The first modern Immortal Battalion costs 400 million Euro

“Victory smiles upon those who anticipate the changes in the character of war, not upon those who wait to adapt themselves after the changes occur.” – Giulio Douhet

Ukraine suffers from an acute shortage of soldiers, which is not surprising since being a soldier may get you killed. Would it not be easier to recruit soldiers if death was off the table? What if we took a larger evolutionary step and created autonomous and remote-controlled fighting units, starting with say a battalion? Such a battalion would enter every new engagement at full strength, always smarter than it was before. It would be immortal, a useful property in war.

Among the elite forces of antiquity, few units carried as much mystique as the Persian Immortals. This legendary corps of 10,000 soldiers formed the backbone of the Achaemenid military machine and served as both the king’s personal guard and shock troops in battle. Their name derived from their unique trait: whenever a soldier fell, another immediately replaced him, keeping the unit’s strength constant and creating the illusion of immortality. This continuity projected an image of inexhaustible power, reinforcing the empire’s claim to eternal rule. The Immortals were meticulously organized and equipped. Their discipline and loyalty made them a symbol of imperial permanence, and their presence on the battlefield was as psychological as physical.

The Immortals fought in some of history’s most famous clashes. They were present both at Thermopylae (480 BCE) and at Plataea (479 BCE) but even more prominently in earlier campaigns under Cyrus the Great, including the conquests of Lydia and Babylon, where their shock tactics helped secure Persian dominance across vast territories. More than warriors, the Immortals were a political instrument. Over time, they became a status symbol among Persian nobility. In art and architecture, such as the reliefs at Persepolis, the Immortals are immortalized as icons of disciplined power, a reminder that in ancient warfare, perception was as potent as steel. As war evolves, the modern equivalent to the Immortals could be a fully autonomous fighting unit which never loses its strength (by always replacing destroyed equipment with new) and where all units always have cutting edge properties as their digital platform enables continuous learning and improvement. Machines don’t sleep, don’t tire, and don’t complain. In modern warfare, they are the new Immortals.

The evolution of war, not the least as seen in Ukraine, goes towards uncrewed battlefields with machines fighting machines with no, or very few, human soldiers in harm’s way. Machines replace humans at the frontline. Counting all maintenance personnel and remote operators, an army of machines does not reduce the total manpower by much but saves lives by putting people at safer distance from the front. Below we make some napkin-level math to estimate how expensive it would be to create autonomous and remote-controlled fighting units, from replacing a pair of soldiers up to a brigade. (The napkin-level math we present here concerns only the hardware costs, it ignores training costs, production facilities, software and AI development, and so on.) We ask ourselves how much it would cost to create the world’s first autonomous battalion as a means of deterrence. Backed with production capacity to maintain it at full strength and operating with a truly “military machine learning” capacity, such a battalion would become immortal and return smarter and more effective after every engagement. An immortal and endlessly improving battalion would certainly be useful for Ukraine and for us as it would be an effective means of deterrence.

Ask any engineer if it would be technically possible to create autonomous fighting units and they will always reply with a yes. They will then typically start talking about hardware and conclude that it is well within humanity’s technical capability to build anything needed. After concluding on hardware they will describe software challenges, noting that there are unsolved pieces, but then write them off as solvable with self-learning AI-based algorithms and systems. Any number of similar interviews leads to the conclusion that if we make a decision today, we could have the unmanned battlefield in three to five years.

Since technology is not a problem in the eyes of the (optimistic) engineers, let us have a look at funding and financing, which always is at the core of major investment decisions. In particular, let us estimate the initial hardware costs. The table below presents a very rough estimate of the numbers of UGVs and UAVs it would take to perform the field tasks of units of various sizes, and some cost estimates. There are hundreds of assumptions made that could and should be argued for, but the table gives some sort of estimate. Our estimates are based on experiences from a portfolio of ten civilian-military projects funded by three Swedish funding agencies, Vinnova, Energimyndigheten and Trafikverket, augmented by material from two NATO Exploratory Teams and one NATO Research Task Group.

Our analysis indicates that replacing one frontline soldier with an autonomous equivalent would cost between 100.000 Euro and 200.000 Euro. The laggards adapting to an unmanned battlefield would probably pay around 100.000 Euro per soldier replaced, the early adopters around say 200.000, while it may cost 500.000 Euro per soldier to be the innovator. Some of the costs could be recuperated through export of hardware, licensing of technologies, and reimbursements based on sinking production costs. In practice, as was observed already from the production of the Liberty class vessels during WWII, the learning curve and hence cost structure improves as technology spreads, production processes improve and production volumes increase.

