability. The demonstration is the easy part.

Expand test infrastructure. Range, airspace, and spectrum capacity for autonomous system testing is a national bottleneck. It requires investment that no individual program will fund.

Accept lower performance deliberately. The hardest institutional discipline. Every performance increment increases cost and reduces quantity. Requirements processes are structurally biased toward adding capability, and resisting that bias requires explicit senior direction sustained over years.

Where this is going

The affordable mass concept has outlived the initiative that popularized it, which is arguably the best measure of success available.

Its influence is visible in CCA Increment 2's reorientation toward low-cost attritable designs following Pacific wargaming, in counter-UAS procurement structured around cost per intercept, in proliferated satellite architectures, and in the naval USV production initiatives moving combat-proven designs into U.S. industrial capacity.

The FY2026 and FY2027 acquisition reforms — drawn from the SPEED and FoRGED Act provisions, expanding Commercial Solutions Openings, easing barriers for non-traditional contractors, and creating pathways from prototype to production — address the mechanical obstacles Replicator encountered. Whether they address the cultural ones is less certain, and analyses of the FY2026 NDAA noted that the most ambitious reform proposals were scaled back in final legislation.

The realistic assessment is that the Department has learned what affordable mass requires and has begun building the mechanisms to deliver it, several years later and at smaller scale than initially promised. That is an unremarkable outcome for an ambitious institutional change effort, and it is meaningfully better than the alternative of never having attempted it.

For industry, the durable signal is clear enough. The customer wants quantity, wants it produced domestically or within allied supply chains, wants it soon, and is restructuring its contracting mechanisms to buy it. Firms that build manufacturing capacity and secure compliant component supply are positioned for that demand. Firms optimized for low-rate exquisite production are not.

The measurement problem

A structural difficulty underlies much of the disagreement about whether Replicator worked, and it will recur in every similar initiative.

Traditional acquisition measures success in delivered articles against a program of record with a defined requirement. That framework does not fit an initiative whose explicit purpose was to demonstrate a process. Did Replicator succeed if it fielded fewer systems than promised but permanently established affordable mass as a departmental priority and drove acquisition legislation? Reasonable people disagree, and the disagreement is largely about what was being measured.

The absence of an agreed metric has practical consequences. Congressional oversight requires accountable numbers. Departmental advocacy prefers narrative progress. The gap between the two produced a public record in which the initiative was simultaneously described as falling short of its targets and making enormous strides, with both characterizations defensible.

For future initiatives, defining success criteria that capture process change alongside delivered capability would improve both oversight and program management. Specific, measurable process metrics — time from solicitation to award, time from award to first delivery, number of non-traditional suppliers transitioned to programs of record, production capacity established — would provide accountability without forcing an innovation initiative into a framework designed for weapons programs.

The broader lesson is that institutional change efforts need to be evaluated as institutional change efforts. Judging Replicator solely on unit counts misses most of what it was attempting and most of what it accomplished. Judging it solely on narrative excuses genuine execution failures. Both errors have been visible in the public discussion, and both make it harder to learn the right lessons for the next attempt.

Frequently asked questions

What was the Replicator initiative? A Department of Defense effort announced in August 2023, led by the Defense Innovation Unit, to field thousands of attritable autonomous systems across multiple domains by August 2025 — intended as both a capability delivery and a demonstration that the Department could acquire at commercial speed.

Did Replicator succeed? Partially. It fell short of fielding thousands of systems by the original deadline, with credible analysis suggesting hundreds were fielded by August 2025. It succeeded in establishing affordable mass as a priority, engaging more than 800 companies, and surfacing the structural barriers that subsequent acquisition reforms address.

What is attritable autonomy? Autonomous systems inexpensive enough that losing them in significant numbers is operationally and financially acceptable. The concept trades individual system capability and survivability for quantity and resilience.

What is Replicator 2? The initiative's second line of effort, announced in September 2024, focused on countering small uncrewed aerial systems rather than fielding offensive autonomous systems.

Why is it hard to produce military drones at scale? Production capacity must be built with capital investment before demand is certain, component supply chains run heavily through China and require expensive requalification, test and certification infrastructure is constrained, and transitioning from prototype to a funded program of record with sustainment and force structure remains the most common failure point.