The Marine Corps has a long and proud history. Every Marine knows names like Smedley Butler, Dan Daly, and Opha Mae Johnson — and knows the significance of battles like Chapultapec, Belleau Wood, and Iwo Jima.

In recent years, the Marine Corps has sought to build on this legacy through a campaign of modernization aimed at preparing for the next fight. These changes have sparked lively debate, and much work remains to be done.

To mark the Marine Corps’ 250th birthday, we’re highlighting how the War Quants team has contributed its analytical and writing skills to the ongoing conversation about modernization and the Corps’ future.

Each year, the U.S. Naval Institute hosts the Marine Corps Essay Contest, inviting Marines and civilians alike to challenge assumptions and advance the profession. This year, the War Quants team was well represented—earning second and third place.

Below are summaries of both essays, along with links to the full pieces.

Subscribe now

🥈 Second Place — “Give the King of Battle Wings”

By Sean Harper

In “Give the King of Battle Wings,” Sean Harper makes the case for giving Marine artillerymen new reach and relevance through an Organic Precision Fires–Heavy (OPF-H) capability. He argues that 08XX Marines—the “King of Battle”—are uniquely suited to employ long-range, one-way attack drones that can deliver precision mass at scale.

By merging the traditions of artillery with emerging drone warfare, Harper envisions a future where Marine fires are more mobile, distributed, and lethal. The essay challenges the Corps to evolve its fires community for the era of autonomous warfare—keeping the King of Battle on the throne.

👉 Read the full essay at U.S. Naval Institute

🥉 Third Place — “First to What Kind of Fight?”

By Aaron Barlow, Patrick Moecher, and Caleb Kadrmas

The Marine Corps can no longer afford both a persistent 3.0 ARG/MEU presence and a fully realized Stand-In Force (SIF) posture. Strategy demands choices. In “First to What Kind of Fight?” the authors argue that the current demands for amphibious shipping to support both the legacy MEU and the modern SIF are unattainable within realistic budgets.

Their solution: emphasize littoral mobility for the SIF and integrate that concept alongside traditional amphibious warfare shipping into a bi-modal naval fleet design that distinguishes between global response forces and theater-focused stand-in forces. The essay calls for a coherent amphibious strategy aligned with the Navy’s Force Design 2045 initiative—one that ensures the Corps remains first to fight in the right kind of fight.

👉 Read the full essay at U.S. Naval Institute

As the Marine Corps celebrates 250 years of adapting to change, we’re proud to contribute to its next chapter—one where analysis and imagination combine to make the Corps stronger for the fights ahead.

The ideas shared in these essays reflect the kind of thinking we value most: quantitative insight, analytical rigor, and intellectual humility. Precisely the type of contributions War Quants strives for.

Subscribe now

The views and opinions expressed on War Quants are those of the authors and do not necessarily reflect the official policy or position of the United States Government, the Department of Defense, or any other agency or organization.

1. Sensor performance varies dramatically with environment, latency, and target profile.

No sensor works perfectly on its own; performance modeling helps balance cost, coverage, and constraints.

2. Layered sensor networks with multiple sensor types can overcome environmental and target profile limitations and are crucial for detecting and countering small UAS threats. Radar, RF, EO/IR, and acoustic systems each bring strengths, but integrated fusion delivers superior operational awareness.

3. Commanders need simulation-based planning tools for CUAS defense to understand these trade-offs and sensor requirements.

The Sensor Layer: Seeing the Swarm Before It Strikes

The first step in defeating a drone threat is detecting it. However, the challenge of detecting small UAS (sUAS), which are often quiet, slow, and low-flying, introduces a layer of complexity that traditional air defense systems were not designed to handle. Environmental clutter, such as bird flights, can confuse sensors and lead to false positive detections. Modern counter-UAS systems require a layered sensor architecture that combines multiple modalities, including RF, radar, acoustic, EO/IR, and data fusion, to integrate all these components. Modeling and simulating these systems-of-systems further helps users optimize their employment. This article is part 2 of the War Quants CUAS Primer, and dives into modeling CUAS sensors on modern battlefields.

The Detect, Track, Identify, and Defeat (DTID) Cycle

The CUAS kill chain can be simplified into four basic steps: detect, track, identify, and defeat. These four steps are outlined in Marine Corps CUAS documents and align closely with the F2T2EA targeting cycle introduced in the introduction to this primer. The cycle, summarized below, requires sensors and coordinated data fusion between them to detect, track, and identify sUAS threats.

The DTID process can serve as a basic model for predicting the probability of success in a CUAS engagement. To achieve a successful engagement, a system must first detect the target, establish a track, identify it, and then conduct a successful engagement. Modeling all of these as independent events, each with its own probability of success, allows analysts to derive the overall probability of success, as shown in the figure below.

In the case shown, the CUAS system has a 99% probability of detecting, a 95% chance of tracking, and a 95% chance of identifying a sUAS. When coupled with a 95% defeat probability, the overall chance of defeating a sUAS is 85%. This probability may seem high; however, it corresponds to only defeating six sUAS on average before one gets through. Future CUAS systems will require significantly higher success rates.

Types of CUAS Sensors

CUAS sensors fall into one of two major categories: passive sensors, which do not emit energy, and active sensors, which emit energy. Under these two broad categories, a variety of sensor options are available.

Passive Sensors:

Visual observation remains a foundational element in CUAS operations, particularly in environments where electronic sensors are degraded or unavailable. As emphasized in the U.S. Army’s ATP 3-01.81 Counter-Unmanned Aircraft System Techniques, trained air guards serve as frontline detectors, providing critical early warning and visual confirmation of aerial threats. These observers, equipped with binoculars or spotting scopes, often work in coordination with RF and acoustic detection systems, cueing their optics toward probable threat axes for rapid identification. Human observers are especially valuable in complex terrain such as urban environments or dense vegetation, where electronic line-of-sight is obstructed and drones may fly below radar coverage. Visual observation also offers a zero-emission profile, which is critical in low-signature operations, and enables high-confidence identification of drone type, payload, and intent. However, the effectiveness of visual observation is constrained by daylight, weather conditions, and the cognitive and physical demands placed on personnel. Despite these limitations, visual observation remains indispensable, particularly for small-unit defense and forward-deployed formations that lack advanced sensor suites.

Acoustic detection systems offer a passive, cost-effective method for identifying and tracking drones by capturing their unique propeller and motor sound signatures. These systems utilize microphone arrays to detect these acoustic signatures, which are then analyzed using machine learning algorithms to classify drone types, such as distinguishing between different sUAS. Triangulation techniques, based on the time-difference-of-arrival (TDOA) from multiple microphones, enable the calculation of a drone’s bearing and range. While acoustic detection is particularly effective in quiet, rural environments, its performance can be degraded in noisy or windy conditions.

A notable example of acoustic detection in practice is Ukraine’s deployment of a nationwide network of nearly 10,000 acoustic sensors, known as “Sky Fortress.” Developed by two Ukrainian engineers, this system uses inexpensive sensors—each costing between $400 and $500—to detect the acoustic signatures of incoming drones, such as the Iranian-made Shahed-136. The data collected is processed centrally and disseminated to mobile fire teams equipped with anti-aircraft weapons, enabling them to intercept drones effectively. This low-cost, scalable solution has garnered interest from the U.S. military and NATO allies as a model for affordable air defense strategies.

Electro-Optical/Infrared (EO/IR) systems are line-of-sight sensor modalities that enable visual confirmation, battle damage assessment, and target handoff in a CUAS system. EO systems, typically cued by RF or radar sensors, provide real-time daylight tracking by detecting motion and shape contrast against background terrain. Infrared sensors enhance this capability by detecting the heat signatures of drone motors and onboard electronics, particularly during early morning or nighttime operations. Modern multi-spectral sensor balls mounted on vehicles, towers, and UAS platforms offer stabilized gimbals with automated tracking, often integrating both EO and IR feeds to increase detection probability. Often, systems integrate machine or computer vision into EO/IR video feeds to automate target detection and identification. Weather, obscuration, and poor lighting conditions can challenge detection, while edge computer vision requires significant power and computer processing resources.

Radio Frequency (RF) detection is a cornerstone of modern counter-UAS (C-UAS) systems, offering a non-emissive, passive means of identifying threats. Passive RF detectors monitor common control and telemetry frequencies, such as 2.4 GHz and 5.8 GHz, to detect command-and-control (C2) links and first-person view (FPV) video feeds—especially effective against manually piloted drones. Advanced systems use protocol-aware decoders that can extract metadata, such as drone IDs and model types, even from encrypted links, leveraging proprietary signature libraries and radio frequency machine learning (RFML). RF triangulation networks geolocate both drones and their operators by correlating direction-of-arrival data from multiple nodes, which is critical for identifying swarm launch points or C2 hubs. Additionally, spectrum analysis tools can isolate anomalous RF activity in congested electromagnetic environments to flag potential threats. While RF-based systems excel in long-range early warning, they are less effective against autonomous or tethered drones that emit minimal or no RF signals. For example, the Ukrainian and Russian militaries both utilize fiber-optic-controlled, one-way attack drones that cannot be detected with passive RF.

Active Sensors:

Radar systems provide an essential layer of CUAS capability by detecting airborne threats in all-weather, day-and-night conditions. As noted in Principles of Naval Weapon Systems, radar excels at identifying isolated targets against low-clutter backgrounds. Modern short-range active radars—operating in Ku-, X-, and Ka-bands—can be specifically designed to detect and track small, low radar cross-section (RCS) drones. These systems often incorporate low-Doppler filters to track slow-moving or hovering sUAS, a common signature of Group 1–2 threats. Passive radar, meanwhile, utilizes ambient signals, such as FM radio, television broadcasts, or LTE towers, to detect reflections, enabling stealthy detection in electronically contested environments without emitting its own signals. Despite these advantages, radar performance can degrade in low-altitude cluttered environments (e.g., urban terrain) and is inherently constrained by the small RCS of many drones.

Software & Signal Fusion

Software and signal fusion form the backbone of scalable, adaptable CUAS networks by enabling enhanced coordination across diverse sensor modalities. Fusion engines integrate inputs from radar, RF, EO/IR, and acoustic sources to produce a unified threat picture, improving detection accuracy and reducing latency through automated cross-cueing. These systems prioritize and track threats dynamically, assigning confidence scores and optimizing sensor tasking in real-time. Advanced AI/ML models further enrich the process by classifying drone roles, such as ISR versus attack, based on behavioral patterns, emissions, and trajectory, while also flagging anomalies like swarm pre-launch signatures. This multilayered integration enhances tactical agility and decision speed across echelons, making fusion software a force multiplier in complex environments. However, these benefits come with significant demands: multisensor fusion systems require extensive integration, substantial bandwidth for real-time data transmission, and high-performance edge computing to process inputs without introducing operational lag. Despite these challenges, fusion remains essential for modern C-UAS operations, particularly as threat volumes and complexity scale beyond what any single sensor can manage.

Modeling Detection, Tracking, and Identification

Modeling detection, tracking, and identification in a counter-UAS architecture underscores a key operational reality: no single sensor provides comprehensive coverage against the full range of sUAS threats. The simple DTID model presented earlier can benefit from additional rigor in capturing the complexities of detection, tracking, and identification.

Each modality—RF, radar, EO/IR, acoustic—brings distinct advantages but also inherent limitations. Mission resilience requires a distributed sensor network that can absorb environmental noise, scale with the volume of threats, and adapt to the diversity of drones being fielded. Integration requires both building networks of sensors and managing tradeoffs:

• Latency: The time from initial detection to engagement cueing is measured in seconds, and at short ranges, in milliseconds; delay directly impacts the probability of defeat.

• Signature Tradeoff: Active emitters, such as radar or laser rangefinders, risk exposing friendly positions to adversary electronic warfare systems that can cue enemy fire assets for strikes.

• Environmental Limits: Weather, terrain, and background noise can degrade acoustic and optical performance, especially at low altitudes.

• Coverage and Cost: No single sensor field can provide 360-degree coverage across varied ranges without creating budgetary and logistical burdens.

No C-UAS sensor network is perfect, but they become exponentially more effective when built around principles of redundancy, low-latency cueing, and intelligent data fusion. The faster a threat is detected, classified, and handed off, the more likely a kinetic or non-kinetic response will disrupt its kill chain before it can be executed. In a high-tempo drone environment, timing is decisive.

Evaluating CUAS Sensor Architectures Through Simulation

To understand sensor employment, analysts can use simple detection probability grids to test different sensor configurations and capability requirements. Detection probability grids split the battlefield into quadrants and estimate detection probability based on sensor availability and laydown. Similar to a fire plan sketch, they allow commanders to array their sensors to best mitigate the sUAS threat. Consider the following scenario: two adversaries face off along a relatively static front line defined by a forward line of friendly troops (FLOT) and enemy troops (FLET) with a no-man’s land in between. The adversary force launches an sUAS from behind the FLET to target critical infrastructure behind the FLOT. In this scenario, the friendly force has a passive RF sensor near the FLOT that detects the sUAS as it passes the FLET. That passive sensor then cues an active radar that tracks the sUAS as it moves over the FLOT towards the target point. Finally, a passive EO sensor identifies the adversary’s sUAS, setting conditions for defeat. The three sensors integrate into the CUAS kill web.

We can compare the sensors using probability of detection curves that show how sensor range affects detection. In this case, the curves compare detection starting from the same point for relative comparison.

This example uses sigmoid detection-decay curves, where detection slowly decreases and then drops off quickly, reaching zero at the maximum effective range for illustrative purposes.

Now imagine this scenario at a larger scale with 30 adversary sUAS.

As the number of potential targets increases, the ability of limited sensors to cover every point generally decreases. In this case, 30 sUAS attack 10 targets. While the CUAS system may detect the adversary’s sUAS moving about the battlespace, tracking and identifying those that don’t pass through the radar and EO fields of view will likely be difficult with the single passive sensor.

Toward Smarter Sensing and Faster Decisions

Modern battlefields require more than just better antennas or isolated sensors. They demand an integrated sensing approach that combines various technologies into a single, responsive system. No individual sensor can fully defend against the growing variety of small drones. However, combining sensors through intelligent fusion and simulation-informed planning can close the gaps. As drones become faster, cheaper, and more autonomous, the performance of the sensor network will play a central role in determining outcomes. Smarter sensor placement leads to quicker decisions, and every second counts.

The views and opinions expressed on War Quants are those of the authors and do not necessarily reflect the official policy or position of the United States Government, the Department of Defense, or any other agency or organization.

Small UAS systems pose new threats to tactical units on the battlefield, and the danger is amplified when drones attack in swarms. For tactical ground units, kinetic terminal defense is often the final protective layer in a CUAS architecture, and arguably the most critical. Other countermeasures, such as electronic warfare (EW), can be more widely effective and thus an essential part of the overall architecture. However, capabilities like EW will be subject to the cat-and-mouse game of innovation and counter-innovation of both tactics and technology. When such countermeasures fail, it is essential to have the final protective measures in place—stopping incoming UAS by physically rendering it unable to perform its mission through kinetic force.

Subscribe now

There are a plethora of kinetic CUAS systems and ideas on the market. Solutions range from nets and shotguns to computer-aided aiming devices and AI-enabled machine guns. Some of these solutions, such as shotguns and computer-aided aiming devices, would be employed at the individual troop level. In contrast, others, like an AI-enabled machine gun, would be an asset assigned to protect a unit, such as at the squad, platoon, or company level. The type of solution or mix of solutions chosen may be more or less appealing depending on the tactical unit’s mission and method of employment. However, when selecting the right solution, it is helpful to understand the level of protection provided and make informed trades against other relevant attributes such as size, weight, sustainability, mobility, and magazine depth, among others.

Framing the Problem

Consider equipping a rifle squad with CUAS capabilities. It may be desirable to equip the squad with individual-level protection, such as a computer-aided aiming device, but does everyone in the squad need one? How many is enough? One per fire team or one per person? How does the level of protection of these individual solutions compare to providing the squad with an AI-enabled machine gun? These questions also apply to higher echelons and other tactical formations. This article introduces a quantitative model for addressing these critical questions.

Drone Swarms

Before proceeding, it is essential to define the term 'swarm' in this context. Some definitions of the term 'drone swarm' hinge on the method of control and behavior of the drones, such as full autonomy or drones in the swarm being aware of the other drones' actions and movements, which creates emergent behavior within the drone formation. These distinctions are important, but for the purposes of this discussion, a swarm is defined as any group of drones that fly together in an echelon of two or more. The defender's perspective shapes the definition in this context. When employing kinetic defense, the method of control is of little importance to the defender. Whether all the drones are autonomous or have a one-to-one drone-to-operator ratio is of less consequence than whether your CUAS can engage the volume of incoming drones.

The Model

The nature of kinetic CUAS is reminiscent of naval missile combat, where ships must plan to intercept incoming missile threats. Wayne Hughes’ famous salvo equations eloquently modeled this form of combat, and as such, these equations serve as a starting point for this CUAS model, where the salvo equations are modified to become swarm equations.

In the swarm equations, we are ultimately concerned with the number of attacking drones the defender can shoot down and the number of hits the remaining drones can inflict. The parameters in the base swarm equations are given below:

Based on these model parameters, the number of drones shot down in an attacking swarm, S, is given as:

This leaves the remaining drones to hit their target.

This model enables comparisons between the levels of protection afforded by different systems, and allowing CUAS practitioners to make informed decisions about the tradeoffs in system capabilities. The model can provide insight into considerations such as minimum protection levels, the swarm saturation point of CUAS defense, and the comparative effectiveness of squad versus individual systems. However, the model does not capture the impacts of these systems on other missions, e.g., the mission opportunity cost of an individual carrying an individual CUAS system versus another weapon like a light machine gun.

Some key assumptions of the model are as follows.

• CUAS shooters have perfect target discrimination and can evenly distribute kinetic effects to all targets. This means that if there are more drones incoming than defending systems, the defending systems will engage individual drones without firing on the same drone. However, if there are more defensive systems than drones, there will be duplicate engagements, increasing the probability of intercepting all targets.

• Drones in an incoming swarm arrive close enough in time to be simultaneous to the defender.

  

Application to a Rifle Squad

To demonstrate the utility of the model, we will consider its application in modeling the CUAS systems in a base tactical unit, a 13-person rifle squad. Force planners must make decisions on the level of protection appropriate to this formation. The easy answer is to give them maximum protection, but depending on the CUAS system's size and cost, it may detract from the other missions and functions the rifle squad is intended to carry out.

In this context, let us consider three kinetic CUAS squad packages. Note that parameters chosen for the model are for demonstration purposes only and are not meant to reflect empirical knowledge of system performance.

1. All members of the squad receive a shotgun. Each squad member is capable of firing two shots at an incoming swarm with a 10% chance of hitting with each shot.

2. Each fireteam (three fireteams in the squad) has a member given a CUAS computer-aided firing device on their service rifle. This means that 3 of 13 squad members have a CUAS system capable of engaging 4 drones in an incoming swarm with a 50% probability of kill in each engagement.

3. The squad is equipped with a single AI-enabled machine gun for CUAS. This system can engage up to 10 drones with an 80% probability of kill.

