The debate you are reading in briefings and think pieces is simple on its face and messy in reality. Cheap, massed robotic systems promise to overwhelm defences and reduce risk to soldiers. Traditional artillery offers predictable lethality, logistics chains commanders understand, and economies of scale honed over a century. If you want a serious cost comparison you cannot just trade unit price for headline glamour. You must break total cost into acquisition, per-shot/attrition cost, sustainment, and the operational context that translates shots into effects.
Baseline arithmetic: traditional artillery
Start with the projectile. A modern unguided 155 mm round has historically been an inexpensive commodity compared with guided munitions. Converting legacy stocks into guided effects with a precision guidance kit has been pursued precisely because it reduces the need for very expensive bespoke rounds. Guidance kits such as the M1156 PGK were fielded as a cost effective middle ground and were priced in the low tens of thousands of dollars per kit when produced at scale, which effectively turns a cheap 155 mm body into a much more accurate round without paying six figures per shot. This matters because it underwrites a core artillery advantage. The unitary cost for a premium guided 155 mm like the Excalibur has been reported in defense budgets and acquisition documentation in the tens to low hundreds of thousands of dollars per round depending on configuration and accounting year. That premium buys precision and minimizes the number of rounds needed for point targets but it sharply changes the calculus when large volumes of fires are required.
You also have launcher and platform costs. A towed howitzer or a tracked battery is a relatively low capital cost compared with sophisticated aircraft, and rocket artillery such as HIMARS or MLRS is more expensive per launcher but provides long range and flexible mission profiles. Those launchers factor into per-effect cost when you amortize platform price across shots, but for most contested campaigns the dominant recurring cost is ammunition and the supply chain to get it to forward units. Commodity shells scale well with surge production when industrial capacity exists.
Baseline arithmetic: robotic swarms
What do we mean by a swarm for this comparison? I use the practical sense: many inexpensive, largely attritable unmanned systems operating together to achieve effects that previously required a smaller number of expensive long-range weapons. The research and experiments that underpin the concept are older than the hype cycle. DARPA and other programs have pursued low-cost tube-launched or mass-launch concepts precisely because smaller per-unit price makes large numbers feasible. RAND and follow-on studies have sketched architectures where many low-cost sensor and strike nodes feed higher-end effectors. The raw point here is simple: if you can field hundreds or thousands of low-cost airframes, the per-unit acquisition price can be orders of magnitude below a guided artillery round.
But numbers matter. Small tactical loitering munitions and single-use strike UAVs have widely varying prices depending on sensors, range, and industrial source. Defense contracts and public procurement in 2022 show that fielded tactical loitering systems can cost from tens of thousands to, in some program accounting, significantly more once launchers, training, and logistics are included. For example, production contracts for tactical loitering munitions in 2022 put program-level prices that translate into tens of thousands of dollars per all-up round in many cases. At the extreme lower end, commercial off the shelf quadcopters used as improvised munitions cost hundreds to a few thousand dollars apiece but lack integrated warheads, hardened comms, and military-grade navigation. The key observation is that a swarm composed of inexpensive airframes looks cheap on the purchase order but incurs additional costs when hardened communications, electronic-warfare mitigation, and logistics are added.
Per-shot and operational costs: a fair comparison
Do not confuse unit price with cost to achieve effect. With Excalibur or other guided artillery, buyers accept a high per-round price because a single guided round can replace many unguided rounds and because it reduces collateral damage and re-attack cycles. The cost per effect is scenario dependent. If your target set is point targets well suited to precision fire, the guided artillery option is efficient despite high sticker price. If you need area suppression, massed fires and cheap rounds remain unrivalled.
Swarms change that calculus by changing the distribution of risk and redundancy. A swarm can mass many small warheads, saturate air defenses, or act as a distributed sensor net to cue higher-value weapons. However you measure it, attrition is expected. If a mission requires sending hundreds of robots against layered air defenses that can shoot or jam them, the per-effect cost can climb quickly. That is where hidden costs show up: replacement manufacture, batteries and spare parts, launcher and reloading logistics, trained operators, mission planning software, and most importantly the countermeasures you must buy or field to make swarms survivable in contested electromagnetic environments. Historical and contemporary studies on swarm employment stress that counter-UAS expenditures and layered defenses are themselves nontrivial.
