To Infinity and Beyond

A Future For The U.S. Army Cavalry

By MAJ R. Atticus Blair

July 1, 2026 | Armor, Summer 2026 Edition

Read Time: < 19 mins

For its first 150 years, the U.S. Army fused reconnaissance and shock action into a single branch because both missions required the same platform — the horse.1 Mechanization inherited the fusion without reexamining it; the tank replaced the horse, and cavalry kept both jobs. That self-induced dichotomy within the Armor branch has forced commanders to fight blind. Observer-controllers at the National Training Center (NTC) and the Joint Multinational Readiness Center (JMRC) report the same failure decade after decade: cavalry squadrons built like maneuver battalions advance, make contact through direct fire rather than standoff sensing, and fight meeting engagements they should have anticipated.2 In the 1990s the diagnosis was poor discipline — scouts fighting instead of reporting.3 In the 2000s and 2010s, two decades of counterinsurgency (COIN) masked the problem by pivoting away from a near-peer adversary capable of punishing bad reconnaissance. When the U.S. Army returned to large-scale combat operations (LSCO) training in the late 2010’s, the old failures returned.

Outmanned and outgunned since Russia re-invaded in February 2022, Ukrainian brigades built dedicated drone companies that fused reconnaissance and strike into compact, low-cost kill chains — collapsing the targeting cycle from hours to seconds, sustaining persistent coverage across a thousand-kilometer front, and killing tens of thousands of Russian soldiers every month.45 In June 2024, Ukraine became the first country in history to establish a military branch for unmanned systems.6 The U.S. Army, by contrast, has $1.4 billion in congressional appropriations for small unmanned aerial systems (sUAS), a program to field 340,000 attack drones in two years, and no branch to receive them, no school to train their operators, and no career field to retain their expertise.7

For the first time since the horse, the U.S. Army can optimize reconnaissance around persistent presence, minimal signature, and expendable sensors — separated entirely from the armored formations built to close with and destroy the enemy. The reconnaissance gap and the unmanned systems gap are one problem: neither function has a branch, a proponent, or a career field. The U.S. Army must create both in a single act — a Cavalry Branch chartered to find the enemy and maintain contact through expendable and autonomous platforms.

This paper examines the reconnaissance problem, the unmanned systems problem, their convergence, and the institutional architecture required to solve them together.

Equestrians

For four thousand years, cavalry performed two jobs under one name. Reconnaissance required dispersion, patience, and stealth. Shock required mass, speed, and violence.8 The two coexisted not because they were compatible, but because both required the same platform: the horse. When the U.S. Army organized mounted forces, they bundled divergent functions into a single branch — the defining skill was not scouting or charging, but horsemanship.9

The tank took the horse’s place, but the U.S. Army never separated the functions the horse had carried. The structural consequences have compounded for nine decades. A cavalry squadron fielding Bradley Fighting Vehicles (BFVs) and Abrams tanks carries roughly the same maintenance burden, fuel demand, and electromagnetic signature as the combined arms battalion it is supposed to precede — the punishing maintenance burden defeats the one thing reconnaissance demands above all else: maintain a persistent presence. Equipping reconnaissance formations with maneuver platforms forces commanders to hold reconnaissance assets back, to fix them. When reconnaissance units are built like maneuver battalions, they fight like maneuver battalions, and the institution rewards them for it. BG Lee Quintas, the 48th Chief of Armor, assessed directly whether the cavalry squadron could accomplish its reconnaissance and security missions. His answer was “NO.”10

The Battle of 73 Easting illustrates how the fusion played out. In February 1991, 2nd Armored Cavalry Regiment led VII Corps into Iraq as a covering force, executing a security mission, not a reconnaissance mission. Eagle Troop, commanded by CPT H.R. McMaster, made contact with the Tawakalna Division of the Republican Guard through direct fire, not standoff sensing. CPT McMaster’s Troop destroyed dozens of Iraqi tanks in minutes. Two M3 BFVs, reconnaissance platforms by designation, killed five T-72s. The regiment succeeded because it was equipped and tasked as a maneuver formation, and it fought like one. The U.S. Army celebrated 73 Easting as confirmation of the cavalry’s value as a fighting force — and it was. But as Brian Steed argues in Battle Analysis: The Battle of 73 Easting, Operation Desert Storm was an aberration, not a model: an outmatched opponent, in open desert, blind beyond eight hundred meters, against a force that could kill at three times that range.11 The institutional lesson the U.S. Army drew — that cavalry’s primary contribution is combat power — left reconnaissance without a champion. If your reconnaissance force is indistinguishable from your maneuver force, same platforms, same signature, same sustainment tail, then it is redundant. Army Structure (ARSTRUC) proved the U.S. Army agreed. The maneuver formations absorb the security missions of the cut squadrons. No formation in the current force structure is designed to see the enemy before making contact with him.12