The table above summarizes the model parameters for each CUAS package and the resulting S, or the expected number of drones that the package can shoot down from an incoming swarm.

We can then use this data to see how many drones would still be able to hit their target, H, across various swarm sizes as shown in the graph below.

Keeping in mind that these are notional performance parameters, we can make some observations. For small swarms, all three packages can prevent any drones from getting through. As the swarm size grows past each package’s kill capacity, H begins to rise sharply—first for the shotgun package, then the aiming devices, and last for the AI-machine gun. While the parameters are made up, this simple model provides a way to compare various CUAS systems. To validate actual system performance, empirical data collection from field testing systems and various CUAS packages would be required.

The swarm equation model presented here is intentionally simple: it uses expected values to estimate the number of drones defeated (K) and those that get through (H) based on force size, engagement capacity, and probability of kill. That simplicity makes it easy to compare systems and visualize saturation points, but it also leaves room for refinement.

Further Validation

Validation against real-world tactics is a key next step. The model assumes all defenders can bring their CUAS system to bear on the swarm and that the swarm arrives simultaneously. In practice, dispersion—either of the defenders or the attacking drones—will affect engagement opportunities. A dispersed defending element might not be able to concentrate fire on the same portion of a swarm; conversely, a dispersed swarm might saturate multiple points at once, reducing the local density of defenders.

The deterministic outputs shown here can also be extended to stochastic versions. A stochastic approach would better capture variance and risk beyond just expected values.

Another avenue is to include mixed CUAS packages in the analysis. A rifle squad might combine an AI-enabled machine gun with a few computer-aided aiming devices, or distribute handheld systems across key positions. Modeling combinations would help identify optimal mixes for different expected swarm profiles and cost constraints.

Also, it may be useful to compare CUAS protection at echelon. Layered defense at higher echelons—platoon, company, or battalion—can incorporate multiple engagement zones, from long-range interceptors down to individual weapons. The swarm equation could be adapted to model each layer in sequence, showing how higher-level assets thin the swarm before it reaches the squad’s terminal defenses.

The bottom line: this model isn’t a substitute for field testing, but it is a fast, quantitative way to explore the trade space, set baseline requirements, and guide live-fire validation. Whether deciding how many systems to issue per squad, how to mix different CUAS capabilities, or how to layer them from squad to battalion, the framework helps shift the discussion from guesswork toward data-driven force design.

Subscribe now

The views and opinions expressed on War Quants are those of the authors and do not necessarily reflect the official policy or position of the United States Government, the Department of Defense, or any other agency or organization.

Subscribe now

For context, FY25 mandatory spending for the DoD totaled $22.41 billion, about 92% of which was tied to expenditures for personnel. Of that, most went to retired military personnel through pensions and benefits. In contrast, OBBBA expands the scope of mandatory spending beyond these traditional personnel costs.

The levels of mandatory spending for FY26 were expected to just exceed FY25 levels at $22.5 billion. Over the next four fiscal years, total mandatory spending was expected to total $88.8 billion. Now, with OBBBA, that figure jumps to $245 billion, marking a nearly threefold increase.


Why It Matters

Why should anyone care if funding is mandatory or discretionary? The answer lies in the complex, multistep process that drives the annual approval of discretionary budgets. Here is an abbreviated and simplified version of that process:

1. Service POMs (Program Objective Memorandum): Annually, each service pours blood, sweat, and tears into a detailed five-year spending proposal (the FYDP), laying out its desired programs and priorities.

2. OSD/Joint Staff Review: These proposals are reviewed, adjudicated, and sometimes rewritten during a combative internal cycle.

3. Presidential Budget Request (PBR): The Office of Management and Budget (OMB) packages a final proposal with the rest of the federal budget to send to Congress.

4. Congressional Action: Congress must pass both:

• The National Defense Authorization Act (NDAA), which provides the legal authority to execute programs.
  

• The Defense Appropriations Act, which allocates funds from the treasury to provide actual funding for these programs.

Both the NDAA and appropriations bill must be signed into law by the President. And while the Defense Appropriations bill is, in theory, a separate piece of legislation, in practice, it is generally not passed in isolation. It is instead subject to the political horse-trading in the other appropriations bills that comprise the overall federal budget. In practice, Congress often fails to produce a budget on time. When such failures occur, a continuing resolution takes the place of the new budget, often staving off a government shutdown, allowing programs to continue at the funding levels of the previous fiscal year. Continuing resolutions have the unfortunate side effect of forestalling the commencement of new programs that may have been authorized in the NDAA or preventing the growth of existing programs.

Take a deep breath. Did that abbreviated explanation seem like a lot to take in? If so, then you can inherently understand the importance of increasing the scope of mandatory spending. Mandatory spending is largely exempt from the process described above. The mandatory funds made available in the OBBBA are available to the DoD through FY29 without the need for the annual budget rigmarole. It would require subsequent legislation, signed by the president, to remove this mandatory funding — an improbable prospect.

This is more than a funding gimmick — it marks a structural change in defense budgets. Whether it is a one-time exception to the normal process or if this becomes the new normal is yet to be seen.

However, with the FY26 budget expected to approach or even exceed $1 trillion, the $156 billion provided by OBBBA over the next four fiscal years may seem trivial. This indicates that the traditional role of approving discretionary spending will remain vitally important, and Congress has not ceded all of its power to the executive with a single piece of legislation.

However, for the areas included in OBBBA, they have some political breathing room and fiscal predictability over the next four years. These funding areas generally have bipartisan support, but the clear exception is the $1 billion made available to the DoD for border security and counter-drug missions. This is not an area traditionally handled by the DoD, and it is unlikely that a different administration or Congress would have sought inclusion of this area of funding for the DoD. Still, it is worth noting that this is a modest amount compared to other mandatory defense spending categories.

On the other hand, other spending areas are mostly benign from a partisan perspective. Shipbuilding comes in at number one in terms of fiscal resources at $29 billion, and few in either party would argue with the need for resources in that area. Integrated air and missile defense is another significant beneficiary, although some of this support is tied to the administration’s Golden Dome initiative. However, while this is a pet project of the administration, it has not encountered outspoken or clear opposition.

The most notable inclusion on the list of defense spending areas in OBBBA is Indo-Pacific Command, which stands out as the only geographic combatant command included. The bill specifically made funds available for Operations and Maintenance and for each service’s exercise budget in the region. Activity in these areas is critical to regional campaigning, demonstrating commitment to partners and allies, and deterring aggression by China. The pivot to Asia is now over a decade in the making, and while the United States has not always been faithful to this promise, these priorities are communicated clearly in OBBBA.

Conclusion

From the perspective of defense budgets, it is too early to gauge the full impact of OBBBA. As stated, selected spending areas are likely to enjoy some fiscal and political stability, and perhaps find new efficiencies as a result. It is also possible that this bill has simply moved the lines of inevitable political posturing as Congress accounts for this new mandatory spending in its annual deliberations over discretionary spending. Regardless of the outcome, observers should be aware of the shift that OBBBA presents for defense spending and be keen to learn the lessons that will follow.

Subscribe now

The views and opinions expressed on War Quants are those of the authors and do not necessarily reflect the official policy or position of the United States Government, the Department of Defense, or any other agency or organization.

1. The Department of Defense must develop robust and task-specific tests to enable effective GenAI integration.

2. Not all GenAI models understand military doctrine equally. GenAI performance gaps should be identified and addressed before consideration for defense applications.

3. This article introduces MarineBench, a tailored benchmark for tasks related to Marine Corps doctrine. Such benchmarks will enable the DoD to make more informed development and investment decisions in AI.

Industry is only beginning to scratch the surface of the potential of generative artificial intelligence (GenAI), and the Department of Defense (DoD) is by its nature several steps behind. Industry's ability to harness GenAI is predicated on domain-specific performance, where performance gains can come from the inclusion of tailor-made models or highly scoped and relevant data in model training methodologies. To make the most effective use of GenAI, the DoD must understand the military domain-specific performance of the models it deploys. This article introduces the concept of benchmarks to assess GenAI model performance, discusses the use of GovBench as an existing solution for DoD-specific domain performance testing, and introduces a novel benchmark, MarineBench, as a proof of concept for Marine Corps-specific domain knowledge.

Subscribe now

Benchmarking

By using domain or task-specific tests called benchmarks, we can quantitatively evaluate the domain-specific performance of models like OpenAI’s o3 or DeepSeek’s R1. Standard industry benchmarks typically consist of generalized tasks or knowledge domains such as math, coding, and reasoning. Benchmarks like the one proposed by GovBench have also initiated an important national security community discussion about GenAI testing. As the Department of Defense accelerates AI adoption for warfighting and enterprise processes, responsible AI testing must become standard practice.

Hallucinations and knowledge gaps have defined GenAI since ChatGPT popularized the technology. Despite exhibiting conversational capabilities that resemble interactions with another human, these models do not actually know anything. They generate text by predicting likely word sequences based on patterns in their training data. As an example, if someone says “peanut butter” while discussing sandwiches you would reasonably assume the next words will be “and jelly.” When data is missing or sparse, knowledge gaps emerge. Hallucinations occur when a model confidently produces false or nonsensical information, and such errors remain a persistent issue even in newer models. Returning to the sandwich example, a hallucination might entail the model confidently responding to the prompt, “peanut butter and ___ sandwiches” with “anchovies” or “light post.” While this example is amusing, other real-world hallucinations are far more concerning, such when a model cites fake studies or offers inaccurate medical advice. Benchmarks offer a consistent and reliable way to detect and evaluate these model failures before making models generally available.

GovBench

Consistent benchmarking provides the GenAI community with valuable guideposts to contextualize and quantify performance. GovBench is an open-source project that aims to establish the standard for evaluating GenAI models within government domains. The project recently tested leading GenAI models against their JointStaffBench test, consisting of 4610 questions covering the six major joint doctrine publications. The GovBench benchmarking questions include multiple-choice and free response questions on several Joint Force topics, ranging from personnel management to military communication systems. For example, a multiple choice intelligence question in the dataset asks, “Which principle for multinational intelligence sharing requires alignment with US policy on releasing classified information?” GovBench presents the question and multiple choice answers to the model and records the generated answer using AI Security Institute (AISI) tools, while answers to free-response questions are evaluated for accuracy and relevance by an AISI GenAI grader model. GovBench found that OpenAI’s o3 outperformed the other models in the test with an overall score of 89%.

The JointStaffBench results show that large industry-leading models such as OpenAI’s o3 and Anthropic’s Claude 3.7 Sonnet outperform smaller models from DeepSeek, Meta, Google, and Mistral. Model size, as measured by the number of parameters, matters because larger models have a greater capacity to capture complex patterns and contextual nuances in language, enabling them to perform better on tasks requiring deep understanding or sophisticated reasoning. However, even the seemingly poor performance of Mistral-Small-24B (where 24B denotes the model’s 24 billion parameters) comprised ~3000 correct answers to multiple-choice and free-response questions. Although we don’t have access to GovBench’s results, it is reasonable to speculate that the small model’s performance likely suffered on the 700 free-response questions. 100B+ models like o3 and Claude 3.7 Sonnet perform better on writing tasks as a result of their large training data sets. For example, in the head-to-head comparison between OpenAI’s o4 and the much smaller Llama 3.1-8B model shown below, OpenAI’s o4 model generates a more concise and contextually appropriate answer.

Marine Corps Bench (MarineBench)

The Marine Corps Bench is motivated by the question, “How well do GenAI models know the Marine Corps?” The Marine Corps’ unique culture is anchored in service doctrine, embodied in a series of documents known as Marine Corps Doctrinal Publications (MCDPs). The MCDPs outline the organization’s guiding principles and strategies on topics ranging from warfighting to expeditionary operations. Similar in design to the JointStaff Bench, the MarineBench is composed of 725 multiple-choice questions and answers generated from each of the eight MCDPs. An example question generated from MCDP 3: Expeditionary Operations asks, “Which factor often restricts the application of military force more in military operations other than war?” corresponding to the answer options “A. Technological inferiority,” “B. Limited resources,” “C. Lack of training,” and (the correct answer) “D. Political Concerns.” For testing and dataset refinement purposes, the generated data also contains an answer explanation.

Model selection is just as important as sound benchmarking practices. Selection criteria generally include performance, pricing, compute resources, and security concerns. For instance, a model’s size determines the compute resources and memory required to run the model, and can also serve a proxy for the model’s performance. While larger models generally perform better than smaller models, performance on nuanced and domain-specific tasks can defy this general trend. Additionally, benchmarking smaller and cheaper models can identify cost efficient alternatives for GenAI systems. Given that a smaller model performs well enough for the AI application, an enterprise GenAI system generating 1 billion output tokens a month with Llama 3.1-8b ($0.08/million output tokens) saves ~$95,000.00 annually when compared to OpenAI o3 ($8.00/M output tokens).

A range of leading GenAI models and smaller, cheaper, and even locally run models were tested on the MarineBench. Leading models including OpenAI’s o3, Anthropic’s Claude 3.7 Sonnet, and Google’s Gemini 2.0 Flash showed comparable and slightly better performance when compared to JointStaffBench. The high correctness scores indicate that the models tested were knowledgeable on Marine Corps doctrine.

There are minimal observable performance differences between the five leading AI models on MarineBench, while JointStaffBench’s results, depicted in the first plot, show slightly more breakout among the same models. Without having the GovBench data to confirm, the higher variance in model performance on JointStaffBench likely owes to differences in each model’s free response answers. Assessing model failures is also an important part of benchmarking. On MarineBench, all models scored the lowest on MCDP 6: Command and Control. Low performance on benchmarks can indicate a model knowledge gap but can also be a result of poorly written benchmarks. In the MCDP 6 question set, there are instances of uncommon acronyms and fictional vignette references that appear to negatively impact model performance. All benchmarks, including the MarineBench, need to be iteratively improved and tailored for specific knowledge sets and use cases.

Conclusion

The DoD’s acceleration of AI integration into warfighting and enterprise processes demands tailored evaluation methods like MarineBench and JointStaffBench for quantifying model comprehension of military topics. Leading models have already shown their knowledge on Marine Corps doctrinal content, but future task-specific benchmarks will further test models’ ability to apply their domain-specific knowledge. Just as other military systems and units train rigorously before deployment, AI systems must undergo domain-specific testing before being trusted for defense applications. While MarineBench and JointStaffBench offer a rudimentary starting point for GenAI benchmarking, robust and task-specific tests must be developed to enable effective AI integration.

Subscribe now

Read More of Our Work Here:

Do We Still See the Same Game Board? Schelling Points and Strategic Salience in an Age of Upheaval

The War Quants Counter UAS Primer: Introduction

From Hours to Seconds: Automating Naval Flight Schedules with Data Science

Marine Corps Loitering Munitions: Ten Tenets for Future Employment Modeling Infantry Organic Precision Fires

The views and opinions expressed on War Quants are those of the authors and do not necessarily reflect the official policy or position of the United States Government, the Department of Defense, or any other agency or organization.

The global strategic landscape is shifting. Today’s environment is characterized by reshaped alliances, changing norms, and heightened tensions among great powers. During the Cold War, superpowers were able to navigate similar strategic shifts and ultimately avoid direct conflict, but this was no small feat, and the Cold War’s peaceful outcome was not nearly as inevitable as it now seems in retrospect. The achievement of lasting stability between Cold War competitors was a tremendous feat of statesmanship and anchored by the ideas of the world’s leading strategic thinkers. Thomas Schelling was one such strategist. As an economist and preeminent War Quant, his strategic theory was grounded in game theory. Schelling’s work helped shape the stability of the Cold War, and his major contributions included developing the related concepts of focal points and strategic salience.

This article uses game theory examples to explain focal points (now commonly referred to as Schelling Points) and strategic salience, as well as the overt and hidden factors that enable understanding and salience. These concepts illuminate the foundations of stable competition, and the lessons are of renewed importance in this new era of strategic competition. The Cold War had geographic focal points from Berlin to Cuba, alongside its broader emphasis on maintaining nuclear peace. Today’s competition has its own focal points that range from Ukraine and the broader European security architecture to the de facto independence of Taiwan, and once again, these issues must be managed amongst nuclear-armed states.

Subscribe now

Perspective and Salience: Overt vs Hidden Differences

Strategic salience emerges when multiple actors can view the potential outcomes of a competitive or cooperative game in the same way, or at least anticipate how other actors will perceive the range of outcomes. But finding strategic salience is not always easy. In cooperative settings, the success of coordination depends not just on the structure of the game but on shared perspective. Differences in perspective—whether overt or hidden—can dramatically change what feels like the “obvious” solution.

The following examples illustrate how perspective affects or influences salience in coordination games. In the first instance, cultural context serves as an overt and commonly understood difference that can be leveraged to achieve alignment. In the second, differences in perception are hidden, even from the players themselves, making coordination far more fragile.

Consider a two-person cooperative game in which two players are rewarded for successful cooperation, but cannot coordinate directly. The players are separated, each given three identical balls: one blue, one red, and one white. The players are instructed to assign an order to the balls by arranging them from left to right. The players win if they arrange the balls in the same order as the other player; otherwise, they lose. A payoff matrix for this game is below:

In this game, the Nash Equilibrium is a mixed strategy, where randomly choosing one of the six orderings with an equal probability gives the players the best chance of maximizing their payout. Thus, the game is just as interesting as two players rolling a standard six-sided dice and being rewarded if the rolls produce the same result.

However, what if the players knew something about one another that gave these seemingly arbitrary colors more meaning? For example, in this game, assume both players are from the United States. In this context, these arbitrary colors go from meaningless to highly significant—these are the national colors of the United States and are always referred to in the specific order of red, white, and blue. Given this cultural context, the game shifts from the equivalent of two random dice rolls to easy money for the players.

It is essential to note that the payoff matrix has not changed—solving for optimal strategies in traditional game theory would not favor the ordering of red, white, and blue over others. Certain information (in this case, specific colors) and a shared context completely alter the game. It is this particular information and context that create the prominence of specific solutions, which Schelling referred to as a focal point in the game.

A focal point is a solution in a game that players are naturally drawn to. Schelling's most well-known example of a focal point is when he would ask students to pick a time and place to meet someone that they could not communicate with in New York City. The most common answer was Grand Central Station at noon. The location was chosen for its prominence as a meeting point, and noon was selected as the most prominent time of day. What about other games? Do all have focal points? What if you were to play this meeting game and the city was Paris, France? It is likely that most readers instantly thought of the same location. Now, what about Paris, Texas? In this case, only people who live in and are familiar with Paris, Texas are likely to converge on a prominent location; the rest of us are lacking context. Again, access to information and context is what creates prominence.

Returning to our ball game, would players find focal points in other contexts? What if the two Americans played with different colored balls, like purple, green, and brown? What if you swapped one of the Americans for a French player? Would it change the game if the players knew each other's country of origin? Does every American know the French flag is blue, white, and red? Does every Frenchman know that in America, it is red, white, and blue? Even if both players are culturally aware of the other player's perspective, the game still fails to have a single focal point. These small hypothetical examples demonstrate how small factors that are external to the game and arbitrary from a pure game-theoretic perspective nonetheless limit the level of salience.

