From Exquisite Platforms to Affordable Mass in the Defence Industry’s Reorganisation Around Drones and Swarms

The Diminishing Returns of the Sixth-Generation Logic

A few years ago, the dominant conversation in defence aerospace circled around sixth-generation fighter programmes—Tempest in the United Kingdom, NGAD in the United States, FCAS in Franco-German-Spanish form, and GCAP as the Anglo-Italian-Japanese successor to Tempest. Each programme promised optionally manned cockpits, low-observable airframes, distributed sensor fusion, and direct interfaces with autonomous "loyal wingmen". The unit economics, however, were always uncomfortable. Each platform was projected to cost in the hundreds of millions of pounds. Each required decades of development. Each would be flown by pilots whose training pipeline was itself measured in years and millions of pounds per aviator.

BCG's 2018 analysis, How AI and Robotics Will Disrupt the Defense Industry, was unusually prescient on this. The firm estimated that even then, more than sixty percent of cutting-edge research and development programmes for unmanned and autonomous technologies were already being captured by non-traditional competitors and smaller prime contractors, rather than the established primes who dominated manned platform development. The implication was clear: the centre of gravity of defence innovation was shifting away from the firms most invested in extracting the marginal performance gain from the next-generation manned fighter, and toward firms building cheaper, faster-iterating, software-centric autonomous systems.

What has changed since 2018 is not the underlying technological trajectory, but the operational evidence. Ukraine has consumed roughly ten thousand small drones per month at the height of its operations. Iranian-pattern Shahed-136 drones, costing around thirty thousand dollars each, have repeatedly forced the deployment of PAC-3 interceptors costing approximately four million dollars per shot. Even at a ninety-three to ninety-seven percent interception rate, the cost-exchange ratio constitutes strategic failure for the defender. The lesson is not that exquisite systems fail technically; it is that they are too valuable to use freely and too slow to replace once used.

The Consulting Industry's Pivot

McKinsey's most recent treatment of this shift, Future Defense Tech: Multidomain Stacks to Build Affordable Mass (2026), provides the clearest articulation of how the consulting industry now frames the problem. The piece argues that the next generation of military advantage will not come from any single platform, but from "multidomain stacks": modular, software-defined architectures that integrate sensing, decision-making, and effects across aerial, maritime, ground, and space domains, populated largely by attritable autonomous systems. The architectural assumption is that no individual node in the stack is irreplaceable; resilience comes from numbers, redundancy, and adaptive coordination, not from hardening any single platform.

This is a substantial reframing of the consulting orthodoxy. For two decades, McKinsey, BCG, and Deloitte built their aerospace and defence practices around helping primes optimise the production economics of major manned platform programmes. The pivot toward affordable mass implicitly concedes that the value pool is migrating elsewhere—toward firms that look more like commercial software companies than traditional defence contractors.

BCG's 2026 publication The New Frontier of Defense Technology and Security, focused on European governments, reinforces this view. It argues that European militaries cannot meaningfully close the capability gap with peer adversaries by continuing to invest in incremental upgrades to legacy platforms. Strategic investment must instead concentrate on emerging technologies where the marginal pound of investment produces a non-marginal capability gain: autonomous systems, AI-enabled decision support, electronic warfare, and the digital infrastructure that ties these together.

Deloitte's 2026 Aerospace and Defense Industry Outlook reaches the same conclusion through a different door. The piece emphasises that defence budgets are now channelling investment specifically into autonomous unmanned systems, drones, and advanced weapons platforms, but flags a critical operational obstacle: most of these new systems must interoperate with legacy platforms already deployed in theatre. Closed ecosystems risk slower adoption regardless of their technical merit. The implication—particularly relevant to anyone working on modular digital twin architectures—is that the architectural openness of an autonomous system may matter more than its peak capability.

The Technical Substance: Why Swarms, and Why Now

The consulting literature provides the strategic framing; the substantive technical and doctrinal analysis comes from elsewhere. RAND's 2024 report on Unmanned Aerial Systems Intelligent Swarm Technology is the most rigorous open-source treatment of why coordinated drone swarms represent a qualitatively different capability from large fleets of individually-operated drones. A swarm is not simply many drones; it is a distributed system in which each agent senses, decides, and acts locally, while contributing to and drawing from a shared situational picture. The behaviours that emerge—adaptive route planning, dynamic task reallocation, coordinated saturation of a defended area—cannot be achieved by centrally controlling many drones because the bandwidth, latency, and operator cognitive load requirements scale faster than the benefits.

CNAS's Swarms over the Strait (2024) operationalises this argument in the context of a Taiwan Strait scenario. The report concludes that hundreds of networked, low-cost drones could guide American long-range anti-ship missiles toward an invading fleet, providing the targeting persistence and resilience that a small number of exquisite ISR platforms cannot. Critically, the report's authors—affiliated with both RAND's modelling community and the US Department of Defense—argue that the United States needs a deliberately heterogeneous drone fleet: a small number of stealthy, high-end systems for the most contested airspace, complemented by a much larger inventory of "good enough" attritable systems whose strategic value lies in their numbers and their disposability.

The Belfer Center's 2025 piece, Autonomous Drones Will Not Replace Fighter Pilots, They Will Augment Them, provides a useful corrective to over-enthusiastic interpretations of the drone revolution. The argument is not that manned aviation is obsolete, but that the optimal force structure is a teaming model in which a small number of manned platforms—still indispensable for the highest-stakes mission elements—operate alongside large numbers of autonomous wingmen that absorb risk, extend sensor coverage, and multiply effects. This is precisely the architecture that GCAP, the US Collaborative Combat Aircraft programme, and similar efforts are nominally being built around.