Two decades of brigade-centric COIN delayed the reckoning. When the U.S. Army returned to division-level operations, it tested the reconnaissance problem head-on at NTC Rotation 20-10 — the first division-level rotation in the center’s history. The 1st Infantry Division fielded a cavalry squadron reinforced with aviation, artillery, and support elements to provide the division’s reconnaissance and security. The hope was that a squadron could cover a division’s frontage. By the end of the rotation, the squadron had been reinforced to the point that it resembled another maneuver brigade — one that required the same sustainment tail and was constrained by the same maintenance tempo.13 No such formation has emerged from the experiment.

The 2024 ARSTRUC made the cut explicit. In a fixed end-strength environment, the Army eliminated fourteen light cavalry squadrons from infantry and Stryker brigade combat teams (BCTs), freeing roughly ten thousand billets for Multi-Domain Task Forces.14 The Armor School called the loss “not ideal.”15LTC Amos Fox was blunter: the U.S. Army was replacing close-combat reconnaissance with theater-level sensor organizations that cannot control terrain — the same substitution that produced the Battlefield Surveillance Brigades of the mid-2000s, quietly shuttered within a decade after proving operationally hollow.16 The squadrons were not cut because reconnaissance was unimportant. They were cut because squadrons built like maneuver formations were expensive and redundant — they were the easiest billets to take. Brigades have responded by standing up multifunctional reconnaissance companies (MFRCs) from existing billets — units without a proponent, a training pipeline, or dedicated personnel, staffed by soldiers whose primary military occupational specialty (MOS) development suffers for the assignment. Every MFRC is different; no doctrinal publication defines their employment. The improvisation validates the requirement. It does not solve it.17 A 2016 United States Army Forces Command (FORSCOM) study warned that reconnaissance expertise takes generations to build and can be lost in a decade or less. That decade started in 2024.18

The underlying problem is not training, equipment, or doctrine. It is institutional identity. Reconnaissance has never had its own branch, proponent, school, or career field.19 It has always been a secondary mission inside organizations whose primary identity, and whose officers’ career incentives, pointed toward direct combat. Until that changes, the reconnaissance problem will reproduce itself in every war, every restructuring, and every round of cuts. Reconnaissance needs an institutional home. The question is what that home should look like.

Experts

Ukraine has been outmanned and outgunned since February 2022. The entire war is an economy-of-force operation sustained for years against a numerically superior adversary. Russia attacked across a thousand-kilometer front with more artillery, more tanks, and more soldiers. That disparity forced Ukraine to find the most efficient way to guard the line at the lowest cost in blood and treasure. The answer was dedicated formations flying persistent, cheap, expendable drones.20

Ukraine institutionalized what combat taught it. In June 2024, it became the first country in history to establish a military branch dedicated to unmanned systems.21 By December 2025, Ukrainian drones were killing an estimated thirty-three thousand Russian soldiers per month.22The 25th Sicheslavska Brigade shows what this looks like at unit level: a UAS company built around reconnaissance and first-person view (FPV) strike platoons, operating in dispersed two-pilot cells, with concealed headquarters handling maintenance and resupply from urban cover. Paired with artillery and encrypted targeting applications, FPV teams collapsed the kill chain from hours to seconds.23 These FPV teams have evolved to fill the security function—the fight to see is no longer confined to the ground; both sides are fighting to control the sky above the front line, where success sets the conditions for maneuver, and failure leaves commanders blind.24

The U.S. Army, meanwhile, is experimenting. The Transformation in Contact (TiC) initiative pushes commercially available drones into select brigades under a “buy, give, inform” model — the U.S. Army fields systems, observes what works, and adjusts.25 The 3rd Infantry Division now operates over 150 drones and is developing what it hopes will become the Army-wide qualification course for sUAS.26The results are promising. They are not institutional. Every TiC brigade bought different drones.27 There is no single authority deciding which systems a given echelon needs, no common training tables, and no career field for the soldiers operating them. The Unmanned Advanced Lethality Course at Fort Rucker trains twenty-eight students per three-week iteration, drawn from four different branches. Its course director built the curriculum in ninety days and called it plainly: catch-up.28Twenty-eight students is not a pipeline. It is an additional skill identifier.