The examples discussed above also demonstrate overt differences in perspective. In this game, knowledge about the other player's nationality and culture was critical in creating salience. What makes the differences overt is their apparent influence on the game. In most strategic games, there will be factors about the players that are overt due to their impact on the strategic actor. But overt differences don’t always have obvious implications.

A real-world example of overt factors contributing to real games of strategic cooperation and their hidden implications is apparent in the Biden Administration’s approach to Ukraine following the Russian invasion in 2022. In this instance, the Biden Administration framed the conflict between Russian and Ukraine as a fight between autocracy and democracy. Yet, given this heady framing, the administration received intense criticism for its slow and calculated delivery of specific weapons to Ukraine. This is because, despite President Biden’s rhetoric, the real focal point in the conflict was keeping the struggle below the nuclear threshold. More than once in 2022, the nuclear situation appeared tenuous, with Russia signaling its willingness to employ nuclear weapons if the United States and Europe crossed lines in intervention. This led the Biden administration to make calculated and slow decisions about weapons delivery to Ukraine, with the provision of certain high-end weapons often conditioned on adherence to acceptable use agreements. Thus, it is clear in this instance that while competing, there was simultaneously a game of cooperation surrounding conflict management, with the clear focal point being maintaining nuclear peace. This game is far more complex and dynamic than the simple ball ordering example, and the stakes are much higher.

Hidden Differences in Perspective

Not all differences in perspective are overt. To illustrate this, consider the following game commonly used to demonstrate focal points, but with a slight modification. Two players are once again separated and then shown the following two-by-two matrix. The players are told to choose one of the four squares in the matrix. If they choose the same square, they win. Otherwise, they lose.

In this game, players are naturally drawn to A2 because it is the only square with a unique color. Now, consider a version of the game where the players play four rounds of the above game and are given feedback after each round on whether or not they chose the same square as the other player. However, only a mutually correct answer in the fourth round counts for a significant prize. For the first three rounds, the players are shown the above matrix, and each time, they are informed that they both selected A2—the same square. Now, for the fourth round, they are playing for keeps. In this final round, they are shown the slightly altered matrix below.

Anecdotally, consider the reaction of both players. Player 1 is initially surprised to see a different matrix, but quickly chooses B2. Player 2 is also surprised the game has been changed, and after a longer pause, selects A2. Both players are shocked to hear that they had different answers. After the game is complete, the two players can communicate. A little incensed, Player 1 confronts Player 2.“What were you thinking? We chose the off-color square for three rounds! Then you just ignored the color?”

Player 2 is confused. “In the last round, there was no difference in color for me. So I just chose the same square we had been picking. That seemed like the obvious choice to me.”
An exasperated Player 1 responds, “No color difference? Are you colorblind?”

As it turns out, in this scenario, Player 2 was indeed colorblind. For people with red-green color blindness, the game matrix appears as shown in the image below.

Had player 1 also been colorblind, how might they have played the game differently? What if Player 1 understood that Player 2 was colorblind and therefore understood how they saw the board? Would Player 2’s logic of returning to past solutions have held even as color left the game?

This example illustrates several key concepts in cooperative games and their relationship to strategy. First, a common perspective is powerful and beneficial. Viewing the world similarly will naturally lead to salience in a relationship. Second, understanding the perspective of other players is important for cooperation. Without this understanding, a difference in perspective on what seems obvious may lead to misunderstanding and conflict. In this case, salience is created by strategic empathy. Last, small shifts in the game can highlight a previously hidden difference in perspective. It could be that you thought you saw the world in the same way as those you were cooperating with, but as the colorblindness example demonstrates, unknown biases can create unexpected outcomes and require players to reevaluate each others' perspectives to recreate the salience that once existed.

As a contemporary example, consider the continued divergence of perspective on U.S. involvement in European security. For decades, the United States has been an essential part of the European security architecture, and as a critical member of NATO and the primary guarantor of nuclear security for Europe, this seems unlikely to change. However, perspectives are drifting on the nature of America’s role in European security. As China has emerged as a threat to U.S. security interests, strategists and policymakers have long discussed a pivot to Asia. Initially, this shift reflected calls to end the long wars in the Middle East, but this narrative has recently expanded to include questions about the U.S. role in European security, with the depth of American aid to Ukraine in its ongoing fight with Russia emerging as an example of what is now negotiable. Just like in our matrix cooperation example, this game of cooperation has been played repeatedly for decades, and a focal point in European security has been the United States' involvement as an external balancer. But again, as in our example, the game board has changed, demonstrating that perspectives may no longer align following such shifts. Many in Europe may be angry or confused as the United States, led by the Trump administration, seems to have experienced a sudden onset of strategic colorblindness. However, the Trump administration sees Europe as having long been an enormous beneficiary of U.S. security guarantees and believes that Europe must take on more responsibility for its own security, particularly as the Indo-Pacific becomes the primary theater.

While this shift is characterized by the unique style and rhetoric of the Trump administration and domestic politics, the diverging perspectives are symptoms of structural shifts in geopolitics beyond a single U.S. administration. Significant questions remain as the old focal point of the United States as a balancing power shifts to new realities. NATO’s Article 5 will likely remain a cornerstone for balancing against Russia, but to what extent is America the underwriter of this guarantee? Will other countries like Germany step into a more prominent role? There is also no small irony in Germany stepping up as the defender of Europe, and with history as a guide, not all will be comfortable with this solution, and others, like France, see themselves playing a leading role in the emerging security order. Can the game of European security architecture settle on a new stable focal point?

Conclusion: Lessons for Today’s Strategic Competition

Since the end of World War II, international politics have been shaped by the Cold War, first through its rigid bipolarity, and then through America’s unipolar moment. As this period comes to an end, we are left with the enduring structures that the Cold War left behind. However, the world is rapidly moving on from this legacy, and actors are more willing than ever to seek advantage by discarding norms and previous assumptions that once underpinned stable competition and deterrence.

Revisionist powers like China and Russia clearly look to reshape the international system, but even the United States has now shown increasing flexibility in its interactions with the world order it predominantly shaped. The shakeup does not just apply to how rivals interact, but also how leverage and coercive power are applied to traditional allies. In this emerging world, strategic salience cannot be assumed to be present. As things evolve and the game changes, it is likely that not everyone’s perception of the problem will keep pace or be synchronized. As in our colorblindness example, today’s shifts may reveal previous blind spots in perception mismatch. Whether hidden or overt, the risk of miscalculation between both allies and rivals increases as this perception gap grows, with consequences ranging from the fracturing of an effective balancing alliance to nuclear conflict. Avoiding such perils in the short term is imperative, and in the longer term, stable competition will require the return of strategic salience.

Subscribe now

The views and opinions expressed on War Quants are those of the authors and do not necessarily reflect the official policy or position of the United States Government, the Department of Defense, or any other agency or organization.

1. Repetitive, manual tasks quietly drain time across the military. Since these automation efforts are not “flashy,” they rarely get the attention or funding they deserve—even when the return on investment is clear.

2. In this example, the author developed a Python-based optimization model for a single squadron that automated weekly P-8 instructor scheduling. This automation reduces a weekly 5-hour manual process to seconds and expands the schedule from one day to one week, saving the squadron hundreds of hours annually, not to mention secondary benefits to personnel readiness and morale.

3. Scaled across the fleet, simple tools like this could save millions and dramatically improve resource utilization, showing that pragmatic automation—not just cutting-edge tech—can profoundly impact warfighter efficiency.

   Subscribe now

   

Purpose

The US military has an automation problem… we do not use it. We require humans to complete redundant tasks that software can solve in seconds. Much of today's rhetoric around automation emphasizes warfighting applications of Artificial Intelligence and Machine Learning—targeting, kill chains, and battlefield decision support. However, this focus neglects the competitive edge that could be gained by applying these techniques to the vast set of tasks across the enterprise of outdated staff work and administrative processes. Automating mundane and error-prone tasks can build efficiency at scale, ultimately saving money and freeing warfighters to focus on lethality. This article shares a recent case study that demonstrates the power of automation in addressing these types of problems.

Aviation squadron scheduling offers a clear example of where straightforward automation can deliver immediate gains. Any servicemember in a planning role can relate to juggling multiple Excel spreadsheets and calendars to produce a feasible training plan. In this instance, that task fell to a Navy pilot who was spending at least five hours each week manually building the flight schedule for his squadron. Repeat this scenario across hundreds of squadrons, and the inefficiency at scale is staggering. As a former Operations Research student at the U.S. Naval Academy, this pilot knew methods existed to provide a better way—but this knowledge was buried beneath years of operational experience. From my position teaching data science and operations research at the United States Naval Academy, I collaborated with him to build a solution.

This project exemplifies how relatively simple, practical applications of automation can drastically improve the daily lives of warfighters. Freeing them from repetitive tasks and providing optimal scheduling of personnel and aircraft creates more bandwidth for training, speeds up the pipeline, reduces waste, and ultimately makes them more mission-effective.

Background

All Naval Aviators begin their training at flight school in Pensacola, Florida, where they learn the fundamentals of aviation by flying training aircraft. Upon graduation, they report to a new squadron to train on the specific aircraft they’ll operate in the fleet. For the P-8 Poseidon community, that squadron is Patrol and Reconnaissance Squadron 30 (VP-30) based in Jacksonville, Florida. VP-30 is staffed by approximately 65 Instructor Pilots (Navy Lieutenants) and a few civilian contractors, who teach students through simulator sessions and training flights.

Problem

As in most Navy and Marine Corps squadrons, building the flight schedule is a highly tedious task. Each day of the week requires a comprehensive schedule to enable the squadron’s 65 pilots to meet their training requirements, which might include as many as ~12 flights, ~13 simulator sessions, and multiple classroom presentations. A good flight schedule must also balance a variety of other planning factors, including the number of flights and simulators available each day, discretionary personnel non-availability, pilot seniority, and mandatory constraints like required crew rest.

For VP-30, the officer in the operations section spent approximately 5 hours a week balancing multiple Excel documents just to generate the flight schedule one day in advance.

Subscribe now

Solution

This problem of optimal scheduling is a textbook example from the field of operations research, a discipline that originated in the military and has since been adopted across virtually every major industry. It applies mathematical modeling and computer programming to find optimal solutions to tedious problems like the one faced by VP-30.

In this case, a specific type of mathematical model called an Integer Linear Program was developed to generate the optimal flight schedule. Integer Linear Programs are well suited to solving complex scheduling and assignment problems like those often faced by the military, and many similar projects have leveraged linear programming models to generate immense impacts against real-world operational problems. For example, this thesis project from the Naval Postgraduate School leveraged an Integer Linear Program to assign enlisted Marines to their “best fit” military occupational specialties at the recruit depot – another manual process effectively reduced from hours to seconds in duration. Once adopted, this model decreased the time new Marines spent training before making it to the fleet, and now serves as an example of how mathematical modeling can help deliver value against soft targets like generating talent management efficiencies.

Whether scheduling Naval Aviators or enlisted Marines, an Integer Linear Program offers far greater efficiency and accuracy than manually balancing spreadsheets. In the solution developed for VP-30, the underlying mathematical model—coded in Python—is embedded in a user-friendly Dash web application. Each week, the VP-30 team collects pilot preferences via a Google Form, uploads the responses into the app along with flight and simulator availability, and simply clicks the Run button. Within three seconds, the model produces an optimized Excel schedule for all instructors for the upcoming week. While the interface could benefit from some basic graphic design improvements, the automation dramatically simplifies what was once a time-consuming manual process.

This practical application of automation reduces a five-hour manual process into a three-second process that expands schedule predictability from one day to one week. As of March 3, 2025, VP-30 is using this application weekly.

The Navy and Marine Corps flight school pipeline takes 3-4 years of training before pilots become effective. The compounding effects of operational inefficiencies like the one solved for here unnecessarily lengthen the pilot training pipeline, wasting time and money, and potentially discouraging interested candidates from pursuing careers as pilots. The most immediately measurable impact of this solution is the time saved by the scheduler, but this system is also more flexible than manual methods in cases where a sudden change in resources requires the schedule to be re-generated, such as in cases where a simulator suddenly requires maintenance or pilot availability changes. A flexible, automated scheduling solution ultimately provides greater predictability for pilots when scheduling their training outside of the cockpit, and could ultimately accelerate the pipeline for placing fully trained pilots in operational units.

Conclusion

Simple automations, scaled across units and services, have compounding effects on mission readiness.

It is crucial to leverage both military and industry capabilities to bring the full range of technology to bear on increasingly complex warfighting problems. However, at the tactical and small-unit levels, the daily grind still consists of tasks like building training plans, coordinating resources, and getting the schedule out. These tasks are necessary, but create constant demands on overworked staff, remove cognitive bandwidth from other tasks, and result in the inefficient use of resources.

Automation does not need a tech breakthrough – the tools already exist. Automation is far better suited and much more efficient than humans to handle repetitive, time-intensive processes like those described above. By offloading these tasks, leaders free up cognitive bandwidth to focus on higher-level priorities, such as identifying competency gaps within their units and planning targeted training to address those deficiencies, ultimately increasing combat effectiveness and lethality.

Major Ryan Martinez is a Marine Corps Infantry Officer from Dallas, TX. He graduated from the US Naval Academy in 2015 with a Bachelor of Science in Operations Research. After commissioning in the Marine Corps, he was stationed at 1st Battalion, 5th Marines in Camp Pendleton, CA. He is a graduate of Naval Postgraduate School, where he earned a Master of Science in Operations Research. Currently, he is serving as a Data Science and Operations Research Instructor in the Math Department of the US Naval Academy.

The views and opinions expressed on War Quants are those of the authors and do not necessarily reflect the official policy or position of the United States Government, the Department of Defense, or any other agency or organization.

1. Militaries worldwide are searching for counter-UAS (CUAS) solutions due to the proliferation of small unmanned aerial systems (sUAS) across modern battlefields.

2. To defeat the threat, a scalable, integrated approach to using the CUAS trinity is required at the appropriate unit echelon.

3. The diversity of modern sUAS requires many CUAS technical solutions accompanied by new techniques, tactics, and procedures.

The modern battlefield is no longer defined by dominance in the air, but by survivability under it. CUAS falls under the larger umbrella of integrated air and missile defense, but this used to be a concern for operational commanders and staff. With the emergence of sUAS, air defense in the form of CUAS is everyone's concern, down to the lowest tactical unit. Group 1–2 small sUAS — once dismissed as hobbyist curiosities — have evolved into pervasive and lethal threats. They reconnoiter, target, jam, and kill — often at a fraction of the cost of the systems they neutralize. Evidence from recent and ongoing conflicts continues to mount: sUAS have become an indispensable feature of modern battle.

Yet despite this evidence, many Western forces continue to underinvest in CUAS systems, clinging to legacy capabilities designed for much larger threats. This decision overlooks the new economics of warfare. Cheap drones bypass expensive defenses, and adversaries have learned that volume and velocity can overwhelm even the best air and missile defense technology if it is not scaled and networked.

A new approach is needed — one that is layered, scalable, and formation-integrated. The recent Royal United Services Institute (RUSI) report, Protecting the Force from Uncrewed Aerial Systems, presents a sobering yet clear-eyed view of how sUAS are shaping the fight in Ukraine, emphasizing the urgent need for echeloned sensor-to-shooter chains and soft/hard-kill layering. Simultaneously, U.S. Army-sponsored studies conducted by RAND make the case for treating sUAS as an integral part of future formations while synchronizing both sUAS effects and CUAS defenses.

Militaries must adapt to the emerging threat posed by sUAS. This article is intended to serve as the introduction to the primer, defining key terms, concepts, and frameworks for dissecting the complex sUAS space.

Subsequent articles will be divided into sensing, soft-kill, and hard-kill sections. They will draw from those insights and add additional perspective by integrating operational modeling, kill chain analysis, and exploring layered countermeasures.

I. Understanding the Threat: Small UAS, Big Disruption

The term sUAS belies the scale of the impact of this capability on modern battlefields. sUAS have proven to be disproportionately disruptive and ubiquitous, with Group 1 and 2 platforms weighing less than 55 pounds, and available in high volumes through commercial channels either as complete solutions or modular kits. In Ukraine and in other recent conflicts, sUAS have performed critical roles in reconnaissance, fires coordination, and direct attack. Their small size, low cost, and flexible payloads allow militaries to deploy them en masse, adapt rapidly, and exploit gaps in traditional air defenses.

Taxonomy of SUAS Threats

To build a resilient CUAS strategy, we must first understand the types of sUAS in play. While each system is unique, threats fall into several general categories:

Methods of Operation:

1. First-Person View (FPV)

Often manually piloted via analog or digital video feeds, these drones are used in loitering kamikaze roles or for rapid strike missions.

FPV attack drones have become Ukraine’s signature battlefield innovation, enabling precision strikes on armor, trench systems, and command posts for as little as several hundred dollars per unit. Increasingly, semi-autonomous variants are emerging with basic object tracking or pre-programmed GPS routes, making them harder to jam.

2. Autonomous

These sUAS use onboard computer vision and decision making, AI target recognition, and other methods to find and engage targets. They can operate without a human in the loop, though rules of engagement may require human approval before they can attack.

3. Preprogrammed Route

In this case, sUAS are preprogrammed with a specific mission and a static target. They execute the mission as designed and cannot deviate.

4. Hybrid

The three methods above can be combined and employed at different mission stages or in different environments depending on operational requirements.

Methods of Control:

1. Line-of-sight communications

Often, VHF and UHF radio frequencies, line-of-sight communications limit range but are the most common in tactical operations.

2. Beyond line-of-sight communications

These drones often use satellite or communications relays. Some commercial UAS even operate on civilian cellular network infrastructure. Mesh networks allow relays of UAS and other assets to be combined, enabling an extended area of operations.

3. Fire and Forget. Autonomous and/or preprogrammed

These systems use onboard software and hardware to find, fix, and finish targets without humans in the loop. Humans can be on or in the loops; however, they are technically not necessary, but a choice of rules of engagement.

4. Tethered-fiber optic controlled sUAS

These systems physically connect the UAS to the controller. This connection cannot be jammed, but the physical connection limits their range.

sUAS Missions:

1. One-Way Attack (OWA)

OWA, or kamikaze, drones have munitions onboard and are designed to deliver their payloads by flying into the target.

2. Intelligence, Surveillance, and Reconnaissance (ISR)

These drones are meant to collect information on the battlefield. Their primary payload is sensors.

Systems like the Orlan-10 perform real-time target acquisition, BDA (battle damage assessment), and artillery spotting.

ISR drones often operate persistently at medium altitude, providing fire control for precision strike systems or guiding FPVs.

Their ability to fly extended missions with minimal emissions makes them particularly hard to detect without a well-networked sensor grid.

3. sUAS Bombers/Multi-attack

Ukraine has successfully run an sUAS bomber campaign against the Russians. Colloquially known as Baba Yaga, these sUAS ideally operate for multiple missions, dropping ordnance on static and dynamic targets across the battlefield.

4. Electronic/Cyber Warfare

sUAS with EW payloads can jam, spoof, or act as decoys. A more emerging technology is sUAS, which can fly near IT networks and inject malware/cyber attacks.