Such a battalion would at first be a mix with mostly remote-controlled units, gradually expanding to autonomous entities. It would at first also be specialized for a single purpose, like maintaining a frontline in a given terrain type (which already can be observed in Ukraine), to evolving into different roles like becoming shock troops. By continually updating software and AI-modes, and adapting the hardware configuration, the Immortals would become more versatile based in their ability to constantly learn and improve. Because unmanned ground vehicles (UGVs) primarily function as carriers, specialization is achieved through the payloads mounted on the platform (e.g., flatbeds, container mounts, weapon stations). A modular approach would enable a single vehicle to support multiple operational roles, such as logistics and direct-combat support.

Humans are versatile while machines are specialized, a fact that shapes assumptions about the battlefield of the present and the future. In general, machines excel in domains of relative certainty: they execute well-defined tasks with precision, speed, and tireless repetition, typically within the boundaries set by their design, training, and operational constraints. In controlled or highly structured environments, where rules are stable and variables are limited, machines can surpass human performance decisively. Yet the fog of war is at times opaque. The battlefield is fluid, adversarial, and saturated with uncertainty. Plans fracture, threats emerge unexpectedly, and decision-makers confront moral and political dilemmas that resist formalization. Contemporary AI systems, particularly those capable of learning from data, fusing multiple sensor inputs, and adapting operating procedures over time, can display forms of situational responsiveness that, while not equivalent to human intuition and understanding, mimic human behavior in complex situations. Even so, such adaptations remain constrained by training regimes, data availability, and the difficulty of encoding or reliably learning normative judgments under pressure. When we, out of convenience, refer to the uncrewed battlefield, this signifies that there are no or few soldiers at the front. Any “autonomous” fighting unit will be a mix of remote controlled and autonomous entities, where ‘remote control’ could be anything from piloting a drone though goggles to giving high-level commands to a group of vehicles. Early versions of autonomous fighting units would have a high degree of remote control and supervision, with more autonomy being introduced over time. It is important to understand that a fighting unit with both soldiers and machines will be a more effective combatant than a unit with solely autonomous and remote-controlled machines. Introducing uncrewed units is not about being the best on the battlefield, it is about avoiding the loss of life. Uncrewed units also offer a wider range of tactical options and can be used more aggressively since any losses can be quickly replaced. This has significant implications for the commander’s decision processes and operational concepts. A decisive skill of the commander would be the ability to use the data gained to improve the unit impact and quickly adapt its operational modes almost in real-time. It should also be noted that replacing frontline manned units assumes that autonomous fighting units can be integrated in Multi Domain Operations (MDO), working in tandem both with long-range artillery and air support.

The Nordic countries face the same structural challenge visible in Ukraine: shortage of soldiers willing to serve on a frontline where the risk of death remains high. Building the world’s first autonomous Immortal Battalion would allow the Nordics to maintain a constant force level and avoid the political vulnerability of human casualties. The cost analysis demonstrates that replacing a soldier with an autonomous unit is expensive for the innovators but later significantly cheaper (please observe that we do not calculate the net cost of replacement, since we do not include the costs of educating and supporting neither soldier nor operator and mechanic). Being the innovator offers strategic and industrial benefits: domestic industries gain competence, production capacity grows, and deterrence strengthens as a credible, always‑ready force takes shape. With enough production capacity to maintain it at full strength, a Nordic Immortal Battalion becomes not just tactically useful but strategically necessary, an insurance policy in a region where geography and limited population make manpower‑heavy defense increasingly unrealistic.

The Nordic region is uniquely positioned to field such a force. Economically, all six Nordic nations rank among the top tier globally, consistently placing within the world’s 20-30 highest‑income countries. Their average GDP per capita, roughly 70,000 USD, sits comfortably above the broader high‑income benchmark. Innovation follows the same pattern. According to the Global Innovation Index, the Nordic countries once again cluster within the global top 20-30, with Sweden standing out as the world’s second most innovative nation. Together, the region forms a dense ecosystem of research universities, strong public funding mechanisms, and a well-developed innovation infrastructure. What the Nordics do lack is large‑scale production capacity in this domain, that gap is precisely what the creation of a Nordic Immortal Battalion would close. Building such a unit would not only meet a strategic need but also catalyze the industrial capability required to sustain it, aligning economic strength with technological ambition.

A modern autonomous battalion revives the same principle as the ancient Immortals, but through machines instead of manpower. Every destroyed vehicle or drone is immediately replaced, and each new unit inherits the accumulated experience of all previous ones. After every engagement, the battalion returns not weakened, but improved. Backed by sufficient production capacity, such a force becomes functionally immortal, never depleted, never exhausted, never forgetting. The same idea that once defined an ancient elite unit becomes a template for the future battlefield: a constant-strength battalion whose temporary losses mean little and whose presence signals resilience. The Immortals embodied imperial continuity; their modern counterpart would embody technological deterrence, shaped not by myth but by engineering. The modern Immortals are immortal because they never were alive.

Per Ödling, Professor at Lund University

Daniel Ekwall, Professor at University of Borås and Swedish Defense University

Per Skoglund, Lieutenant Colonel and senior lecturer at Swedish Defense University

Tore Listou, Associate Professor at Norwegian Defence University College