The institutional logic behind this diffusion is that drones should be organic to every formation at every echelon — that the U.S. Army does not need drone specialists because every soldier will be a drone operator.29 The logic is appealing but wrong. It assumes latent capacity exists within formations and that assumption has been proven false. Three independent experiments confirm this. The British Army’s Defense Science and Technology Laboratory (DSTL) Urban Phalanx study found that infantry section commanders in urban combat could not simultaneously command their sections and process digital sensor information; the solution was a dedicated systems operator billet and a restructured section built around reducing cognitive load on junior leaders.30 The U.S. Marine Corps reached the same conclusion independently, creating a squad systems operator whose primary task is managing the squad’s drone and digital systems so the squad leader can focus on the fight.31 At JMRC, even U.S. cavalry squadrons that retained their scouts after ARSTRUC struggled to absorb UAS — reconnaissance experts, handed drones, could not fly them and maneuver BFVs at the same time.32

A soldier manipulating a controller is not manipulating a rifle, a mortar tube, or a helicopter. A squad leader processing a drone feed is not commanding his squad. The generalist model forces a dilemma that the technology does not require. A drone operator’s sensor is airborne. There is no reason to put the operator at the point of contact when a covered position hundreds of meters back provides the same capability with none of the dilemma. Specialization does not remove unmanned systems from the rest of the force. It makes them work.

Congress recognized the gap before the U.S. Army did. In June 2024, the House Armed Services Committee proposed a Drone Corps as a basic branch responsible for Groups 1 through 3 UAS and counter-UAS (CUAS).33 The U.S. Army opposed it. Under Secretary Gabe Camarillo called it “premature.” The Administration called it “an unwarranted degree of specialization.”34 The provision was stripped from the final legislation. The word “premature” has a history. It was premature to separate the Aviation Branch from the Transportation Corps — until 1983, when the U.S. Army acknowledged it had wasted critical human capital distributing aviation assets without a dedicated proponent. It was premature to create the Cyber Branch — until 2014, when growth under the Signal Branch proved inadequate. It was premature to establish Special Forces — until 1987. In every case, the capability stagnated under someone else’s roof until it got its own.35

The money is already flowing. The U.S. Army has $1.4 billion in congressional appropriations for sUAS industrial base expansion, a Drone Dominance program fielding 340,000 attack drones over two years, and a Secretary of War directing every division to be equipped with unmanned systems by the end of 2026.36 What it does not have is a branch to receive these systems, a school to train their operators, a proponent to write their doctrine, or a career field to retain their expertise. The institutional architecture to spend that money well does not exist. Unmanned effects need an institutional owner. The question is who.

Charter

Two problems, one solution. The capability that unmanned systems perform best — persistent, low-signature, expendable sensing and strike — is the capability that reconnaissance has always required but never had the platform to deliver. Drones do not tire. They can be lost without losing a crew. They can loiter for hours over a named area of interest that a mounted patrol must pass through in minutes. For nine decades, reconnaissance needed persistent presence at acceptable cost and risk. Unmanned systems provide exactly that. The branch that owns reconnaissance should own unmanned effects — binding them — not because the missions are adjacent, but because they are the same. The battlefield forced Ukraine to converge reconnaissance and unmanned effects. The question for the U.S. Army is not whether this works — Ukraine has answered that — but whether it will make the institutional decision deliberately or wait for a war to impose it.