5. Logistics and Resupply

Soldiers on foot and vehicles often struggle to get supplies forward on contested battlefields. sUAS can bring small amounts of high-value supplies forward at lower risk to units.

This taxonomy forms a modelable framework for sUAS classification and engagement. Take the drone below from a recent Forbes article. The article covers the recent Russian battlefield innovation of modifying drones with fiber-optic cables to allow operation in electronic warfare-saturated environments where jamming stops radio frequency control.

Using this framework, we would classify its method of operations as FPV, its method of control as fiber-optic, and its mission as a one-way attack. These three different parameters require specific countermeasures to defeat.

Characteristics That Complicate Defense

SUAS aren’t just disruptive because they’re new — they’re disruptive because they invert traditional airpower economics and defense logic.

1. Cost Asymmetry

A $500 drone can destroy a $2 million IFV or trigger the launch of a $150,000 interceptor. This cost imbalance incentivizes volume attacks and forces defenders into difficult trade-offs: overreact and exhaust resources, or underreact and take losses.

2. Size, Altitude, and Flight Profile

SUAS typically fly in the low hundreds to thousands of feet, relatively slowly compared to modern aircraft, and with potentially erratic flight paths. They can avoid radar detection by staying below the radar horizon or blending into ground clutter. Their small radar cross-section (RCS) and low thermal signature further complicate tracking.

3. Rapid Iteration and Modularity

These systems are evolving faster than most defense acquisition cycles. New drone variants appear weekly on the battlefield, often built with commercial parts, custom payloads, and open-source flight software. Modularity allows adversaries to modify drone capabilities. With a new radio or software, a drone you can jam today may be unjammable tomorrow.

II. The Targeting Cycle: How Groups of sUAS Hunt and Kill

Small UAS do not operate in isolation—they form part of an evolving, decentralized kill web, executing coordinated tasks in the targeting cycle with speed and precision. This cycle, from traditional military doctrine, unfolds in six core stages: Find, Fix, Track, Target, Engage, and Assess.

In practice, this cycle is now often distributed across multiple drones. An ISR quadcopter might “find” a target, a relay drone may help “fix” its position, an FPV drone will launch to “engage,” and a secondary ISR asset confirms the “assessment.” This separation of roles enables massed, asynchronous attacks, overwhelming defenders who react only when the terminal drone is too close. Key vulnerabilities will be covered in later installments of this primer as they form the basis for defeating sUAS.

III. The Counter-UAS Trinity: Sensor – Soft Kill – Hard Kill

Effective counter-UAS defense relies on a layered approach, which we call the Counter-UAS Trinity: Sensing, Soft-Kill, and Hard-Kill. First, drones must be detected and classified using a network of sensors — radar, RF, EO/IR, and acoustic — to enable rapid cueing. Once detected, defenders may attempt a soft-kill by disrupting the drone’s control, navigation, or sensors using jamming, deception, or cyber tools. A soft-kill does not physically destroy the drone, but renders it useless. If soft-kill fails or isn’t feasible, hard-kill options — including lasers, bullets, missiles, or interceptor drones — must physically neutralize the threat before it reaches its target. Hard-kill options physically destroy the drone, rendering it unusable. Success depends not on any single layer, but on the speed, integration, and adaptability of all three working in concert.

The proliferation of sUAS demands a structured, model-driven response. This primer has defined a taxonomy for classifying threats by operation, control, and mission. The next step is to apply this taxonomy to each layer of the CUAS Trinity — sensing, soft-kill, and hard-kill — and model their performance, limitations, and integration across echelons. We can only design effective, scalable CUAS systems that keep pace with the threat through layered analysis.

Read More of Our Work Here:

Rearmament Along the Rhine

Damn the Torpedoes: Smaller, Cost Efficient, Less-than-Lethal Systems

Precision Mass at Sea: A Preliminary Analysis

Capability Analysis: AI Machine Guns for Drone Short-Range Air Defense

The views and opinions expressed on War Quants are those of the authors and do not necessarily reflect the official policy or position of the United States Government, the Department of Defense, or any other agency or organization.

• Winning the undersea fight requires robust acoustic surveillance.

• Waiting for conflict will be too late. Effective AI-enabled surveillance capabilities require data that must be collected now.

• Traditional submarines and small autonomous vehicles will coexist in the future subsurface environment. To train effective AI models, datasets need to include the widest possible variety of targets.

  ---

The still scenes of the ocean below the surface seem to be a quiet and peaceful place with soothing white noise. But this soothing quiet is turned into a cacophony by all forms of human maritime activity – and a navy that can effectively detect this activity will have an enormous warfighting advantage. The challenge is making sense of the noise. This problem’s difficulty is dramatically increased by the myriad of small, electric unmanned surface and autonomous underwater vessels that stand ready to prowl contested waters. Fortunately, as the development of seemingly signatureless unmanned watercraft has advanced, so too has the means to detect such threats.

Recently, undersea sensor packages equipped with artificial intelligence (AI) have been developed with the singular purpose of cutting through ambient noise to locate and identify adversary maritime threats. However, AI does not create itself, but rather relies on massive amounts of data to train models. Like a pre-season game hunter setting up trail cameras, an extensive network of acoustic sensors must be proliferated to establish a baseline pattern of activity and data library for AI model building. In this article, acoustic sensor employment considerations and the technical constraints of AI models are presented as key motivators to establishing sonar networks as soon as possible.

Subscribe now

SOSUS Legacy Echoes through Time

The US Navy has been capable of listening to adversary naval contacts for decades. During the Cold War, the Sound Surveillance System (SOSUS) could detect and locate numerous undersea targets and anomalies. For example, in the 1960s, the SOSUS community detected the sinking incidents of the USS Thresher, USS Scorpion, and the Soviet K-129. The SOSUS originally focused on detecting Russian subsurface threats transiting the Greenland-Iceland-UK gap, but proved valuable in other areas of operations. While the methods were classified at the time, we now know the operators and analysts in the SOSUS network relied on “LOFARgrams” for crucial acoustic information. Low-Frequency Analysis and Ranging (LOFAR) grams, like the one shown below, used a combination of black bars and spaces to represent acoustic sounds.

Fixed sensors in the SOSUS network transmitted acoustic collections to listening stations where operators printed the LOFARgrams on large spools of paper. Exploiting LOFARgrams for intelligence purposes was an arduous human task requiring analysts to carefully interpret the abstract markings on these spools of paper and apply subject matter expertise to identify anomalies. In addition to these time and labor intensive exploitation requirements, the SOSUS network had widely known fixed locations, making it unsuitable for future conflict.

Today’s AI models1 such as neural networks also render such manual interpretation of acoustic detection data obsolete. When properly trained, AI acoustic detection models outperform humans at an astonishing rate. Advanced AI-enabled software packaged in a hardened, versatile, and mobile form factor is table stakes for the next generation of acoustic sensor networks, whose chief advantages include global deployability, wireless communications, and foundational AI algorithms that will be responsible for identifying acoustic anomalies. Similar AI models developed for hobby applications like the Merlin Bird Identification phone app demonstrate the type of inputs required to build such models, and how these models can be operationalized.

What We Can Learn from Bird Watchers

Merlin Bird ID, the popular bird watching phone app from Cornell Lab, gives us a glimpse of the functionality, data requirements, and software that are foundational to acoustic collection networks. The application helps users identify birds by their calls using a specially trained model that predicts a bird’s species from user-submitted sound recordings. While the underlying model is invisible to the user, the application processes and interprets the recording input like an AI version of a SOSUS analyst. When a user uploads a sound recording, the recording is converted into a visual representation of sound waves over time called a spectrogram, a closely related cousin of the LOFARgrams created by SOSUS. The spectrogram serves as the primary input to the application’s AI model, which analyzes the spectrogram for anomalies and outputs the probability that the presented sound originates from a specific bird species.

The Merlin Bird ID is powered by a deep convolutional neural network model called Sound ID. As asserted before, high quality data for AI model training is foundational to performant neural networks. Without good data, these models are subject to the “garbage in, garbage out” principle. By extension, without quality undersea data, AI-enabled undersea acoustic sensors will miss threats and create exploitable situational awareness gaps. According to the Sound ID website, Cornell Lab claims that Merlin Bird ID can correctly identify 458 different birds, but only after training the underlying Sound ID model on 266 hours of unique bird calls and environmental audio data. In an ideal world, sensors like Anduril’s Seabed Sentry or Optics11’s OptiBarrier would boast similar identification capabilities, correctly identifying every surface and subsurface threat that an adversary operates. However, the time, labor, and resources required to collect, tag, and aggregate 266 hours of threat acoustic signatures will pose an extreme challenge. The next generation of stealthy autonomous unmanned vessels like the Chinese HBU-001 and Anduril’s Copperhead AUVs will project less noise into the water column, resulting in smaller acoustic signatures. Furthermore, the challenge of detecting quiet undersea threats becomes impossible if acoustic sensors are not emplaced and exploited before competition turns into conflict.

Layered, Deployable, and Connected Networks

Undersea surveillance networks must be layered, deployable, and connected to ensure actionable information supports timely defense operations, and several recent industry innovations indicate an acknowledgement of this reality. For example, European university researchers have recently tested a new sensing technique called Distributed Acoustic Sensing (DAS) that utilizes existing fiber optic sea cables to detect seismic and acoustic signatures. The DAS technique also benefits from near real-time reporting, since the sensors are the cables themselves and are therefore already low-latency and hard-wired to a shore station.

Some fiber optic sensing companies like Optics11 have advocated that new defense-focused fiber optic cables should be emplaced around routinely targeted undersea infrastructure like the Baltic Sea data communication cables. Optics11’s fixed seabed monitoring system, OptiBarrier, advertises a hydrophone-based system that can detect the complete array of undersea threats from divers to submarines. OptiBarrier breaks effective system ranges into three envelopes with associated threat detection objectives from <5km, <30km, and >100km.

Lastly, the aforementioned Seabed Sentry is an AUV-delivered sensing option that advertises active and passive sonar, months-to-year persistence, and an acoustic communications relay capability. None of these systems adequately address the undersea sensing gap on their own, but layered together, both fixed and mobile sensing options start to address the challenge.

For example, layering fixed and mobile undersea sensing assets in a Baltic undersea infrastructure defense scenario could achieve redundancy, deployability, and persistence. To illustrate this concept, the minefield planning simulation from the Damn the Torpedoes: Series on Naval Mining Operations is repurposed to model an acoustic sensor field that collects in the vicininty of the undersea cables connecting Helsinki and Tallinn. In the figure below, fixed systems are depicted as the primary protective assets while mobile systems like the Seabed Sentry are conceptualized as flexibly deployed systems.

This scenario anticipates integrating systems similar to the Optics11 OptiBarrier with a cable-based DAS capability to establish a persistent and fixed sensing capability. Fixed systems will also provide the foundational AI model training data that will ultimately enable more complex threat detection models. Meanwhile, mobile systems should provide flexible surveillance options and be leveraged to enlarge or target specific collection areas and events. Much like the SOSUS, the locations of fixed assets established near critical infrastructure will be known and avoided by adversarial actors in the region. Therefore, unfixed assets must be creatively employed to collect and survey the other areas of interest.

Sense and Make Sense First

Undersea acoustic surveillance is not a luxury or grand advantage in today’s competition or tomorrow’s conflict. The next conflict will integrate exquisite autonomous platforms that evade traditional detection. Only by deploying and training layered, AI-enabled acoustic sensor networks now can we ensure our forces are prepared. The SOSUS proved the value of undersea surveillance and continues to provide foundational lessons to the community. The lesson today is clearer still: deploy late and your network is worthless.

Victory will belong to those who sense and make sense first.

Subscribe now

Read More of Our Work Here:

Rearmament Along the Rhine: Part 1 of a Series on Rebuilding the German Military

Precision Mass at Sea: A Preliminary Analysis

Gone MAD: How the Golden Dome Threatens the Logic of Nuclear Deterrence

One Million Suicide Drones with Chinese Characteristics

The views and opinions expressed on War Quants are those of the authors and do not necessarily reflect the official policy or position of the United States Government, the Department of Defense, or any other agency or organization.

Subscribe now

Capabilities and Costs

Like all weapons, loitering munitions come in a variety of different sizes and configurations, with immense cost differences between them. For our analysis, we will examine two different loitering munition "tiers." These are definitions created specifically for this article, primarily categorizing loitering munitions by warhead weight, which will be important to our analysis. This will allow for better ease of reference than the U.S. military's Group 1 through 5 UAS categories, which include maximum takeoff weight in their definitions, but do not account purely for payload. Most loitering munitions fit into the DoD's Group 1 and 2 definitions, with a few, such as the Shahed-136, bleeding into the characteristics of Group 3 UAS systems.

For our analysis, Tier 1 loitering munitions encompass small systems that carry fairly light warheads, usually weighing less than 10 pounds, and in land combat are generally found at the squad and platoon levels. Tier 2 loitering munitions are larger systems that carry heavier warheads weighing between 30 and 40 pounds that are often used for higher-level operational engagements, such as strikes on critical infrastructure. There are, of course, loitering munitions that carry even heavier warheads such as the Israeli Delilah, which are not discussed as part of this analysis since their capabilities bleed closer into the domain of cruise missiles.

The chart below categorizes the two tiers that will be examined in this article. The cost was calculated by taking the average, rounded to the nearest thousand, between the estimated costs of the three different systems listed in each of the example columns. While not perfect, these estimates provide us with baseline costs of the correct order of magnitude to conduct preliminary analysis

Both Tier 1 and Tier 2 loitering munitions cost significantly less than virtually all missiles employed by modern ground forces. Most anti-tank weapons cost anywhere between 1 and 200,000 USD (e.g. the American BGM-71 TOW and FGM-148 Javelin cost $218,000 and $351,000 respectively), while rocket artillery and tactical ballistic missiles such as the Guided Multiple Launch Rocket System (GMLRS) and Army Tactical Missile System (ATACMS) can cost in the hundreds of thousands to over 1 million dollars. Significant amounts are therefore saved through the use of loitering munitions when they are able to supplement traditional missiles and rockets. The question then is whether the same cost-effectiveness can be replicated at sea.

Naval Strike Salvos and Loitering Munitions

Modern naval combat fundamentally revolves around salvo fire. This has important implications for the role that loitering munitions will play in maritime conflicts. Loitering munitions offer the ability to mass fires at lower costs than most anti-ship missiles. However, the sheer volume needed to actually take warships out of action generally makes the cost benefits of the use of these systems lower than they are on land. We can illustrate this using the "cube root rule" which holds that the number of thousand-pound bomb equivalents needed to catastrophically kill a warship is equal to the cube root of 1/1000th of the ship's displacement in tons. For example, if a ship displaces 5,000 tons, roughly 1.7 1,000-pound bombs, or 1,700 pounds of ordnance, would be needed to take the ship out of action.

The chart and table below show the number of loitering munitions in both tiers that would be needed to destroy five different warships given their displacements. These displacements are meant to broadly reflect a rounded displacement of most warships classified as such, and do not necessarily reflect the displacement of every single vessel labeled a patrol ship, corvette, frigate, destroyer, or cruiser by the various navies of the world.

We can see from the chart that it takes significantly more Tier 1 than Tier 2 loitering munitions to destroy each warship due to the much smaller size of the Tier 1 loitering munition's warhead. In an actual engagement, these numbers would also have to be significantly increased in order to account for air defenses and other countermeasures that the opposing warships would undoubtedly employ. Nonetheless, we can use these numbers to compare the relative costs of the loitering munition engagements compared to traditional anti-ship missiles. For this, we will use two common ASMs in the American inventory: the Harpoon anti-ship missile, mounting a 488 lb warhead and costing roughly $1.4 million, and the Long-Range Anti-Ship Missile, mounting a 1,000 lb warhead and costing roughly $3 million. The number of systems needed in each engagement are provided in the graph below:

The number of anti-ship missiles needed to take out each warship is significantly smaller than the number of either tier of loitering munition. We can then use the number of loitering munitions needed in each engagement and compare the associated cost of their employment with those of the three anti-ship missiles. These costs are shown in the chart below:

The most expensive engagement was the Tier 1 loitering munition salvo. At the cruiser level, we see a 49.6 percent decrease in cost between the number of Tier 1 loitering munitions needed to destroy the warship compared to the number of LRASMs. In other words, while an individual Tier 1 system costs significantly less than a single LRASM, the sheer volume of Tier 1 systems needed to destroy a warship ultimately makes the LRASM significantly more cost-effective as an anti-ship weapon system, at least in theory (more later).

However, the Tier 2 salvo stands out as the most cost-effective option of all the ones listed. Of course, this only accounts for the hypothetical number of systems that would be needed to take one of the warships out of action. In an actual engagement, the number of these systems would have to be significantly increased to penetrate through adversary defenses while still bringing sufficient firepower to bear against their targets for a successful strike. While gauging the "real" number of loitering munitions needed in such an engagement is impossible, we can obtain a rough estimate using Captain Wayne Hughes' Salvo Equations, which provide a simple model for estimating the number of losses that two opposing naval forces would endure in a missile exchange. In this case, we will apply Colton Byers' methodology and use a single Force B equation, substituting loitering munitions for missiles:

For our example, we will apply the equation to a single warship of frigate size, which fires 20 air defense missiles against a Force A salvo of 52 loitering munitions (i.e. the exact number needed to theoretically take the ship out of action). We will assume that each loitering munition is "well aimed," meaning that any munition not downed by Force B's air defenses will hit its intended target. Flight tests indicate a roughly 80 percent effectiveness rate across all American air defense weapons, which we will use as the effectiveness for Force B's defenses. Plugging these into our equation, we obtain:

Practically speaking, this means the ship survived the salvo, though its staying power is reduced after the engagement. Increasing the number of loitering munitions in the salvo would increase the chances of sinking the Force B ship, but this would present similar problems to those we had with the Tier 1 systems, namely that cost savings diminish as the number of systems increases. For example, if we were to expand the size of the salvo by just half, for a total of 78 loitering munitions, the cost of the engagement would come out to $5.2 million, slightly lower than the cost of the LRASM engagement, and on par with the Harpoon engagement.

However, as was also noted, the frigate's staying power was substantially reduced by the loitering munition salvo, despite the fact that it was not actually sunk. Had heavier ordnance also been involved in the engagement, the ship would have likely been destroyed, given its need to deal with such a substantial number of threats coming towards it. We can demonstrate this using a modified version of the Salvo Equations that includes two types of weapons in a salvo. Because the amount of ordnance on each weapon differs, we will express the number of weapons fired in terms of ordnance to account for this. For example, if each ship in Force A fires two LRASMs, each with a 1,000-pound warhead, then we would write this as 2,000. The modified version of the Force B equation is thus given as:

We can return again to the frigate example. Force A fires a salvo of 35 Tier 2 loitering munitions (i.e., 1,400 lbs of ordnance at a cost of $2.6 million) and 2 LRASMs (i.e., 2,000 lbs of ordnance at a cost of $6 million). The total cost of the engagement would amount to $8.6 million. Let us assume that of the sixteen air defense weapons fired by Force B, 1 manages to hit an LRASM (i.e., 1,000 lbs of ordnance), while the rest intercept loitering munitions (i.e., 525 lbs of ordnance). The cube root rule tells us that it takes roughly 1,817 pounds of ordnance to destroy the frigate. Plugging the values above into our modified equation, we obtain:

Our results have manifested as a positive value which in practical terms means that the frigate was sunk in the salvo. Because modern conventional warships are highly visible on radar and emit a large electromagnetic signature, they can successfully be targeted by strike salvos. The large volume of targets that the warship then needs to intercept once it is identified in turn presents a significant challenge to the warship's air defenses.