Sustainment, scale, and industrial base implications
Artillery has a mature industrial base that can produce simple rounds in large volumes when ordered and when industrial capacity is prioritized. That produces predictable unit costs at scale for unguided rounds and for PGK conversions. Guided heavy munitions and precision rockets are bottlenecked by more complex supply chains. Robotic swarms trade that bottleneck risk for a different one. Cheap airframe components are broadly available from commercial markets, but military-grade electronics, hardened radios, and secure supply chains for warheads and propulsion become the limiting factor when you try to scale beyond demonstration numbers.
From a procurement and budgeting view the choice is rarely either or. Swarms promise the ability to generate mass at lower per-platform cost but at the price of new supply chains, high churn rates, and training and doctrine changes. Traditional artillery requires heavy investment in production capacity for quantity munitions and modern guided rounds if precision is needed. The two approaches therefore stress different parts of the industrial base and different budget lines.
Vulnerabilities that tip the math
There are three recurring vulnerabilities that often get glossed over in raw cost-per-unit arguments.
1) Electronic warfare and jamming. Precision guided rounds and many robotic systems rely on GNSS and data links. A contested EW environment degrades both, increasing miss rates and therefore the number of weapons you must expend to achieve an effect. Loss of GPS or degraded comms multiplies cost-per-effect for both swarms and guided artillery, but cheap swarms built around commercial components tend to be more fragile unless engineered for hardness.
2) Countermeasures and defense costs. Large-scale drone attacks force investment in C-UAS which can be costly per defended installation. The logic of attritable swarms assumes you can accept losses. That only works if the attacker values the effect more than the defender values the cost to stop it. In some contexts the defender will force the attacker into diminishing returns and drive up the attacker cost. Conversely massed artillery fires can also be countered by mobility, counter-battery, and air interdiction. Both sides of the ledger are active and expensive.
3) Logistics and operational tempo. Artillery logistics are well known and scale linearly. Swarms require rapid reconstitution and potentially distributed manufacturing or forward rearming. Those logistics lines are not free and they are operationally sensitive; high-tempo swarm operations create a recurrent bill that procurement accountants should not ignore.
So which is cheaper?
Simple scenarios exist where one clearly dominates. If you need a single precise strike on a hardened high-value target in permissive airspace a single guided artillery round or a long-range precision rocket will likely be cheaper and more reliable than sending dozens of small drones. If your objective is distributed sensors to locate fleeting targets across a wide area or to saturate an area to force the defender to disperse and reveal, swarms can be cost-effective relative to launching high-value guided munitions. In attrition-heavy environments swarms lose their price advantage fast unless they are engineered and budgeted as attritable systems with cheap, redundant production and hardened comms.
My practical bottom line
Do not accept the mantra that swarms are categorically cheaper. They can be cheaper per platform. They are not automatically cheaper per effect. If your accounting stops at procurement price you will mislead commanders. Build a cost-per-effect model that includes expected attrition rates, reconstitution time and cost, countermeasure expenditures, and the value of precision. In many realistic operational profiles in 2023 the answer will be a mixed force. Use artillery for mass fires and precision guided rounds for point kills when you need single-shot certainty. Use swarms where massed, distributed, or persistent effects provide operational advantage and where you have a plan to accept and finance attrition.
The military practicalities matter more than the hype. If you are buying for contingency, fund both the comfortable, industrially backed artillery lines and a modest, rigorously tested swarm capability that is measured by cost per effect not by flashy demos.
Acknowledgements and sources
I relied on open acquisition and program documents, program press releases, and established research on swarming concepts and their tradeoffs. The numbers cited for Excalibur, PGK and tactical loitering munitions reflect publicly available budget and contract material and reporting through 2022. The broader discussion of swarm doctrine and experimental programs draws on RAND and DARPA program descriptions that predate 2023. The key point remains pragmatic and simple: buy the effect you need and measure cost in outcomes not in unit sticker price.