The forward observer precedent makes the logic explicit. The First World War created the three-dimensional battlefield; from the trenches emerged the forward observer, a dedicated career that made indirect fires work for the whole force.37 In the century since, the forward observer has survived every technological transition — from visual observation to laser designation to digital fire networks — because the function requires the expert. Forward observers create a persistent presence by embedding into supported units; other artillery formations deliver fires into the deep area.38 Unmanned systems should work the same way: specialists who attach to every commander who needs them, while dedicated UAS formations handle the security zone via reconnaissance and strike. Every combat arms soldier today can call for fire in extremis, but the U.S. Army does not pretend that every infantryman is a fire supporter. The same principle applies to drones. Baseline unmanned literacy belongs in every combat arms and combat support training pipeline — but mastery requires a dedicated career field, just as it does for artillery.

The cavalry squadron is the conversion vehicle. The doctrinal tasks transfer cleanly: screen, area reconnaissance, maintain contact.39These are descriptions of what drones already do in Ukraine.40 The fourteen squadrons cut under ARSTRUC provide the billets.41Positions currently filled by Bradley mechanics and vehicle crew members convert to drone operators, firmware technicians, and electronic warfare adaptation specialists. The organizational architecture — headquarters troop, forward support company, command relationship to the brigade — remains intact. Technical skills change; mission, structure, and tactical expertise do not.

This proposal does not diminish the Armor branch. It liberates Armor from a self-induced dichotomy — mass, speed, and violence against dispersion, patience, and stealth — by giving the quiet half its own home. The U.S. Army’s deepest identity is combined arms: the strength of the U.S. Army is that we can take two hundred civilians off the street and six months later execute a combined arms live fire. Each soldier masters one function. Combined, they become the most lethal tactical formation on the planet. A Cavalry Branch adds an arm to the combination — an arm, worth mastering.

The branch charter should read: reconnaissance, unmanned effects, and counter-unmanned effects across all domains and echelons. Not a drone branch. Not an FPV branch. A branch defined by the function of finding the enemy, maintaining contact, and delivering effects through expendable platforms — while denying the enemy the ability to do the same. This charter gives every brigade commander a reason to demand the capability and every division commander a reason to resource it. For the first time since the horse, reconnaissance and its enabling platform share a single institutional home.

The U.S. Army should establish the Cavalry Branch at the Maneuver Center of Excellence, Fort Benning. Proximity to Infantry and Armor — the primary supported branches — ensures combined arms integration from the start. The first commandant should be a combat arms officer with operational UAS experience, selected at brigadier general, charged with standing up the branch school, writing initial doctrine, and managing the career field within twenty-four months of authorization.

Sourcing begins with conversion, not creation. The fourteen cavalry squadrons cut under ARSTRUC provide the billets.42 The first unit to convert should be an armored BCT (ABCT) cavalry squadron — a formation that retains its organizational architecture and reconnaissance mission — equipped with systems already entering the force through TiC. One squadron converts within twelve months of branch authorization as proof of concept. One squadron per brigade combat team within five years achieves initial operating capability across the force.43

The MOS structure requires three tracks. First, an operator track for UAS pilots and sensor operators across Groups 1 through 3, with initial training built around the twenty-to-twenty-five-hour simulator proficiency threshold already validated by the Unmanned Advanced Lethality Course and Ukrainian training data.44 Second, a maintainer and electronic warfare technician track responsible for field repair, firmware management, frequency adaptation, and cUAS operation — the sustainment depth that no additional-duty model can provide. Third, a warrant officer technical integrator track modeled on the 390A Robotics Technician warrant officer that U.S. Army Special Operations Command (USASOC) established in 2025: a career-long specialist bridging tactical employment and system acquisition, ensuring operational feedback reaches program managers and doctrinal updates reach operators.45 All three tracks feed a single proponent, eliminating the fragmentation that currently spreads UAS responsibility across the Aviation Branch, the Military Intelligence Branch, and individual brigade initiatives.46

The risks are real. In a fixed end-strength environment, every billet assigned to the new branch comes from somewhere else. The deeper concern is that the U.S. Army sinks institutional energy, facilities, and personnel into a branch that becomes obsolete. The character of war will change, but it will not regress from unmanned systems — and the uncertainty that defines the nature of war will always force commanders to demand to see their enemy before they fight him. The Aviation Branch, the Cyber Branch, and Special Forces faced every one of these risks at establishment. All three survived because the capability they owned was too important to leave as someone else’s secondary mission.47

Decision

The U.S. Army has the money, the technology, and the operational evidence. What it lacks is the institutional decision to put them together. Make it.

Notes