While the $8.6 million cost is higher than the theoretical number of LRASMs needed to destroy the same ship, we must remember that naval combat occurs in salvos. If a frigate armed with 20 anti-air missiles is capable of shooting down 16, then at least 18 LRASMs would be needed in the engagement, coming out to a cost of $54 million. If we assume that the LRASM’s advanced features, such as stealth technology, enable this salvo size to be reduced to, say, five or ten, the price still amounts to $15 and $30 million respectively, all significantly more expensive than the $8.6 million cost of the engagement which included a lower number of LRASMs and a large volume of loitering munitions to complicate targeting.

We thus see that loitering munitions hold the potential to be a powerful force multiplier for the Navy's anti-ship missiles. A strike salvo involving a medium-sized number of loitering munitions, along with a sufficient number of anti-ship missiles needed to actually sink an opposing force's ships, could present significant targeting challenges for defensive systems, alongside those already offered by stealth technology on weapons like the LRASM.

In addition to presenting new cost-effective targeting challenges for enemy warships, loitering munitions could also give way to new tactics, techniques, and procedures for the employment of precision strike capabilities at sea. Whereas anti-ship missiles like those described in this article almost exclusively aim to simply strike and destroy enemy vessels, the extreme accuracy of loitering munitions could pave the way for tactics that attempt to target specific parts of a warship. Loitering munitions have allowed both Russia and Ukraine to target each other with pinpoint accuracy. Loitering munitions equipped with AI-enabled precision guidance could well be used for the same purposes, specifically targeting an enemy warship's bridge, VLS cells, antennas, and other vital components. Thus, while not necessarily being able to destroy the warship outright, a loitering munition-only salvo could be used to take the warship out of action in other ways.

Implications for the U.S. Navy

Loitering munitions hold the potential to significantly impact naval warfare. While they likely do not present an outright threat to most warships unless procured in large numbers, their ability to overwhelm air defenses at relatively cheap costs is significant. With an eye towards exploiting the capabilities offered by loitering munitions while also defending against adversary use of them, U.S. Navy leaders should:

• Aim to acquire large numbers of "middleweight" loitering munitions with sufficient range to operate alongside anti-ship missiles and large enough warheads to present at least some degree of threat to enemy warships while remaining small enough to be procured at low costs.

• Adopt capabilities that can overcome adversary loitering munition swarms while saving more advanced defense systems for use against actual anti-ship missiles. The new Coyote and Roadrunner interceptor systems present one such capability.

• Examine new tactics and operational concepts centered on combined anti-ship missile and loitering munition salvos.

War at sea is changing. The Navy must change with it.

Subscribe now

Luke Widenhouse is a research assistant at the Yorktown Institute. His research at Yorktown focuses on naval strategy, emerging technologies, and the defense budget. He is the author of multiple articles and monographs on U.S. national security policy and great power competition. He has previously been published in RealClearDefense and as a guest author for the Center for Strategic and International Studies.

The views and opinions expressed on War Quants are those of the authors and do not necessarily reflect the official policy or position of the United States Government, the Department of Defense, or any other agency or organization.

1. Germany is embarking on its fourth major military rearmament in the past 125 years.

2. Berlin plans to invest over $1 trillion over the next decade, potentially expanding the Bundeswehr from 180,000 to more than 400,000 personnel by 2035.

3. A shrinking military-eligible population may increasingly challenge traditional force generation in Germany and Europe as a whole. Autonomous systems may offer a path to a more capable, less manpower-intensive force.

Germany’s war machine showed early signs of sputtering back to life last month with the announcement of a sweeping $1 trillion rearmament plan. If executed, this plan would mark the most significant transformation of the Bundeswehr since the Cold War, positioning Germany at the center of a broader European defense resurgence in response to a shifting geopolitical landscape.

Historic Context

Since unification in the late 19th century, Germany’s military posture has followed a cyclical pattern of rapid buildup and subsequent demobilization. Across three distinct eras — from the last Kaiser, the Third Reich, and the Cold War — Berlin has demonstrated a capacity for fast mobilization, enabled by a deep industrial base and strong state coordination. The current rearmament effort fits within this historical rhythm yet diverges in key ways, particularly in its democratic governance, integration within NATO, and emerging reliance on expensive, advanced technologies over mass mobilization.

Subscribe now

The most recent demobilization began with the end of the Cold War, as Germany pivoted from territorial defense to international peacekeeping and stability operations. Over the following decades, the Bundeswehr was reduced from more than 500,000 active personnel to 182,000 today and now represents the 28th largest military globally. This downsizing was reinforced by the suspension of conscription in 2011, which signaled a broader shift toward a leaner, professional force. Notably, this military contraction occurred alongside substantial economic growth, with Germany emerging as Europe’s largest economy, highlighting the growing gap between Germany’s economic weight and military capacity.

While the Bundeswehr shrank in size, German defense spending gradually increased in absolute terms. However, this funding was disproportionately directed toward procuring high-cost, technologically advanced systems and constructing a smaller professional force rather than expanding overall military capacity. As a result, Germany’s military became increasingly expensive to maintain, and yet often lacked deployable readiness and depth. These shifts mirrored broader trends across Western Europe, with nations investing in quality over quantity even as emerging threats began to require both.

To quantify how Germany’s defense spending has evolved for modeling purposes, I have calculated Germany’s defense budget allocated per uniformed service member, shown in the chart below. In 1989, at the tail end of the Cold War, Germany spent the equivalent of $73,000 (adjusted to 2023 dollars) per active-duty member. Today, that figure has ballooned to nearly $350,000. To forecast future military spending across the coming decade, I have also leveraged time-series exponential triple smoothing. Triple smoothing accounts for long-term trends, seasonal variation, and recent accelerations in expenditures. If current trends persist, per-capita military spending could exceed $430,000 by the 2030s.

Many factors contribute to this growth in cost per uniformed person, including the increasing technological complexity of military capabilities and the growing challenges of training and retaining qualified personnel in an increasingly complex military ecosystem amid global demand for technical professionals. For example, the financial incentives necessary to build and maintain competitive human capital may be driving rising personnel costs. The Bundeswehr is also an all-volunteer force in an era with a low propensity for national service.

Future Plans: Scenario-Based Forecasting

To model the Bundeswehr's potential future, I developed a 13-year scenario-based analysis, anchored on defense spending per uniformed service member. Using known 2023 defense expenditures, Germany’s 2025 draft budgetary plans, and historical spending data from 1989 to the present, I have estimated 2024 spending figures and constructed three forecasts through 2035, each based on a different trajectory of political will, fiscal prioritization, and defense commitment.

Scenario 1: NATO Target, Minimal Impact (Worst Case)

Germany meets the NATO benchmark of 2% of GDP in defense spending but significantly dilutes the widely publicized $1+ trillion supplemental rearmament package. A large share of these funds are diverted to non-military priorities, in the same way that the German Green Party recently reallocated $100 billion from defense funding for environmental initiatives. In this scenario, the Bundeswehr remains under-resourced and fails to meaningfully expand or modernize. Force size stagnates, readiness remains limited, and Germany's contribution to NATO deterrence remains primarily symbolic.

Scenario 2: Measured Rearmament (Baseline Case)

Germany meets the 2% GDP threshold and retains approximately 60% of the supplemental rearmament package for actual military use, representing Germany's currently planned trajectory. In this scenario, the Bundeswehr undertakes a substantial expansion, supports modernization programs, and invests in new capabilities such as unmanned systems and cyber defense.

Scenario 3: Strategic Shock Response (Aspirational Case)

In response to escalating global threats and pressure from allies (e.g., former President Trump's proposal for NATO members to spend 5% of GDP on defense), Germany accelerates spending dramatically. Defense investment ramps from 1.8% to 5% of GDP by 2029 and holds that level through 2035. In this high-intensity scenario, Germany rebuilds a scalable, modern force, investing heavily in both manpower and emerging technologies. Germany positions itself as a central pillar of NATO’s European deterrence and power projection capability.

The graph below shows each of the three scenario projections from the model. Scenario 1 drives a small increase in troop numbers over current levels, but does not represent substantive change from the status quo. Scenario 2 ramps personnel levels to over 400,000 by 2035, a meaningful and substantive improvement. Scenario 3 peaks at over 600,000 personnel, representing a transformative reordering of German society and commitment to defense. Under this scenario, the German military would exceed its Cold War troop numbers, with 0.72% of the German population serving in uniform.

The first and third scenarios place plausible lower and upper bounds on future German troop numbers, while Scenario 2 represents a middle ground. Two key observations:

1. $1 Trillion Enables Real Growth – The proposed $1 trillion supplemental package—if fully allocated to defense—has the potential to scale the Bundeswehr dramatically.

2. Defense Inflation Outpaces GDP Growth – In both the 2% and 5% GDP spending scenarios, troop numbers initially rise, but decline over time as per-soldier costs continue to outpace national economic growth. This reflects a structural trend: as military equipment, training, and personnel costs inflate faster than GDP, fixed percentages of national output buy fewer soldiers and systems each year. Without sustained increases in real defense spending, force size will atrophy within a decade, even after an initial buildup.

A key modeling assumption in Scenario 2 is that the German government backloads spending, allocating a more significant portion of the $1 trillion package in the later years of the forecast. This approach is designed to offset rising personnel and equipment costs, enabling a more sustainable and gradual force buildup over time. Because Berlin has not yet released a detailed annual appropriation plan for the complete rearmament initiative, yearly budget allocations are estimated using historical defense inflation rates, capital expenditure patterns, and political signaling. This assumption is critical for projecting troop numbers and capability development in the 2030s.

Manpower Constraints: A Strategic Bottleneck

While Germany may successfully allocate capital for rearmament, a critical question remains: does it have the manpower to fill its ranks? The answer is far from certain. According to the German government forecasts, the largest share of the military-eligible population — men aged 18 to 30 —is projected to shrink relative to the aging population by 2030. This demographic cohort, highlighted in the graphic below, forms the core recruiting base for the Bundeswehr. Even under favorable economic conditions and increased defense budgets, force expansion will be severely limited unless Germany reforms its recruitment model, reintroduces some form of national service, or rapidly integrates automation, unmanned systems, and non-traditional personnel models.

Germany’s Rearmament Problem

Germany’s fourth rearmament effort arrives with political momentum and deep pockets—but faces hard limits imposed by demographics, rising per-soldier costs, and labor market competition. Without a scalable approach to manpower and more adaptive force structures, the Bundeswehr may find itself flush with funding yet unable to generate lasting combat power. Rearmament will require more than money — it will demand a rethinking of what a modern, sustainable force looks like in the heart of Europe.

Subscribe now

Read More of Our Work Here:

Sustained Russian Shahed Swarms: The War of Precision Mass Continues

Damn the Torpedoes: Smaller, Cost Efficient, Less-than-Lethal Systems

One Million Suicide Drones with Chinese Characteristics

Capability Analysis: AI Machine Guns for Drone Short-Range Air Defense

The views and opinions expressed on War Quants are those of the authors and do not necessarily reflect the official policy or position of the United States Government, the Department of Defense, or any other agency or organization.

End Note:

German Military Size Graphic Sources:

1. GDHI

2. International Encyclopedia of WWI

3. Encyclopedia Brittanica

4. Müller, Rolf-Dieter (2016). Hitler's Wehrmacht, 1935–1945. Lexington: University Press of Kentucky. ISBN 978-0-81316-738-1.

5. German Army

6. World Bank

Finding accurate and transparent data was difficult. Early Germany military size estimates from World War II and before may not capture the full size of the entire armed forces, including the German Navy.

In January 2025, President Trump signed an executive order titled Iron Dome For America, which has since been renamed Golden Dome for America. This order begins actions that will presumably culminate in the development of a new integrated air and missile defense system for the American homeland. The logic behind Golden Dome also offers insight into the strategic vision of the Trump Administration. While seemingly defensive in nature, Golden Dome represents the conviction that continuous competition between nuclear powers is not an enduring feature of international relations, but rather a game that can be won. By this logic, Golden Dome will upset the nuclear deterrence framework of mutually assured destruction (MAD), under which the world has enjoyed decades of nuclear peace. This article provides a brief introduction to MAD and its historical context, examines how Golden Dome will disrupt MAD, and briefly discusses what this disruption may mean for the future of strategic competition.

Subscribe now

Understanding MAD

Mutually assured destruction or MAD refers to the conditions of competition between nuclear states where any nuclear attack would lead to the destruction of both states and their respective societies. In these conditions, there is no advantage gained from being first to strike an adversary, and there is no incentive to disarm.

Throughout the Cold War, game theory was often used to understand the framework of nuclear weapon employment decisions, as championed by preeminent 20th century polymath and War quant John Von Neuman. In a simple two-player context, the game is simple: each player has the decision to cooperate or strike. Let us call these players A and B. We can then represent the payoffs of each player given the other player's decision in matrix form, where the players’ utility is shown for each decision combination.

We can then represent MAD in this form, as shown in the figure below. The utility values here are arbitrary representations of the negative consequences of an attack by either party, since any strike will result in retaliation from a second-strike capable force. The dominant choice – or Nash Equilibrium – for each player here is clearly cooperation.

Early in the Cold War, the conditions for MAD did not exist. During his administration, President Eisenhower felt that the United States held a nuclear advantage over the Soviet Union, while the Soviet Union maintained an advantage in conventional mass. For the limited time that the United States possessed a nuclear advantage, the game of nuclear competition was unstable, since the United States possessed a legitimate incentive to strike first to avoid fighting a conventional conflict.

A game theory representation of a situation in which one party possesses a nuclear capability advantage can be seen in the payoff matrix below. Player A has a large incentive to strike first, as he can hurt his enemy without much fear of counterattack by potentially knocking out his adversary’s entire strike capability in the first action. Player B has more to lose by striking first, but still may have incentives to do so. For example, Player B can inflict damage on Player A in a first strike, but without the ability to destroy Player A’s ability to counterattack, Player B will still suffer disproportionately in retaliatory strikes, though less severely than if Player A had struck first. This represents a very unstable situation for both players. Player A has clear incentives to strike first, and Player B also has clear incentives to strike first if he believes Player A’s strike is inevitable. Such an unstable nuclear paradigm was arguably the case for much of the early Cold War years in some form or another.

Strategic Defense or Security Breakdown?

Building the Golden Dome means MAD is over. The executive order specifically mentions layered defense against countervalue attacks – where countervalue attacks refers to attacking targets of value to the adversary’s society, such as population centers, economic hubs, and critical infrastructure rather than targets of direct military significance.

Defense against countervalue attacks is not entirely new for the United States. However, since the end of the Cold War, American nuclear posture has been focused on rogue nuclear states like North Korea. The Golden Dome order envisions going beyond defending against threats from rogue states, and focuses instead on defending against countervalue strikes from America’s most capable nuclear adversaries. While North Korea has been improving its strike technology, defending against North Korean nuclear attacks does not compare in scale or complexity to defending against strikes by Russia and China.

If successfully implemented, Golden Dome would remove the condition of mutual vulnerability and alter the decision-making framework for all belligerents. When one side becomes invulnerable, the game changes entirely. Consider an alteration to our earlier payoff matrices. In this case, Player A is now invulnerable to strikes and counterattacks by Player B.

In this example, reaching invulnerability is clearly in the interest of both players. However, reaching invulnerability does not happen without advance warning, and the signal that a country is pursuing significant defensive upgrades invites other actors to react in the meantime. In the case of Golden Dome, America’s adversaries may perceive that the United States is trying to gain an advantage in a nuclear conflict, and this fact alters the decision-making calculus for all players and introduces new instability in nuclear competition. For example, if Russia or China perceive that the altered nuclear paradigm created by Golden Dome would be to their long-term disadvantage, they may be incentivized to strike preemptively before the Golden Dome is in place. Additionally, should Golden Dome be successfully implemented, actors could perceive a nuclear strike as less escalatory if there is reason to believe the attack would be intercepted, making this form of nuclear brinksmanship a part of crisis escalation (consider the missile attacks of Iran on Israel as a non-nuclear parallel).

If another nuclear power develops its own countervalue defense system, nuclear dynamics could become even more dangerous. Other nations could see a massive build-up in nuclear weapons as the only way to offset increasingly capable nuclear defense systems. Likewise, in a world in which every nuclear power must develop their own countervalue defense systems, the decision space may become hugely complex as every player compares their relative offensive and defensive advantages to every other player. While some potential outcomes are speculative, changes in the dynamics of nuclear competition will greatly increase the chances of a new nuclear arms race among great-powers – and this may have already begun with China’s nuclear modernization.

Extended Deterrence and Non-Proliferation

MAD didn’t just deter adversaries – it also reassured allies under the concept of extended deterrence. Under MAD, Boston and San Francisco are just as vulnerable to nuclear attack as Berlin and Tokyo. Under such conditions, allies of nuclear powers could be reasonably assured that their nuclear ally would not drag them into a nuclear conflict in which the non-nuclear state absorbs the strikes. This is of course because the nuclear state is not able to control or defend against any attacks either on its own homeland or the homeland of its allies.

Under these conditions, non-proliferation arguments make sense, since non-nuclear states can fall into the MAD architecture without arming themselves. However, with Golden Dome protecting the American homeland, the United States could potentially climb the nuclear escalation ladder with few consequences for its own safety, leaving adversaries like Russia or China with no credible option but to strike America’s more vulnerable allies instead. In this world, non-nuclear states could feel obligated to begin developing their own nuclear weapon programs as a deterrent.

Imagining Victory in Competition

Despite the disruption and risk possible due to the development of Golden Dome, there are potential upsides and opportunities. MAD has always had its critics. Some argued that MAD – and adjacent policies such as détente – implicitly accepted bipolar great power competition and its associated spheres of influence as enduring features of international relations as the permanent status quo.1

It took some imagination and disruption to challenge Cold War logic and envision what victory under a different paradigm might look like. President Ronald Reagan could be credited with having such a vision. Reagan rejected détente, amped up pressure on the Soviet Union, and departed from diplomatic norms that created Cold War stability. Examples of this strategy included the Strategic Defense Initiative, Reagan’s plan to build a capability similar to Golden Dome. Had it been successful, this missile defense capability would have disrupted the nuclear deterrence framework as described above, but the technology underpinning the initiative did not materialize.

While the stability of the existing order secured by MAD may be under threat, President Trump and his strategy advisors would not take this step unless they believed there was an inherent advantage for America and its interests. Indeed, select criticism of President Biden’s approach to strategic competition with China observed that the administration simply sought to compete rather than define or achieve victory. An ambitious plan like Golden Dome could be seen as an attempt to move beyond status quo competition and towards victory.

