Saturation warfare with drones and cluster munitions is breaking the interception only model, and the replacement is not a bigger shield but a multi domain architecture with damage limitation standing as a co equal pillar.
For four decades, military planners assumed the equation was solvable: identify an incoming threat, assign an interceptor, and treat the engagement as a per-shot engineering problem. That assumption is now failing, not because any single interceptor has broken, but because the threat set has evolved faster than the architecture built to answer it.
The pressure shows up in three recent episodes with named corroborating sources. In October 2023, Hamas fired 5,000 rockets at Israel in an initial barrage — a volume that analysts at the Modern War Institute described as simply too much for Iron Dome to manage. During the May 2025 India-Pakistan crisis, Pakistan's military said it destroyed a deployed S-400 battery using CM-400AKG missiles launched from JF-17 aircraft — the first time the system came under direct attack. And in 2026, Iranian ballistic missiles carrying cluster warheads repeatedly breached Israeli airspace, with the Guardian identifying at least 19 that struck urban areas, injuring at least 15 people in a single incident. Each is a specific kind of lesson: not that the system broke, but that the demand curve outran the supply curve.
The expired assumption, stated plainly, is that an adversary's salvo will look like a textbook. Saturation warfare inverts that. A modern threat package can include cruise missiles, one-way attack drones, loitering munitions, and ballistic reentry vehicles with decoys, all arriving on overlapping timelines. Defending against it per-shot burns through magazines faster than logistics can refill them, and the doctrine that grew up around this approach — the interception-centric model — is now openly under reconsideration in defense planning circles.
What is replacing it is not a bigger shield. It is a multi-domain integration architecture. The components have names: space-based tracking for early custody of cold-launch signatures, AI-assisted target recognition to sort warheads from decoys, adaptive radar networks that can re-task on minutes rather than hours, and sensor fusion layers that hand off tracks between space, air, surface, and undersea nodes. None of those are new as line items in procurement documents, but they have not historically been built to operate as one system. The current shift is to design them as a system, with damage limitation standing as a co-equal pillar next to interception rather than as a fallback when magazines run dry.
Damage limitation, treated seriously, is a different kind of problem. It means hardening the assets an adversary wants to deny, dispersing high-value nodes, building redundancy into command and control, accepting that some salvos will get through, and planning the post-strike picture before the first interceptor fires. Military planning can and must evolve beyond interception-centric models, and this is a design choice by a defense establishment, not an automatic outcome of new technology. The distinction matters: a country can buy the component parts of integrated defense and still field them as a collection of single-shield systems, with the same saturation vulnerability it had before.
The strategic stakes of getting this wrong are concrete. A state that interprets the saturation problem as a call to buy more interceptors, or to build one marquee national shield, will arrive at a more expensive version of the architecture that saturation has already stressed. A state that treats the shift as doctrinal first and technical second will spend the next decade on a different curve: one where the cost of attack rises because defenders see further, decide faster, and accept that no shield is leak-proof, and where the operational job of defense is to keep the state functioning through a strike, not to pretend the strike cannot happen.
The open question worth asking of any national missile defense program that comes up for budget review is whether it is organized around interceptors or organized around integration. The first model is now a known quantity. The second is still mostly a promise on slides, and the next several years of fielded performance are what will separate the promise from the product.