The Last Generation of the Infantryman?

By Matthew Parish

Monday 28 September 2026

There is something almost timeless about the infantryman. Alexander the Great would have recognised him, Napoleon would have recognised him and the soldiers who crossed the Somme in 1916 would recognise him immediately. He is tired, frightened and overloaded with equipment. He sleeps badly, eats when he can and spends much of his existence waiting for something terrible to happen. Eventually he walks across a piece of ground carrying a weapon while another human being tries to kill him. The equipment has changed beyond recognition over several thousand years, but the essential proposition has survived: territory is controlled ultimately because armed human beings stand upon it.

Aircraft may destroy positions, artillery may pulverise them and tanks may break through them, but somebody eventually has to occupy the trench, building, road junction or village. That somebody has traditionally been the infantryman. Ukraine is now conducting an extraordinary experiment in whether this ancient proposition can finally be qualified. The answer, at least for the moment, is that the infantryman remains indispensable, but across the Ukrainian battlefield human beings are gradually withdrawing from tasks that only a few years ago necessarily required soldiers to expose themselves to enemy fire.

Drones conduct reconnaissance and strikes. Ground robots carry ammunition, food and water. Machines evacuate casualties and recover the dead. Unmanned vehicles undertake engineering work and reconnaissance-strike missions. Artificial intelligence increasingly assists in interpreting the vast quantity of imagery and battlefield information generated by these systems. By July 2026 Ukraine’s Defence Procurement Agency had contracted more than 22,000 unmanned ground vehicles for the year, nearly twice the number contracted during the whole of 2025. In June alone, Ukrainian units conducted more than 16,600 logistics and evacuation missions using ground robotic systems.

This is no longer experimentation at the margins of warfare. Ukraine is beginning to construct an army in which the machine goes first and the human being follows only when necessary.

The empty battlefield

Imagine a Ukrainian defensive position somewhere in Donetsk Oblast ten years from now. A Russian reconnaissance drone approaches and is detected not by a soldier staring into the sky but by sensors feeding information into an artificial-intelligence system. An interceptor drone launches, manoeuvres towards the intruder and destroys it. Several kilometres ahead, autonomous reconnaissance aircraft patrol the approaches to the Ukrainian position while ground sensors detect movement. Other drones investigate, and software compares their imagery against millions of frames of previous battlefield data to distinguish vehicles, equipment and soldiers.

Russian troops begin to advance. The first Ukrainian response does not necessarily involve a Ukrainian soldier firing a weapon. FPV drones emerge from concealed positions, robotic vehicles move towards likely avenues of approach and artillery receives targeting information from an increasingly integrated network of sensors. Human operators remain behind the system, supervising it, interpreting what it sees and authorising decisions that machines cannot or should not make for themselves. The infantry may meanwhile remain some distance behind this electronic screen until human presence becomes necessary.

This is not yet the Ukrainian battlefield in its entirety, but neither is it science fiction. In April 2026 the Ministry of Defence described its Drone Line programme as the implementation of a new doctrine of technology-driven warfare, intended to provide continuous aerial support to infantry and establish a contested zone extending approximately 10 to 15 kilometres from Ukrainian positions in which Russian forces cannot advance without being exposed to attack.

The significance of that distance should not be underestimated. Historically the infantryman’s principal danger began as he approached the range of enemy weapons. Ukraine is attempting instead to populate that dangerous space increasingly with machines, sensors and remotely operated weapons. The infantryman does not disappear, but the battlefield immediately in front of him begins to empty of friendly human beings.

The robot carrying the ammunition

Revolutions often begin with unglamorous things, and the most consequential battlefield robot may not be a machine carrying a machine gun. It may be a small tracked vehicle carrying boxes of ammunition, batteries, food and bottles of water. Supplying frontline positions has become exceptionally dangerous because Russian drones observe roads, tracks and fields far behind the traditional contact line. A vehicle repeatedly travelling towards a position establishes a pattern; eventually somebody sees it, follows it and identifies both the vehicle and its destination.

The obvious solution is to remove the driver, and Ukraine is doing precisely this on an increasingly substantial scale. Of the more than 22,000 unmanned ground vehicles contracted for the Defence Forces by July 2026, the overwhelming majority were logistics systems, followed by engineering and combat platforms. Thousands had already reached military units, and the Ministry of Defence was simultaneously making robotic systems directly available to units through the DOT-Chain Defence procurement system.

The implications are profound because a considerable proportion of battlefield casualties historically arise not from dramatic assaults but from the mundane necessities of keeping an army alive. Somebody must carry ammunition forward, deliver food and water, recover damaged equipment and travel along roads watched by the enemy. If an inexpensive machine can make some of those journeys instead, the economics and morality of warfare change simultaneously. Destroying a robot is expensive and inconvenient. Destroying a soldier is irreversible.

The robotic stretcher bearer

Casualty evacuation provides an even more compelling example. The traditional battlefield creates a terrible secondary danger: a soldier is wounded and his comrades must then expose themselves to rescue him. One casualty may consequently produce several more, while persistent drone observation has made the problem still worse because an enemy operator may observe both the casualty and those attempting an evacuation.