One way to understand this logic is to envision Golden Dome as a tool to prevent vertical escalation in a future regional war. Consider a scenario in which China invades Taiwan. China could threaten nuclear strikes against the US homeland to deter intervention. Under MAD, this creates a credible threat of mutual destruction and may successfully deter US involvement, especially if the sovereignty of Taiwan is perceived as being secondary to the consequences of nuclear war.

Similar logic was apparent in the Biden Administration’s cautious response to Russian nuclear threats in the Ukraine conflict. President Biden showed real concern over the risk of nuclear escalation, and this often translated into a slow, carefully calibrated, and risk-averse approach to Ukrainian aid.

In contrast, a functioning Golden Dome defense may embolden the United States to be more assertive in situations involving risk of nuclear escalation. Reduced homeland vulnerability could give US leaders more freedom of action abroad, particularly when pursuing national objectives in confrontations with nuclear-armed powers. Of course, this shift would create new risks for US allies, who may remain vulnerable to countervalue attacks, but the overall logic of assertiveness remains and aligns with an America-first approach.

Towards a New Nuclear Future

Golden Dome may mark the end of MAD, but its implications for global security are far from certain. The creation of Golden Dome demonstrates the United States’ desire to push out of a stable equilibrium to a position of advantage. Other states will certainly respond. Indeed, China is already embarking on overhauls to its nuclear programs. Perhaps a new equilibrium will emerge, but for now, the risk of miscalculation has increased as global nuclear powers transition from mutual vulnerability to asymmetric capability.

Subscribe now

The views and opinions expressed on War Quants are those of the authors and do not necessarily reflect the official policy or position of the United States Government, the Department of Defense, or any other agency or organization.

Bottom Line Up Front:

1. Across 2024, US, UK, and Israeli strikes against Houthi forces in Yemen contributed to a reduction in the rate of Houthi attacks on maritime shipping in the Red Sea.

2. The reluctance of shipping firms to return to operations in the Red Sea en masse suggests that simply halting Houthi attacks on merchant shipping may be insufficient to restore the perception of stability.

3. With the confidence of shipping firms remaining low even as Houthi attacks have stalled, Houthi actions remain a playbook for malign actors in the vicinity of maritime chokepoints worldwide.

The Red Sea Crisis in Brief

Since late 2023, Houthi rebels in Yemen have prosecuted a regional campaign of sea denial in the Red Sea, Bab al-Mandab Strait, and the Gulf of Aden. Although nominally conducted in response to US support of Israeli military operations in Gaza, this campaign has formed a part of the decades-long Iranian-led axis of resistance to Israel and the West in the region. On October 19, 2023, the Houthis opened their campaign by launching several salvos of one-way attack drones and land-attack cruise missiles over the Red Sea, many of which were intercepted by the USS Carney (DDG-64) in what the US Navy described as “the most intense combat engagement by a U.S. Navy warship since WWII.”

In November 2023, the rebels seized the Galaxy Leader, a British-owned cargo ship, taking its crew hostage and commencing the series of attacks on merchant shipping that would continue through the next year. In response, the United States announced the launch of Operation Prosperity Guardian, a multi-partner initiative to reestablish the security of maritime trade in the Red Sea and Gulf of Aden. In January 2024, the United States and United Kingdom carried out a series of air and naval strikes against Houthi targets, with further strikes by the United States and partners conducted throughout the remainder of the year.

However, these strikes did not halt Houthi aggression. Notable Houthi attacks on merchant shipping throughout 2024 included the respective sinkings of the Rubymar bulk carrier and the MV Tutor coal carrier in March and June, alongside the near-destruction of the oil tanker MV Sounion, fortuitously towed to safety by a EUNAVFOR task force before its contents could cause an environmental catastrophe.

In January 2025, coinciding with an announced ceasefire between Israel and Hamas, the Houthis announced that they would pause attacks in the Red Sea, and released the crew of the Galaxy Leader. However, the group reversed their position in early March, prompting sudden US military strikes on March 15, apparently conducted to degrade the Houthi fires capabilities before attacks on shipping could resume. Where the conflict goes from here remains uncertain.

Global Economic Consequences of the Red Sea Crisis

Houthi attacks on merchant shipping in the Red Sea have emphasized the dependencies of efficient global trade on the security of key maritime choke points. The crisis has caused dozens of major shipping firms to alter their operations, with many firms choosing to navigate around the Cape of Good Hope rather than accept the physical risk and surging insurance premiums currently charged for Red Sea voyages. As of 2025, many shipping firms are still taking a wait-and-see approach before returning to operations in the region due to continued uncertainty around regional security, with the firms’ operations now aligned to the status quo for at least the next fiscal year. Both container and tanker ship passages through the Red Sea remain at levels far below 2023 baselines, as shown in the graphic below.

With many firms still committed to avoiding the Red Sea in the near term, thereby accepting higher operating expenses associated with longer, less efficient routes, shipping customers have continued to bear the additional cost. For example, 40-foot-equivalent (FEU) container shipping rates surged across 2024 as constant demand for goods collided with the inefficiencies caused by the Red Sea conflict, as shown in the graphic below. As of this writing, FEU rates have declined as firms have adjusted to their new operational constraints, but rates still remain elevated above 2023 levels.

Analysis of US, UK, and Israeli Strikes

Using curated data from the Yemen Conflict Observatory, a special project of the Armed Conflict Location & Event Data (ACLED) initiative, we can compare the volume of Houthi attacks against maritime shipping against the volume of the US-led international response. This analysis also pairs the uniquely identified merchant vessels in the ACLED dataset with ship metadata from the Electronic Quality Shipping Information System (Equasis) to better understand the characteristics of the ships ultimately targeted by the Houthis.1

The graphic below shows the merchant ships attacked by the Houthis over the course of 2023 and 2024 by dead weight tonnage (DWT), a measure of the total carrying capacity of merchant ships. No clear patterns emerge in target selection by either ship type or size, although the rate of attacks appears to decline beyond August 2024, with the plot becoming increasingly sparse by the end of the year.

The declining rate of Houthi attacks on merchant shipping becomes more clear when we compare the number of cumulative Houthi attacks against cumulative US, UK, and Israeli strikes. As shown in the graphic below, although the Houthis conducted frequent attacks on shipping in 2024, with a noticeable acceleration over the summer months, the accumulation rate of attacks ultimately flattened towards the end of the year.

Meanwhile, US-led strikes built up rapidly early in 2024, and after a months-long deceleration in the spring, continued at a relatively stable rate for the remainder of the year as the Houthi attack rate began to level out. This analysis suggests that Houthi capabilities and capacity may have been slowly degraded by sustained US-led strikes over the course of 2024, with the Houthis ultimately ending the year in a weaker position relative to the start.

However, it is challenging to disentangle the specific contribution of these strikes from the contributions of other geopolitical developments. Examples of other potential contributors include the decimation of Hezbollah leadership and the sudden collapse of the Assad regime, both of which may have convinced Iran that a regenerative break in hostility in the region would be to its benefit. Likewise, reports emerged in late 2024 that exposed Russia’s direct operational support to the Houthis, including supplying the militia with targeting data on merchant ships, and it is possible that Russia suspended or reduced this support to avoid provoking the incoming US administration. The true relationship between the US-led strikes and the deceleration in Houthi attacks therefore remains unknown.

Last year, plenty of debate surfaced as to whether the US strategy for returning stability to the Red Sea via limited strikes could ever prove successful. If the United States primarily aimed to halt attacks on merchant shipping, these strikes outwardly appear to have been a success – for now. However, as shipping firms have indicated this week, it may take a lot more than a cessation of attacks on merchant ships to reassure shipping companies and the firms that insure them that their ships and crews are safe. The possibility of violence against shipping, however remote, is a powerful deterrent to commercial firms, and US hard power methods seem ill-suited to meet this asymmetry. For each day that massive international companies believe a major global trade corridor is too unsafe to navigate, the Houthis and Iran chip away a small fragment from the deterrent credibility of the United States. Now with US strikes on the Houthis resuming, and with another aircraft carrier headed to the region as of this writing, it seems any return to perceived stability in the Red Sea remains in the distant future.

Subscribe now

1. Ukraine is on track to almost 27x its monthly drone production from January 2024 to December 2025, with recent 2025 plans to purchase 4.5 million FPV attack drones.

2. These new orders increase FPV density from a 2024 daily average of 7 FPVs per mile of frontage to over 20 FPVs per mile in 2025.

3. Concurrent advancements in FPV terminal guidance using edge artificial intelligence models combined with increased production numbers could drive a 900% improvement in Ukrainian lethality per mile of front.

Production Boom

Ukraine’s battlefields witnessed an unprecedented density of first-person-view (FPV) drones in 2024. The conflict has a well-defined World War I-style frontline extending 600 miles, but these one-way attack drones are changing the character of war as part of the emerging precision mass regime. Ukraine reportedly acquired 1.5 million FPV drones in 2024, and now says it will triple domestic FPV purchasing in 2025 to 4.5 million drones.

Subscribe now

The Ukrainian drone industry could not produce at this scale at the beginning of 2024. In January of that year, almost two years into the war, Ukrainian industry ramped up to produce 20,000 FPV drones. This number would break records in any Western military, and represents about 1.1 FPV drones available daily for each of the 600 miles of battlefield frontage in Ukraine. However, this early 2024 production rate proved insufficient to meet Ukraine’s battlefield requirements as Russia continued to quickly replace destroyed equipment and soldiers.

Across 2024, Ukraine’s drone industry made immense capital expenditures (CAPEX) to expand its production capabilities and received additional funding from the United States. By the end of 2024, the industry had achieved a monthly production rate of 200,000 FPVs. This 10x increase in production capacity improved FPV density to approximately 11 FPV drones per day for each mile of frontage.

The graphics below show estimates of 2024 Ukrainian FPV production alongside 2025 production forecasts derived from open-source reporting and War Quants factory-to-frontline modeling. The 2024 estimates use the January 2024 and December 2024 production numbers quoted by the Ukrainian government, while the 2025 forecast reports monthly production capacity using the top-line 4.5 million cumulative production numbers. Monthly production is roughly linear, and modeled to meet known monthly production rates and overall production targets.

Overall, the modeling effort reveals that a production target of 4.5 million FPV drones is a lofty goal. Ukraine must increase monthly FPV production numbers through massive capacity expansion to meet its production targets. Base production in 2025 represents a continuance of the 200,000 FPV monthly December 2024 production rate. Increment 1 includes 50% of the additional expansion, with Increment 2 representing the final 50% incremental production required for the 4.5 million FPV target.

If it meets its stated goals, Ukraine’s drone industry will produce over 500,000 FPVs per month by December 2025, equating to 29.4 daily FPVs per mile of frontage using an average 30-day month and 600-mile frontage. This terminal monthly run-rate is almost 27x larger than January 2024. If achieved, this mass may provide an immense advantage in attriting Russian assaults and preserving scarce personnel for critical missions.

FPV Drone Deployment and Effectiveness

Concurrent with increases in monthly drone production rates, the end of 2024 saw improvements to Ukrainian FPV quality and effectiveness. Russia and Ukraine began integrating fiber-optic cables to control FPV drones and avoid jamming. Fiber-optic cable FPVs, however, have range and employment limitations due to their tether with the operator. To overcome these limitations, Ukrainian companies are improving terminal guidance systems that allow drones to maneuver to and attack their targets autonomously. Now, with AI-enabled terminal guidance, drone operators can designate targets from outside EW jamming bubbles. Drones can then fly into the bubble and engage the target autonomously so there is no link between the drone and operator to jam. Furthermore, we expect AI-enabled terminal guidance systems to improve as more battlefield training data becomes available.

We can model this improvement using the FPV kill chain from previous War Quants work. In the model, we estimate the success probability of a set of actions: the launch, movement, kinetic kill avoidance, non-kinetic kill avoidance, and strike of the FPV. In this case, terminal autonomy substantially increases the success rate of non-kinetic kill avoidance. In the example below, mission success rises from 15% to 60% similar to a recent CSIS report.

In aggregate, according to recent RUSI reporting, Ukraine’s FPVs account for 60-70% of all destroyed Russian equipment. While not an all-weather fire asset like artillery, confirmed numbers over 50% suggest that FPVs are quickly becoming the new “King of Battle."

Factory-to-Frontline Pipeline

Combining trends in efficacy and mass, we can measure the improvement in Ukraine's drone war lethality from 2024 to 2025. Assuming an average FPV attack success probability of 20% and seven FPVs per mile of frontage daily, as seen in 2024, we expect 1.4 successful daily attacks per mile. Scaling to 21 daily FPVs per mile of frontage based on 2025 production forecasts, we expect successful attacks to increase to 4.2 daily per mile. After adjusting the probability of a successful attack to 60% to account for the potential benefits of AI-based FPV terminal guidance, the successful attack numbers jump to 12.6 daily per mile. This capability and capacity improvement is illustrated below.

Improving from 1.4 daily successful attacks per mile of front to 12.6 is a 900% year-on-year growth. Is this decisive for the war effort? Time will tell.

Unprecedented Build Up

By the end of 2025, Ukraine has the potential to become a military-industrial powerhouse that can punch well above its relative size by deploying unprecedented masses of semi-autonomous robots along the front. If successful, this deployment could create an unmanned barrier that Russian attacks must breach to capture Ukrainian territory, and there is no doubt that nations and non-state actors across the globe have taken note. The future will likely see the export of AI-enabled drones to other battlefields, kickstarting an era of autonomy, artificial intelligence, and robotics in war.

“The Future is here, it’s just not evenly distributed yet.” - William Gibson

Subscribe now

Read More of Our Work Here:

Sustained Russian Shahed Swarms: The War of Precision Mass Continues

Damn the Torpedoes: Smaller, Cost Efficient, Less-than-Lethal Systems

One Million Suicide Drones with Chinese Characteristics

Capability Analysis: AI Machine Guns for Drone Short-Range Air Defense

The views and opinions expressed on War Quants are those of the authors and do not necessarily reflect the official policy or position of the United States Government, the Department of Defense, or any other agency or organization.

1. Propulsion disablers can offer nations a less-than-lethal mine solution and flexible, cost-effective response options supporting sea denial and control.

2. Advanced mining systems offer increased minefield performance, area coverage, and manageable mine delivery requirements at a lower cost.

   ---

Naval mining technology is advancing. Today, navies have a broad spectrum of choices in mining technology, especially compared to previous eras when minelaying warships dropped spherical contact mines off the stern. Contemporary visions for the future of mine warfare promise persistent, reliable, and surgically accurate systems like the propulsion disabler (PD). The PD is a hypothetical system in a small form factor with a less-than-lethal torpedo-like warhead payload. PDs could disable a vessel’s propulsion system without catastrophically damaging the hull, thereby creating a variety of unique strategic opportunities for navies in possession of this capability. In this article, I compare PDs to other contemporary mining technologies using a simulated Baltic Sea scenario first introduced in a previous article (Part 1: The Return of Naval Mining) to better understand how novel mining capabilities can be employed in the Gulf of Finland to control sea lanes and deny adversary maneuver.

Subscribe now

Setting the Stage: Naval Mining in the Baltic Sea

In a 2024 article from Proceedings, Commander Laanemets of the Estonian Navy asserts that the Baltic’s “new setting” requires the Northern Atlantic Treaty Organization (NATO) to possess an organic mining capability to enable anti-access/area denial (A2/AD). History and geography support the benefits of mining in this region. The Baltic Sea’s relatively shallow depth and abundance of maritime chokepoints create an ideal environment for mining operations. Indeed, the Crimean War and both World Wars demonstrated the effectiveness of mining operations in the region. Today, Russian capabilities at Kaliningrad include more than 50 vessels, alongside stores of land-based anti-ship missiles and numerous combat aircraft. Yet all of these capabilities possess limited utility if they cannot break out of the Baltic Sea at the outset of conflict. This fact makes the value of mining operations in the Baltic Sea self-evident. Moreover, illegal shipping practices and submarine cable sabotage linked to Russia present emerging reasons for Western militaries to establish sea denial and control in the Gulf of Finland. While the need for a deterrent mining capability is clear, what specific mine systems NATO should invest in is less obvious.

Options With Modern Mining Technology

The scenarios below assess the Quickstrike (QS), Hammerhead (HH), and Propulsion Disabler (PD) mining systems in a Baltic Sea use case. As the only widely fielded system, the legacy QS bottom mine serves as a reference point to the others. The HH, as described by Strikepod, is a moored mine with an encapsulated MK54 lightweight torpedo, sophisticated sensors, a communications suite, and high-capacity energy modules, primarily designed to target submarines. While conceptually similar, PDs do not yet exist. For the purposes of the scenario below, the system is loosely envisioned as an encapsulated, less-than-lethal version of the US Navy’s Compact Rapid Attack Weapon (CRAW). The CRAW system is considered a “Common Very Light Weight Torpedo (CVLWT),” capable of serving in both anti-submarine missions and as an anti-torpedo countermeasure. In this hypothetical scenario, the same housing is reprogrammed and fitted with a smaller warhead to disable ship propulsion systems. The assumed system characteristics for each mining capabilitity are summarized in the table below.

For modeling purposes, the advanced HH and PD mine systems are assumed to possess a higher probability of detection and actuation against enemy ships over the legacy QS system. The PD is also depicted with a slightly higher likelihood of actuation due to its less-than-lethal nature. The underlying assumption here is that nations might be more willing to target vessels with lower track qualities using less-than-lethal munitions. Finally, the PD cost, hypothesized at approximately $200k per unit, assumes that each system will cost less than the highly capable but expensive HH, and more than the legacy QS mine.

Gulf of Finland Simulations

To model these capabilities in the Baltic Sea, the baseline simulation from Part 1: The Return of Naval Mining returns with new enhancements. The simulation’s visualizations have been improved to depict simulated minefields on a world map and display General Bathymetric Chart of the Ocean (GEBCO) data. The two maps below show the model’s geographic visualizations for the Gulf of Finland minefield.

The top plot gives planners an appreciation for the minefield’s surrounding coastlines and territorial boundaries, while the bottom simulation plot depicts the Baltic Sea’s bathymetry, simulated routes, and mines. Each minefield configuration is simulated 1000 times, and each simulation generates 100 sequential vessels to transit the Gulf of Finland. In the simulation, once a vessel is successfully engaged, no other mines are expended. The baseline model assumptions are also tailored for the systems and environment simulated.

List of Model Assumptions

1. Minefield is preemptively established.

2. Advanced mine systems deconflict with each other through undersea communication networks and target vessels individually.

3. Mine systems are effective at simulated depths.

4. Mine systems detect target vessels according to system detection probability.

5. Mine systems will attempt to actuate when a target vessel passes within system employment ranges.

6. Vessels become casualties when successfully engaged by a single mine system.

Performance and Cost of Simulated Minefields

Minefield performance is assessed according to expected casualties, initial threat on sea route, and casualty density distribution measures of effectiveness (MOEs). These MOEs are outlined in Part 1: The Return of Naval Mining. The descriptions of the simulated blocking scenario and associated performance are summarized below. As discussed in the previous article, a blocking minefield should “present a high initial threat on sea route, have a sizable left skew in its casualty density distribution, and have high expected casualties.”