The Coordination Problem at the Tactical Edge

The technical bottleneck in all of this is coordination under adversarial conditions. A swarm that cannot maintain coherent collective behaviour when its communications are jammed, when individual agents are spoofed or destroyed, and when its sensor inputs are deliberately corrupted, is not a swarm—it is an expensive collection of independent drones. This is where the academic and defence-research communities are converging on a set of hard problems: distributed inference under intermittent connectivity, real-time fusion of heterogeneous and adversarially perturbed sensor data, and the construction of operationally meaningful latent states from indirect observations.

These are precisely the problems that Bayesian graphical methods, dynamic data-driven systems, and modular digital twin architectures are being developed to address. The fact that the consulting industry's strategic framing now converges on the same technical desiderata—modular, software-defined, multidomain, attritable—suggests that the research agenda being pursued in academic settings and the procurement reality being articulated in industry white papers are no longer running on separate tracks. The white papers tell us what the industry is trying to buy; the technical literature tells us what is hard about building it.

What This Means for the United Kingdom

The United Kingdom's position in this transition is more delicate than its public communications suggest. The 2025 Strategic Defence Review committed to increasing defence spending from 2.3 to 2.5 percent of GDP by 2027, with a further trajectory toward the NATO five-percent target. The Review explicitly identified autonomy, drones, and AI-enabled systems as priority capability areas. The Royal Air Force's Rapid Capabilities Office, the Defence Science and Technology Laboratory, and the recently restructured Defence Innovation organisations are all moving in the same direction.

Yet the United Kingdom simultaneously remains committed to GCAP, a programme whose underlying logic is the very sixth-generation manned platform paradigm that the consulting industry is now describing as a diminishing-returns proposition. The tension is not necessarily fatal—GCAP is explicitly designed to operate as the manned node in a teaming architecture with autonomous collaborative combat aircraft—but it does create a difficult prioritisation problem. Every pound spent on GCAP development is a pound not spent on the attritable mass, the software stacks, and the coordination infrastructure that the affordable-mass doctrine demands.

The United Kingdom also faces a structural choice that other European nations are confronting in different ways. The September 2025 Palantir partnership, framed as a 1.5 billion pound strategic agreement to "unlock investment" and accelerate the deployment of AI-powered military capabilities, represents one model: rapid capability acquisition through dependency on a foreign commercial partner. The alternative, exemplified by some of the indigenous European autonomy startups now receiving European Defence Fund and individual member-state backing, is slower but preserves greater technological sovereignty. Neither path is obviously correct, but the choice should be made deliberately rather than by inertia.

The Cost-Exchange Logic and Its Consequences

Underneath all of this is a single piece of arithmetic that has reorganised the strategic conversation. When a thirty-thousand-dollar drone can force the expenditure of a four-million-dollar interceptor, the side with the cheaper replenishment curve sets the operational tempo. This is not a marginal observation; it is the central insight of the affordable-mass doctrine. The implication is that the historical Western preference for exquisite, low-volume, high-capability systems is not merely expensive—it is strategically self-defeating against an adversary willing to absorb attritional losses among cheap unmanned systems.

The consulting literature is converging on this point from several directions. McKinsey frames it as a "multidomain stack" architectural problem. BCG frames it as an industrial-strategy and AI-adoption problem. Deloitte frames it as an interoperability and sustainment problem. The defence think-tank literature, less constrained by client sensitivities, states the underlying logic more bluntly: the side that can produce and sustain attritable autonomous mass will dictate the terms of any prolonged engagement against a peer adversary.

What is striking, reading these documents in sequence, is how rapidly the consensus has hardened. Three years ago, drone swarms were a speculative future capability discussed primarily in academic and DARPA-adjacent literature. Today, they are the organising concept around which the world's most influential consulting firms structure their defence industry advice. This is not a slow drift; it is a discontinuity, and the procurement systems of major Western militaries are not yet organised to absorb it.

Conclusion: A Procurement Problem, Not a Technology Problem

The most useful framing emerging from this body of literature is that the pivot to drones and swarms is not, fundamentally, a technological problem. The relevant technologies—autonomous flight, distributed decision-making, sensor fusion, low-cost airframe manufacturing—exist in mature or rapidly-maturing forms. The constraint is the procurement, integration, and doctrinal infrastructure required to absorb them at scale.

This is why the consulting white papers matter. They are not, on the whole, advancing novel technical claims; they are signalling to their defence-ministry and prime-contractor clients that the procurement logic must change. The accompanying technical literature from RAND, CNAS, and the academic community describes what must be built. The combined message is unambiguous: the next decade of defence advantage will accrue to the militaries that can field, coordinate, and sustain large numbers of attritable autonomous systems faster than their adversaries can.

For the United Kingdom, sitting at the intersection of a manned-platform tradition (GCAP), a partnership-dependent capability acquisition model (Palantir, F-35), and a growing indigenous autonomy research base (DSTL, RCO, university-based programmes), the strategic question is not whether to pivot toward affordable mass, but how to do so without sacrificing the legitimate advantages that the existing platform investments still provide. That is a question the consulting industry has framed but not answered. Answering it will require the kind of deliberate procurement and architectural choices that the next phase of the Strategic Defence Review will need to make explicit.

References