A ground robot does not care that the road ahead is frightening. It does not have a wife, husband, parents or children. It can move into a dangerous position, retrieve a casualty and attempt to return. If the machine is destroyed, another can eventually be manufactured. Ukrainian procurement authorities reported that logistics and evacuation missions by ground robots had risen by 122 per cent between January and June 2026, suggesting that these systems are moving rapidly from occasional experiments towards routine battlefield tools.

This may prove more revolutionary than the robotic machine gun. The history of military technology is usually written around increasingly effective methods of killing people, but robotics may be equally significant because it offers increasingly effective methods of preventing soldiers from having to be present where people are being killed. A military revolution need not consist solely of discovering a better way to destroy the enemy. It may also consist of discovering that a machine can undertake the journey that previously killed your friends.

The robot with a gun

The next step is more uncomfortable. Ukraine is also developing and deploying armed ground robotic systems intended for reconnaissance and strike missions. In July 2026, for example, the Ministry of Defence authorised the Tanchik Droid 12.7, an unmanned ground vehicle equipped for reconnaissance-strike operations whose stated purpose includes undertaking dangerous missions in areas exposed to intense enemy fire while reducing risk to Ukrainian personnel.

The engineering logic is compelling. A human machine-gunner requires food, water, body armour, sleep, medical support and evacuation if wounded. He experiences fear and exhaustion, while training him takes months and losing him may mean losing years of accumulated military experience. A robotic firing platform principally requires electricity, maintenance and communications. The comparison is nevertheless misleading in one important respect: machines remain remarkably stupid compared with people.

Mud defeats them, rubble defeats them, electronic warfare interferes with them and communications disappear. Batteries fail, tracks break and cameras become obscured. An infantryman encountering an unexpected ditch may climb through it, walk around it or improvise another solution. A robot may simply stop. Human beings possess an extraordinary general-purpose intelligence that engineers still struggle to reproduce: a frightened 20-year-old soldier can interpret terrain, understand a commander’s intention, recognise an injured civilian, improvise when equipment fails and distinguish a suspicious movement from a genuine threat. That collection of abilities remains enormously difficult to automate.

Artificial intelligence enters the battlefield

Yet the gap is narrowing, and Ukraine is deliberately trying to narrow it further. In March 2026 the Ministry of Defence announced that real Ukrainian battlefield data would be made available for training artificial-intelligence models used in unmanned systems, explicitly identifying greater autonomy as an objective. In August it opened the Avengers Labs platform to Ukrainian defence companies. The platform contains an annotated dataset of some five million battlefield frames covering tanks, artillery, air-defence systems, infantry and aerial targets.

The scale of AI integration is already considerable. By April the Ministry reported more than 200 companies involved in producing AI-enabled drones, more than 300 relevant developments registered through Brave1 and more than 70 AI or computer-vision systems being actively used at the front. In June the Ministry established the Defence AI Center A1 around the explicit ambition of creating an ‘AI-driven army’, using artificial intelligence for data analysis, decision-making and target engagement.

By August the Ministry was describing a battlefield information system in which the Avengers AI platform could identify military equipment by analysing more than 100,000 video streams each month. In September A1 was already testing SPECTR with combat units, an AI system intended to analyse operational information, identify patterns and generate recommendations for military planning. These developments do not mean that a computer has replaced the Ukrainian commander, but they demonstrate how quickly machines are moving from carrying weapons towards helping humans interpret the battlefield upon which those weapons are used.

One soldier, ten machines

This may prove to be the decisive transition. The first generation of military drones generally required something approaching one human operator for one aircraft, which did not eliminate the soldier so much as relocate him. The operator remained indispensable throughout the mission. Increasing autonomy changes the arithmetic because if one soldier can supervise five machines, then ten and eventually dozens, the relationship between manpower and military power begins to separate.

At that point a country with fewer soldiers might nevertheless place more weapons and sensors on the battlefield because each human being commands a constellation of machines. Navigation, target recognition and routine manoeuvring can increasingly be delegated while the human operator supervises several systems and intervenes at the points at which judgement remains necessary. The soldier moves upwards in the chain of abstraction: instead of flying the drone, she commands it; instead of driving the robot, she assigns its destination; instead of watching every camera continuously, she receives the information the machine considers important.

The implications for Ukraine are obvious. Russia possesses the larger population and can therefore attempt to exploit a demographic advantage in a prolonged war of attrition. Ukraine cannot manufacture millions of additional Ukrainians, but it can manufacture machines. The strategic purpose of autonomy is consequently not technological elegance. It is to increase the military power generated by every Ukrainian who must remain physically exposed to danger.

The last kilometre

There remains an enormous obstacle to the dream of an empty battlefield because somebody still has to occupy the ground. A drone can destroy a trench, a robot can fire into it and an autonomous vehicle can deliver ammunition beside it, but none of those things necessarily establishes political control over the territory. Occupation remains fundamentally a human activity because territory is not merely geography: it contains people.