Scenario Descriptions:

• QS Scenario: a densely seeded minefield using legacy QS mines. This scenario requires the most systems at the highest cost.

• HH Scenario: a minefield employing advanced HH systems at emplaced at a distance of 60% the system’s maximum effective range. This scenario has the least number of mines.

• PD Scenario: an advanced minefield employing less-than-lethal PDs, emplaced at a distance of 60% at 66% of the system’s maximum range.

• PD Reinforced Scenario: reinforces the PD scenario with nearly twice the mine systems and covers more area.

The simulation results support the thesis that advanced mining systems offer better-targeting performance, wider minefield coverage, and manageable delivery considerations at a lower cost. The QS configuration, serving as a baseline for cost and effectiveness, achieves fewer expected casualties and covers a significantly smaller area at the highest price point. Setting the minefield cost aside, delivering 1542 mines places a significant, perhaps infeasible, burden on delivery aircraft, subsurface, and surface vessels.

The next three advanced minefield configurations offer comparable performance with similar costs and area coverage. All minefields create 14 expected casualties and are substantially cheaper than the QS minefield. Notably, the PD and PD Reinforced configurations produce the same number of disabled vessel casualties. The simulated vessels incrementally degrade the minefield integrity until a shipping lane is cleared, allowing successive vessels to transit safely. In reality, the vessels’ willingness to transit the minefield would likely decrease with each subsequent casualty. Reseeding operations would be used to maintain the minefield, replenish expended mines and mitigate vessel penetrations. Additionally, the feasible delivery requirements for the HH and PD minefields are more appealing to planners and cover 100x the minefield area.

The minefield casualty density distribution is assessed as a measure of minefield capacity and effectiveness. The three advanced mine configurations, shown in red, green, and blue, reliably produce higher casualty numbers across simulations. These minefields experience 14+ casualties in the majority of simulation runs. Conversely, the QS mines, shown in purple, perform slightly worse and produce fewer casualties more often.

The Case for Smaller, Cheaper Mines, and Larger Mine Magazines

Smaller, cheaper, and more capable mines are needed to provide minefield effectiveness and response flexibility. PDs seemingly meet all the requirements for flexible mining capabilities and could minimize civilian minefield casualties. If developed and proliferated, PDs would conceivably be outfitted with AI targeting algorithms for enhanced target discrimination, high-capacity batteries for persistence and propulsion, extended employment range, and remote control capability. The possibilty of leveraging these sophisticated capabilities at a scalable cost in large minefield use cases is the main value proposition of the PD.

By comparison, the HH mine possesses a similar suite of high-tech capabilities, but while the existing HH mine’s form factor and capabilities provide evidence that the PD is technologically feasible, the HH does not support the same operational goals. For example, the HH is tailored for anti-submarine warfare rather than surface targets. Presumably, the system could be reprogrammed for anti-surface applications, but the ~$1.1M price tag makes it relatively unscalable and cost-inefficient for distributed use. Furthermore, the MK54 lightweight torpedo employed by each HH eliminates the flexible response options that a PD provides. The victim of a PD attack can be sunk by supporting land or air-based fires, seized, or assisted and repaired. As Captain Dismukes, USN (ret.) explains, “disabling an enemy’s naval ship rather than sinking it will almost always be the superior choice, certainly for the U.S. Navy” and, presumably, NATO members.

Successfully employed and supported Baltic PD minefields could provide flexible options for Western militaries to disrupt illegal shipping from St. Petersburg and Kaliningrad and constrain the Russian Baltic Fleet’s maneuver options in the event of a conflict, potentially at lower cost and risk than other like capabilities. Targeted vessels with propulsion systems rendered inoperable could present opportunities for follow-on seizure and exploitation or subsequent targeting by lethal means.

Subscribe now

The views and opinions expressed on War Quants are those of the authors and do not necessarily reflect the official policy or position of the United States Government, the Department of Defense, or any other agency or organization.

1. Chinese sabotage of the Panama Canal in a Taiwan contingency scenario could delay force closure by more than 12 days, according to a War Quants model.

2. Should the United States lose access to the Panama and Suez Canal, the total distance the fleet must sail to a Taiwan contingency could increase by almost 26%, or approximately 800,000 kilometers.

3. The increased rhetoric and engagement over the Panama Canal reflects a recognition of its importance to US interests in the Western Hemisphere.

The Panama Canal - the Western Hemisphere’s vital maritime connection between the Atlantic and Pacific - currently commands levels of geopolitical attention not observed since President Carter negotiated the canal’s turnover to the Republic of Panama in the late 1970s. The United States retains strategic and economic interests in the canal, but the return of Great Power Competition has reinforced its importance to the US Navy and the joint force. No scenario better illustrates this importance than a possible Chinese invasion of Taiwan.

The White House has recently expressed concerns about a Chinese company operating ports on both sides of the Panama Canal. These hubs provide China with regional economic influence and could potentially serve as launch points for cyber and special operations against the Panama Canal in the event of a Taiwan invasion scenario. Should China successfully sabotage the canal or its operations, it could delay the movement of US forces into the Pacific and enable a fait acompli strategy to quickly seize Taiwan before the United States can halt an invasion.

Where in the world is the US Navy?

To understand the Panama Canal's military significance, one must first understand the force laydown of the United States Navy. According to the homeport registry for Active and NRFA ships, the US Navy moors 248 hulls (including both surface vessels and submarines) across 25 homeports, as summarized in the table below.

Most ships are clustered in ports in New England, the Mid-Atlantic, the Southeast, Southern California, and the Pacific Northwest. In the event of a Taiwan contingency scenario, the most efficient route to the Western Pacific would encompass a journey through the Panama Canal for many of the ships stationed on the East Coast of the United States, with some exceptions. Notably, the five carrier strike groups stationed in Norfolk, Virginia would need to find an alternate route. Although the Panama Canal completed a massive capacity expansion in the last decade, its width still does not accommodate the flight decks of Ford and Nimitz-class carriers. Likewise, due to the sensitivity of their missions, SSBNs are unlikely to leverage the Panama Canal in crisis. After accounting for these constraints, I estimate that if responding to a contingency in Asia, approximately 69 US warships would need to pass through the Panama Canal to take the fastest route to the conflict theater. But what would happen if the fleet could not use it?

How might the loss of the Panama Canal affect the US Navy?

To understand the impact of a canal closure, I’ll analyze a scenario in which the US Navy moves the 69 ships that might reasonably traverse the canal in crisis from their homeports to the Philippine Sea. Using minimum-cost flow and shortest path algorithms, I’ll compare scenarios in which the Panama Canal and Suez Canal are unavailable to US forces to quantify the impact of a canal closure in terms of both total distance traveled and voyage days.

The model makes several assumptions:

1. All ships start at their homeports. In reality, the US Navy always has ships underway and forward deployed. However, because their locations change often and are not usually published, ships are modeled starting at their home ports.

2. All registered ships will be available, subject to the constraints on canal usage outlined above. This simplifying assumption ignores the fact that ships have a readiness cycle which includes dry dock (and the inability to get underway for its duration).

In the best-case scenario in the model, the Panama Canal and Suez Canal remain open to the US Navy. Under these conditions, the algorithm selects the shortest path from US homeport clusters to the Philippine Sea, with ships on the East and Gulf coasts traversing the Panama Canal into the Pacific. Where applicable, ships avoid non-permissive areas like the South China Sea and instead use an alternative route through the Timor Sea. In this best-case scenario, the fleet must sail approximately 3.07 million kilometers to get to the fight in the Philippine Sea.

In a scenario where the Suez Canal is denied, carrier strikes groups from Norfolk must sail around Africa to reach the Pacific Theater. This routing increases the total sailing distance to 3.16 million kilometers.

If the Panama Canal is denied but the Suez Canal remains open, the total distance sailed increases to 3.6 million kilometers. All US warships on the East and Gulf coasts route through the Mediterranean.

In a worst-case scenario with both the Suez and Panama canals closed, significant numbers of ships would need to reroute around the Cape of Good Hope to get to the fight. This transit increases the total sailing distance to 3.87 million kilometers. Interestingly, the Strait of Magellan is never a good option, though this result may not surprise any true mariner.

The table below captures the impact of the four route permutations described above on force closure distances in terms of total fleet kilometers traveled.

But might this impact force flow timelines?

The longer it takes for the United States to respond to Chinese aggression, the more time the People’s Liberation Army will have to gain a lodgement on Taiwan. The US Navy remains the critical combat force that can contest the invasion. Conducting a time-phased force deployment analysis further reveals how canal closures could impact US Navy ship arrival times in theater.

Assuming an average ship speed of 20 knots to account for port calls, resupply, and other delays, analysts can calculate a ship's closure time from its homeport to the Philippine Sea along the optimal paths from above. For simplicity, I assume that all ships depart simultaneously, and no vessel breaks down on the way. Across the four scenarios outlined above, the fleet arrival times diverge on day 21 of the conflict.

If the US cannot use the Panama Canal, up to 30% of its fleet could be delayed. Even if both the Panama and Suez canals are usable, US forces in the Chinese weapon engagement zone may need to persist for 30-60 days without significant support as the fleet musters its assets for decisive action. Conversely, the loss of the Suez seems less critical to US naval mobility unless it is the only canal available.

What is Old is New Again: Grading US Posture for Conflict

In 1964, a great debate raged about the value of the Panama Canal in the nuclear age. In a contemporary column, the New York Times examined arguments that the canal’s military significance had waned in a world in which the United States possessed a two-ocean navy with enough ships to contest adversaries simultaneously in both theaters, even as the Soviets clearly sought to influence and control maritime chokepoints globally through subversion. We have entered a new era of Great Power competition and maritime chokepoints are definitively back in focus. Force closure logic should inform US policy and drive a closer relationship with Panama to ensure the security of the canal. The United States cannot take sea lanes for granted and no longer has a two-theater navy. Just as it has been for a century, the Panama Canal remains a lifeline that enables the United States to project power and protect its interests with maximum efficiency.

Read More of Our Work Here:

Sustained Russian Shahed Swarms: The War of Precision Mass Continues

One Million Suicide Drones with Chinese Characteristics

FPV Math: Precision Mass for 21st-Century Warfare

Capability Analysis: AI Machine Guns for Drone Short-Range Air Defense

The views and opinions expressed on War Quants are those of the authors and do not necessarily reflect the official policy or position of the United States Government, the Department of Defense, or any other agency or organization.

In our previous analysis, we noted that from 2022 until the late summer of 2024, the Ukrainian interception rate against Shahed strikes had been high, consistent, and stable. When launched at low rates, the Shahed's slow speed and large noise signature appear to make the drone a relatively easy target for various Ukrainian air defense systems. However, in August 2024, as Russian forces began rapidly increasing the rate and volume of Shahed launches, Ukrainian interception rates fell precipitously. At the time, we assessed that the Russian employment of these OWA drones in mass represented a key innovation in making loitering munition attacks a more effective form of long-range fires and that increased domestic production was a key factor in achieving sustained mass.

With updated data now available, we wanted to see if the Russians had sustained high rates of Shahed launches over the early winter months, and if the Ukrainians had been able to build additional air defense capacity against this new threat. Unfortunately for the Ukrainians, the new data reveal that the Russians have continued to increase their Shahed launch rate over late 2024 levels. Moreover, the Ukrainians appear to have been unable to achieve the air defense gains necessary to return to the high shootdown rates they enjoyed earlier in the war.

The figure below demonstrates the continued increases in Russian monthly Shahed launches alongside the Ukrainian “new normal” interception rates. For example, in January 2024, the number of Russian Shahed launches hit a conflict-high of 2,641. Of note, the number of Ukrainian interceptions has also significantly increased since the beginning of the war, with the Ukrainians now conducting thousands of Shahed interceptions per month. January 2024 also represented a conflict-high in Shahed interceptions, with the Ukrainians shooting down 1,591 drones. Unfortunately, these gains in interception successes have simply not kept proportional pace with the new numbers of Russian OWA launches.

A clearer upward trend in Russian Shahed salvo sizes also emerges in the new data. The minimum Shahed salvo size has risen from 3 in September 2024 to 30 so far in 2025. Meanwhile, of the 48 Shahed attacks thus far in 2025, only 4 have had swarm sizes of less than 50 — the average attack size being about 88 and the median 81. This evidence suggests that Russia no longer considers smaller attacks to be a valid tactic.

A direct examination of Ukrainian shootdown rates across the conflict indicates that Ukrainian air defenses remain stuck in an interception-rate trough. We do note a positive change in the interception rate from December to January, and it is possible that the Ukrainians are turning the corner on their air defense capacity and effectiveness. However, more data will be required to confirm the presence or direction of any trend.

The rapid increases in Russian Shahed launches over the last few months raise questions as to whether this pattern is sustainable. According to recent CNN investigative reporting, a new drone factory in the Alabuga Special Economic Zone in Russia produced 5,760 Geran-2 drones between January and September 2024 (the Geran-2 is the domestic name for the Shahed). This factory — now believed to be the primary source for Shahed-style drones for the Russian military — may have also produced up to 10,000 new Gerbera drones in 2024. The Gerbera is a low-cost decoy version of the Geran-2, and has been increasingly launched alongside its armed counterpart to further stress Ukrainian air defenses, which likely cannot easily distinguish the difference. In the graphics presented above, it is assumed that the launch totals include both Geran-2 drones and Gerbera decoys in unknown proportions to one another, just as they would appear to Ukranian air defenses.

Using the total estimated production figures quoted in the CNN report, we can back into some rough monthly production totals for Geran-2 drones and Gerbera decoys. Assuming the rate of production for all drones and decoys is constant, we can estimate that the Alabuga plant produced a total of 7,680 Geran-2 drones in 2024 by scaling the Jan-Sep 2024 total of 5,760 drones by three additional months, giving us a per-month production rate of 640 OWA drones. Similarly, an annual production total of 10,000 Gerbera decoys suggests a production rate of 833 decoys per month.

Finally, a Ukrainian source featured in the CNN article suggested that in a Russian salvo of 150 Shahed-style drones, only 20 or 30 total OWA munitions would actually be launched, with the rest being decoys. For our purposes in estimating capacity, we’ll assume that in a given month, only 20% of the total launches represent Geran-2 drones. For simplicity, we’ll also make the conservative assumption that the Russians started 2024 with no existing drone inventory, and we’ll assume that any unexpended inventory can be stored from month to month. These calculations and assumptions give us the following 2024 production and expenditure estimates for Russian Shahed-style drones.

Several simplifying assumptions have been made here, but overall, these production and expenditure estimates look favorable from a Russian perspective. Even though the Russian military may be shooting more Gerbera decoys than it produces in certain months, there may be existing inventory to support the monthly launch rates observed since late summer 2024. Moreover, if we believe that Geran-2 drones only make up around 20% of every salvo, the Russians may now possess considerable reserves of these munitions — allowing them to reserve OWA capacity for their most critical targets.

Taken together, these updated data add additional evidence to our previous assessment that employing OWA munitions as precision mass represents an increasingly valid tactic to overwhelm and defeat air defenses. Furthermore, this tactic likely remains highly cost-effective for the Russians. At an estimated cost of $20,000 to $50,000 per munition, and at the production rates estimated in this analysis, the Russians can likely sustain the recent trend of large attacks and continue to impose heavy costs on Ukraine in the form of strikes and forced air defense expenditures.

Subscribe now

The views and opinions expressed on War Quants are those of the authors and do not necessarily reflect the official policy or position of the United States Government, the Department of Defense, or any other agency or organization.

This article begins with the First Offset, the foundation of Eisenhower’s ‘New Look’ strategy. This strategy relied heavily on policies of massive nuclear retaliation and covert operations, but this analysis will focus primarily on the nuclear strategy component. The First Offset illustrates how the interplay of grand strategy, technology, and force design shaped America’s Cold War posture and offers lessons for today’s challenges.

Historian John Lewis Gaddis defines strategy as aligning potentially unlimited aspirations with necessarily limited means. When the scope of a given strategy becomes sufficiently large, it becomes a grand strategy.1 An offset strategy evaluates the means available to both one’s self and the adversary and recognizes a strength or capability that can negate an advantage held by the adversary. The Eisenhower administration's New Look defense strategy contained all the elements of a grand strategy and the characteristics of an offset strategy.

Subscribe now

Ends and Means: Elements of Grand Strategy

Eisenhower came to office in 1953 determined to end the Korean War, which had been stuck in protracted negotiations. In Eisenhower’s view, the Korean war had been far too costly in American blood and treasure: a slippery slope of over-committing American forces and resources to contain the spread of global communism. Containment had been the strategy of the Truman administration, but Eisenhower believed that large, frequent, and expensive deployments of US conventional forces globally would limit the full potential of American domestic prosperity. With these beliefs in mind, and with the Korean War finally ending in the summer of 1953, Eisenhower called together a group of national security experts to design a new strategy at Project Solarium. The project included high-profile defense and foreign policy luminaries such as George Kennan and John Foster Dulles. Project Solarium provided the essential inputs for NSC 162/2 and the foundations of the New Look strategy.

Eisenhower’s strategists first determined that the Soviet Union and the potential spread of communism represented long-term existential threats, and adopted the containment of both the Soviet Union and the spread of communism globally as a desired strategic outcome. The strategy planners also identified a prosperous American economy as a second goal, while recognizing that limited means were available to achieve both ends. Any resources invested to grow military capacity to achieve foreign policy goals would necessarily limit resources available to grow the domestic economy.

Defense spending was at historic highs during WWII, with the defense budget representing over 40% of US gross domestic product (GDP). After sharp post-war declines in defense budgets, large outlays became necessary to meet the demands of the Korean War, both in nominal terms and in defense spending as a percentage of GDP. The Korean War defense budget peaked in 1953 at $56.9 billion and 14.55 percent of GDP.2 Expenses like these drove Eisenhower’s belief that open-ended commitments to limited wars with conventional troops would slowly drain America’s economy and strength and jeopardize American domestic prosperity.

Force Posture and Nuclear Strategy

Recognizing the tension in limited means, the Eisenhower team also evaluated the resources available to the Soviets. The Soviet Union possessed a conventional mass advantage over the United States and its NATO allies. The Red Army could muster huge reserves of troops, and Soviet industry provided staggering volumes of war material. Eisenhower’s strategists assessed that the United States would be playing a losing game by trying to match conventional force levels in Europe while simultaneously responding to other Communist advances throughout the world. Instead, the administration leveraged America’s technological edge to develop an asymmetric strategy of nuclear deterrence that became known as the First Offset. By relying on superior nuclear capabilities, the US sought to neutralize the Soviet Union’s advantage in conventional forces while maintaining a more sustainable defense budget.

At this time, US military intelligence collected from U-2 spy planes indicated that the United States maintained the lead in nuclear technology and possessed the means to strike the Soviet Union. Although the Soviets also possessed nuclear weapons, they lagged behind in production and offensive strike capacity compared to Strategic Air Command. The policy of massive retaliation outlined in NSC 162/2 stated that in the event of Soviet aggression, the United States would respond with nuclear strikes (specifically tactical nuclear weapons) against Soviet military targets. American strategists believed that the mere threat of such a response would serve as a powerful deterrent, allowing the United States to maintain a smaller ground force in Europe and elsewhere.

Increasing the Nuclear Lead, Shrinking the Active Service

Eisenhower and his advisors correctly assessed that the United States could outproduce the Soviet Union in numbers of nuclear weapons. When Eisenhower took office, the US nuclear arsenal comprised 1,169 warheads. This arsenal would grow to 22,229 by 1961, a nineteen-fold increase. By comparison, the Soviet arsenal increased from 120 warheads in 1953 to 2,492 warheads by 19613. Thus the United States maintained a large lead in nuclear arsenal size throughout Eisenhower's administration.

As America’s nuclear lead grew, the size of its active duty military force shrank. Data from the Department of Defense Manpower Center shows a decrease of over 825,000 active duty personnel (about 25% of the force) between 1954 and 1960. The graph below shows how the follow-on policies of the Kennedy and Johnson administrations (discussed later) led to increases in the size of the US military force following the decreases of Eisenhower's New Look.

The Central Intelligence Agency and Covert Operations

Covert operations led by the Central Intelligence Agency formed another component of the New Look strategy, providing a means to actively contain the spread of global communism without costly ground wars. Under Eisenhower, the CIA led operations in Iran, Guatemala, Vietnam, and began operations in Cuba that would lead the Bay of Pigs invasion. Through the lens of long-term historical hindsight, the results of these operations were clearly mixed. However, in the short term, these operations enabled Eisenhower to achieve multiple foreign policy goals by restoring the Shah of Iran to preserve oil interests, installing a pro-United States military dictator in Guatemala, and ensuring a separate government in South Vietnam to oppose the Communists in the North when the French departed. Moreover, these achievements came at fraction of the cost of employing conventional forces for the same ends, and avoided the domestic political ramifications of far-flung wars.

Evaluating the New Look

Many critics of the New Look and the nuclear offset strategy have emerged in the intervening years. For one, the historian John Lewis Gaddis criticized the strategy of massive response as non-credible for any threats beneath the threshold of total annihilation – meaning that the United States would not resort to nuclear weapons except in cases where its adversaries brandished truly existential threats. According to this critique, there existed a wide range of Soviet behaviors that fell beneath the threshold of nuclear employment for which the United States lacked a credible response. For example, the United States and NATO had no acceptable military means by which to halt the heavy-handed Soviet suppression of the 1956 Hungarian revolution.

A narrow focus on the nuclear offset strategy may have also led to the Eisenhower administration’s failure to anticipate the impending shift in the strategic landscape from advances in Soviet rocket and missile technology. In 1957, the Soviets put the United States on notice with a successful test of the R-7 Semyorka, the first-ever intercontinental ballistic missile (ICBM). American fears were further exacerbated by the launch of Sputnik later in 1957. These advances provided the Soviets with their own offset, effectively eroding America’s strategic advantage. Armed with ICBMs, the Soviet Union could carry out nuclear strikes on the American homeland, and despite the size of America’s nuclear arsenal, any use of its weapons now carried a far graver risk of retaliation.

Under Eisenhower, the United States raced to catch up to the Soviet Union's new technological capabilities. The United States launched its first satellite on January 31, 1958, followed by its first Atlas ICBM in 1959. Though the United States still maintained a numerical advantage in warheads, both sides possessed roughly equivalent strike capabilities. Thus began a tense form of nuclear brinksmanship, or what strategist, game theorist, and war quant Thomas Schelling called a game of nuclear chicken. Using the vocabulary of game theory, Schelling argued that nuclear brinkmanship did not have a dominant strategy and thus represented a very dangerous strategic situation.

These developments led the incoming Kennedy administration to criticize the Eisenhower administration for allowing a missile gap to occur4 and to adopt a strategy of Flexible Response, focused on providing options at the strategic, operational, and tactical levels rather than relying solely on massive nuclear retaliation. The strategy of Flexible Response also included a build-up of conventional forces. Of course, like its successor, this new strategy faced its own pitfalls. The Lyndon B. Johnson administration remained committed to Flexible Response and ultimately became entangled in the Vietnam War.

Despite the criticisms leveled, Eisenhower’s New Look strategy passed the most critical test of any strategy - the effective alignment of desired ends with available means, a point that Gaddis acknowledged despite his criticism. The New Look carefully balanced competing foreign and domestic goals, ensuring that economic prosperity was not consumed by strategic competition.

Eisenhower’s farewell address is best known for its warnings of the growing influence of the military-industrial complex. The speech has had many interpretations, but it should be thought of as a reminder of what defines an effective strategy — balancing ends and means.

Conclusion

Today’s strategists must leverage emerging technologies to offset adversary advantages while balancing domestic priorities and avoiding unsustainable force structures. New emerging technologies include autonomy, artificial intelligence, space capabilities, biotech, and quantum computing. Harnessing these technologies for hard power will play an outsized role in this round of great power competition – even while the world still lives under the shadow of nuclear threats. Strategists will have to identify and exploit relative advantages and remain vigilant to disruptions in the strategic landscape from new innovations. The fundamental challenge remains the same: how to deter aggression, maintain strategic flexibility, and ensure economic sustainability in an era of prolonged competition.

Subscribe now

The views and opinions expressed on War Quants are those of the authors and do not necessarily reflect the official policy or position of the United States Government, the Department of Defense, or any other agency or organization.

Is a 35% probability of failure acceptable for a high-stakes hostage rescue mission?

According to the declassified Holloway Report on the Iran hostage rescue operation of 1980, this is exactly the risk unknowingly assumed by the planning team for Operation Eagle Claw. Due to a fundamental miscalculation regarding how many helicopters should be sent, every advisor and commander in the operational chain of command, including the Chairman of the Joint Chiefs and the President of the United States, approved a rescue mission that had only a 65% chance of even arriving at the first staging area successfully. Many years later, when asked if there was anything he would have done differently during his presidency, President Carter responded, "I wish I would have sent one more helicopter to get the hostages." In this article, I will share two classic operations research lessons: how varied numbers of helicopters would have affected the likelihood of Operation Eagle Claw’s success and the importance of communicating risk-to-mission through probabilities.

Context For a Complex Rescue

During the Iranian Revolution, armed Iranian students captured 53 American hostages at the U.S. Embassy in Tehran. Rescue mission planning began on 4 November 1979 and ran for nearly five and a half months. While the U.S. was pursuing diplomatic solutions, the rescue team faced a myriad of complex planning challenges, many of which stemmed from Tehran’s geographic isolation. To insert a hostage rescue team from the Gulf of Oman via the USS Nimitz, U.S. military helicopters would first need to complete a nearly 750 nautical mile flight to reach the hide site, Desert Two. This distance exceeded the helicopters' range, necessitating a rendezvous with C-130s at an intermediate staging area, Desert One, to refuel and embark the assault force. Finally, after a second flight to reach Desert Two, four RH-53Ds would need to wait until nightfall and launch again to conduct the rescue.

How Many Helicopters to Send?

At the time of the operation, the RH-53D had a fleet-wide mission capability rate of 74.0%. In an attempt to compensate for the potential failure of the rescue aircraft, military planners reasoned that if four helicopters needed to launch from Desert Two, five helicopters would need to arrive at Desert Two, six would need to arrive at Desert One, seven would need to be operational at launch, and eight helicopters would need to be staged aboard the USS Nimitz. In other words, if approximately three-quarters of RH-53D aircraft were mission capable, then with eight aircraft launched, six aircraft could be expected to reach Desert One on average.

Beware of Averages in Planning

Unfortunately, hindsight reveals that the rationale of multiplying the average fleet-wide mission capability rate by the number of aircraft launched fails to accurately capture risk. While intuitive, averages don't sufficiently account for the independent randomness inherent to each aircraft or communicate a likelihood of success. To quantify and understand risk-to-mission, commanders must consider probabilities of success, not average outcomes or expected values.

To better understand the impact of average-based planning assumptions, let's quantify the probabilities of success for different launching scenarios. For the scope of this article, I define mission success as six or more helicopters reaching Desert One in mission-capable condition. I will also assume that the determination of each helicopter's operational readiness is an independent event, like a coin toss. These events are called Bernoulli trials in probability theory, where the probability that a certain number of events achieve a desired outcome is modeled with a binomial distribution.

In this scenario, calculating the probability of at least six helicopters reaching Desert One is identical to calculating the probability of getting six or more heads in eight independent coin tosses, except instead of a 50/50 coin probability, we use the 74% mission capability rate of the RH-53D helicopters. The table below quantifies the risk-to-mission for potential flight plans under these assumptions, illustrating how the number of helicopters launched dramatically affects the likelihood of mission success.

This analysis highlights the rescue plan’s inherent vulnerability: with eight helicopters launched, there was only a 65% chance that six mission-capable helicopters would reach Desert One. Even increasing the number of launched helicopters to ten would have raised the probability to just 91%.

Execution

The Holloway investigation found that after the launch of eight helicopters on the evening of 24 April 1980, Helicopter #6 registered cockpit warnings for compromised rotor blade integrity, forcing the crew to abandon the aircraft. Vital flight and navigation instruments failed on Helicopter #5, forcing a return to the USS Nimitz. Helicopter #2 encountered a partial hydraulic failure but continued to Desert One. At this point, only five helicopters reached Desert One in operational condition (below the planned no-go threshold), and the mission was aborted.

Aftermath

During refueling for withdrawal from Desert One, a maneuvering RH-53D impacted one of the C-130s, igniting ammunition and fuel tanks onboard. Both aircraft were destroyed, and five Airmen and three Marines lost their lives. All remaining helicopter crew and ground forces retrograded on the unaffected C-130s. The hostages were not released until 20 January 1981, and a commission led by Admiral Holloway was ordered to investigate the rescue operation. The Holloway Report identified the need for better integration among the services and the acquisition of a long-range vertical lift capability, leading to the establishment of the V-22 tiltrotor program in 1982 and the creation of U.S. Special Operations Command (USSOCOM) after the passage of the Goldwater-Nichols Act in 1986.

Conclusion

Operation Eagle Claw highlights the critical importance of quantifying risk-to-mission and building resilience into mission plans. By relying on averages rather than probabilities, planners underestimated the likelihood of failure, and this vulnerability ultimately resulted in an aborted mission and a tragic loss of life. These lessons extend beyond a single operation. Effectively quantifying and communicating risk is an essential tool for commanders and planners, enabling more informed decisions in complex and high-stakes environments.

Subscribe now

Further Reading on Operation Eagle Claw and Military Readiness:

Moving Toward A Holistic, Rigorous, Analytical Readiness Framework

Operations Readiness Rollup

Military Readiness Toolkits from the MORS Community of Practice in Probability Management

The Flaw of Averages by Sam L. Savage.

Read More of Our Work Here:

FPV Attack Drone Proliferation: Unfreezing Conflict in Syria

FPV Math: Precision Mass for 21st-Century Warfare

Capability Analysis: AI Machine Guns for Drone Short-Range Air Defense

Carrier 2.0: The Drone Carrier Revolution

The views and opinions expressed on War Quants are those of the authors and do not necessarily reflect the official policy or position of the United States Government, the Department of Defense, or any other agency or organization.

Few US Navy sailors in the fleet today can earnestly swap sea stories about their encounters with naval mines. Naval mining can seem as though it belongs to another time: an era of boilers and bunker fuel, of minelayers and minesweepers. Recent advancements in the United States Navy’s (USN) mining and counter-mining capabilities have received little public interest. Yet mines are far from being a relic of the past. New Russian mines in the Black Sea have damaged commercial shipping vessels as recently as 2023, and mines will undoubtedly be a feature in any future conflict between the USN and the Chinese People’s Liberation Army Navy (PLAN) as the militaries of both nations invest in new naval mine technology. For example, the USN’s Orca XLUUV and the PLAN’s UUV300CD unmanned, subsurface mine vehicles will each allow operators to covertly seed and maintain naval minefields and carry out countermining operations.

Without the wisdom of sea stories as a guide, we must look elsewhere for insights into this resurgent capability. This article offers a whirlwind introduction to the basics of naval mining, outlines minefield measures of effectiveness (MOEs), and presents a baseline minefield simulation. Simulations like the one demonstrated here can help planners evaluate potential minefield configurations and make informed investment decisions concerning emerging mining technologies.

Subscribe now

The Basics: Mines and Missions

The three types of mines — moored, bottom, and moving — offer critical naval capabilities for achieving sea denial. These mines can be laid or seeded by aircraft, surface vessels, or subsurface vessels. Moored mines, like the one seen chained below, are perhaps the most frequently represented in media and laid by surface vessels. Bottom mines excel in relatively shallow water, and can be more easily laid in high quantities by bomber aircraft to achieve a dense minefield (i.e. one with no transitable gaps between effective mine radii). The distinction between these types of mines is vital because the physical characteristics and employment methods determine the suitable delivery vehicle, appropriate minefield location, and cost of each mine system. In this article, the simulation I will present only represents a homogeneous minefield consisting of bottom mines to simplify underlying assumptions.

Subscribe now

Naval mining operations are separated into three minefield operation types and four mission tasks. The Joint Force1 delineates minefield operations types based on the minefield’s location and intended effects, as shown in the table below. Planners must consider a minefield's covert and overt characteristics when selecting a mine delivery vehicle and scheduling the timing of operations.

Every minefield has a specified mission. The mission2 determines the minefield’s composition, shape, and desired minefield metrics. The table below lists the four minefield missions and their associated purposes.

Generally, blocking minefields require densely seeded mines and should be resilient to countermining efforts. Alternatively, disruption missions may require only a few well-placed mines to impede traffic or motivate vessels to alter routes. For example, consider Russian naval mining operations in the Black Sea. These minefields are in enemy territory and thus characterized as offensive, and given their density and effects, likely support a disruption mission. Correctly identifying the underlying mission of an adversary’s minefield is critical, since this mission analysis can give naval planners an idea of the minefield’s likely density, shape, vulnerabilities, and supporting land or sea positions.

Minefield MOEs

Minefield effectiveness is evaluated using domain-specific MOEs. This article considers three: initial threat on sea route, casualty density distribution, and expected casualties. Initial threat on sea route captures the minefield's ability to produce the first casualty. Casualty density distribution and expected casualties assess effectiveness over successive vessel transits and consider minefield degradation. Naval planners should use these MOEs to evaluate minefield performance and properly resource minefield missions. For example, an effective blocking minefield requires a dense configuration. It should present a high initial threat on sea route, have a sizable left skew in its casualty density distribution, and have high expected casualties. In the table below, each MOE is described for computational purposes.

Minefield Simulation

The simulation presented in this article uses an offensive minefield with a blocking mission as a base case. Each simulation generates 100 sequential vessels that transit along an identified sea lane. The plot below shows an example of the generated minefield and vessel tracks.

In the simulation, a vessel can only degrade the simulated minefield by a single mine. In other words, no other mines explode once the ship is hit, and simultaneous mine detonations cannot occur. As with wide shipping routes, the randomly generated vessel track can potentially avoid the minefield altogether.

List of Model Assumptions

1. Minefield is preemptively established by covert subsurface vessel(s).

2. Mine type uses a 1000lb bottom mine, similar to middle-sized Quickstrike and the retired Mk 52 Mod 63.

3. Bottom mines are laid at 100 feet.

4. Mines are spaced to cover the intended 600 x 400 yard minefield area.

5. Mines will actuate when the appropriate vessel passes within 120 feet.

6. Vessels transiting the minefield have ~4000-ton displacement, similar in size to the United States Freedom-class LCS or the Chinese Jiangkai II Frigate .

7. Vessels become casualties when passing within 80 feet of an actuated mine.

Minefield Experimentation

This offensive blocking scenario is used to set up a simple experiment. In this experiment, the minefield capacity and probability of actuation are assessed against each other. In other words, the experiment looks to answer the question “Which has the bigger impact on minefield performance: minefield capacity or probability of actuation?”. This simple example can provide insight into whether investments should be made in improving technology, like actuation probability, or in obtaining larger quantities of lower-performing mines. The table below outlines the varied minefield parameters.

The first two scenarios assume mines are spaced by 80-yard intervals and vary the probability of actuation from 40% to 80%. Increases in actuation probabilities serve as a proxy for performance increases, like those that could result from a new generation of mine technology. We could also choose to vary the effective casualty range, minefield size, and actuation distance as further proxies for performance and underlying technological changes. Here, Scenarios 3 and 4 continue to vary the probability of actuation, but also increase the minefield capacity to 71 mines. Note that this increase in capacity results in an increase in the density (not the area) of the minefield. The chart below shows the resulting minefield’s casualty density distribution given this variation in input parameters.

The casualty density distributions indicate similar minefield performance for the two sets of scenarios despite the increase in actuation probabilities from 40% to 80%. In simple terms, dense minefields produce more casualties than sparse minefields. While this concept is easily digestible, the underlying requirements for sourcing, seeding, and maintaining 71 instead of 41 mines can be substantial. Assuming an XLUUV configured like the Orca can carry and emplace 8 tons of cargo or 16, 1000-lb naval mines, five total XLUUVs would be needed to seed the denser minefield in a single operation. While two additional vehicles may sound small, at $450M per XLUUV, the added capacity means the fictitious 5-vehicle squadron would cost over $2B.

The table below shows each scenario's initial threat and expected casualty MOEs.

Reinforcing the casualty density distribution, Scenarios 3 and 4 produce higher expected casualties. The simulation also indicates a negligible difference in expected casualties as a result of increasing probability of actuation. This is most likely due to the dense configuration of all scenarios, which creates casualties on subsequent mines after initial failures to actuate. Further research would be necessary to discount the need for investment in such advances.

Lastly, the four scenarios produce very little variation in initial threat on sea route outcomes. The constant minefield area likely contributes to this consistent performance. To confirm this relationship, a minefield covering a larger area with 37-mines was simulated. The minefield, with an increased 1400-yard frontage, produced an 85% initial threat on the sea route. Planners should seek to maximize the initial threat of a blocking minefield while balancing expected casualty performance. Potential options include increasing the minefield area or choosing a sea lane that requires vessels to transit the minefield.

In this example, the increase in expected casualties from seeding more mines might be worth the investment, given this mission and its operational implications. Simulations like the one above can build context around mining and countermining operational-level problems and can provide insights into strategic-level investments.

Conclusion

Though often neglected, naval mines remain a powerful and versatile tool in maritime warfare. While acknowledging the cost, conventions restricting mining operations, and complexities of mine warfare, the necessity of its capability can not be understated. With advancements in technology and the prominence of great power competition, the operational significance of naval mining will only increase. Models like the one presented can help answer important questions when investing in this critical capability. In this initial look, we can intuit that the density of mines plays an outsized role in effectiveness when compared to the actuation rate. This type of insight is invaluable for guiding capability investment.

The views and opinions expressed on War Quants are those of the authors and do not necessarily reflect the official policy or position of the United States Government, the Department of Defense, or any other agency or organization.