Soldiers search buildings, question civilians, accept surrender, guard prisoners, distinguish friends from enemies and exercise judgement amidst situations that refuse to fit neatly into algorithms. This is why Ukraine’s Drone Line does not purport to eliminate infantry. Its official purpose includes providing continuous aerial support and protection to infantry while attacking the enemy before Russian forces can reach Ukrainian positions.

The infantryman therefore survives, but his geography may change. Instead of spending days in the foremost trench, he may increasingly remain behind a screen of sensors and machines until human presence becomes necessary. Instead of carrying ammunition through the most dangerous kilometre, he may receive it from a robot. Instead of walking forward to discover whether an enemy position remains occupied, he may send a machine first. Robotics does not need to abolish infantry to transform infantry warfare; it merely needs to reduce the proportion of a soldier’s life spent exposed to the enemy.

The economics of expendability

There is another brutal advantage to this transformation. Military organisations have always treated equipment and people differently because equipment is replaceable. A commander may rationally accept the destruction of ten inexpensive drones to eliminate one valuable enemy system, whereas no comparable calculation involving ten soldiers is morally or politically tolerable.

Machines therefore permit a kind of tactical experimentation that human warfare cannot. Send a robot down the road and, if it encounters a mine, something has been learned. Send another by a different route and, if it is jammed, something else has been learned. Eventually a surviving machine may discover a path through the battlefield. The process is almost Darwinian, except that the things being destroyed are manufactured objects rather than young men and women.

This may eventually reverse one of warfare’s oldest and ugliest assumptions. Industrial warfare frequently expended soldiers in order to preserve or exploit expensive equipment. Twenty-first-century warfare may increasingly expend enormous quantities of cheap equipment in order to preserve soldiers. That would represent an extraordinary improvement for the army capable of achieving it, even if the battlefield itself becomes no less destructive.

The terrible alternative

Unfortunately Russia can build robots too, and every Ukrainian technological innovation creates an incentive for a Russian equivalent. The increasingly empty battlefield may therefore become a place where enormous populations of machines fight one another while human beings retreat progressively farther from the immediate point of contact. That does not necessarily make warfare humane; indeed it may create new dangers by reducing the immediate human cost perceived by governments contemplating the use of force.

There is also the more profound question of autonomy. A machine that carries ammunition raises few moral difficulties, while a machine that identifies a person and independently decides to kill him raises enormous ones. Between those extremes lies an expanding territory of partial autonomy in which humans select areas, objectives or categories of targets while software performs increasing portions of navigation, identification and engagement. The technological pressure will always be towards speed because a machine that waits five seconds for human approval may lose to a machine that decides in five milliseconds.

The ethical and legal problem therefore grows directly out of the military advantage. The more successfully humans remove themselves from the immediate battlefield, the more decisions must either be made remotely or delegated to machines. Saving the soldier’s body from exposure may consequently create a different problem: determining how far the soldier’s judgement can also be removed before meaningful human control disappears.

The disappearing front line

The strangest consequence of this technological transformation may ultimately be conceptual. For centuries armies possessed front lines because soldiers had front lines: there was somewhere that one mass of human beings confronted another. Drones have already begun dissolving this distinction in Ukraine. A soldier tens of kilometres behind the contact line may still be attacked, logistics routes have become battlefields, headquarters are battlefields and drone operators hunt other drone operators.

As autonomous systems proliferate, the battlefield may become less a line than an enormous contested volume. Machines will patrol it continuously, sensors will observe it and algorithms will attempt to make sense of what those sensors see. Human beings will enter the most dangerous portions when necessary rather than necessarily remaining there continuously.

This would reverse the traditional relationship between soldier and machine. During the twentieth century machines accompanied soldiers onto the battlefield. During the twenty-first, soldiers may increasingly accompany machines.

Not the last infantryman

So is Ukraine witnessing the last generation of infantrymen? Almost certainly not. Human beings remain unmatched general-purpose military systems. They climb, crawl, improvise, interpret and decide; they understand context, recognise when the plan has become absurd, comfort frightened civilians and accept the surrender of enemy soldiers. No Ukrainian robot yet possesses anything approaching that collection of abilities, and there is little reason to imagine that infantry will disappear merely because increasingly sophisticated machines surround it.

But the infantryman’s role is changing with extraordinary speed, and Ukraine is now institutionalising that change rather than merely improvising it. The relevant question is therefore not whether robots will replace soldiers. It is how many dangerous tasks can be removed from soldiers before human presence becomes indispensable. Every kilometre of ammunition delivery transferred to an unmanned vehicle matters, every wounded soldier extracted by a robot matters, every reconnaissance patrol performed by a drone matters and every dangerous approach investigated by a machine before a Ukrainian soldier walks into it matters.

Technology does not need to abolish the infantryman to transform warfare. It merely needs to keep him alive for the moments when only a human being will do. Ukraine’s most consequential military technological revolution may therefore prove not to be the creation of machines capable of fighting like people, but the construction of a battlefield upon which people increasingly do not have to stand.