Everything an AI Knows About Drone Warfare

Pellegrin Foiskola
Drones are rewriting the economics of warfare. Everyone involved knew this was coming. March 2026.

The United States just used a $35,000 drone to blow up targets in Iran.1 The drone was reverse-engineered from an Iranian design. Iran, meanwhile, used the original version of the same drone to kill six Americans in Kuwait.2 Both sides are, in the most literal sense possible, hitting each other with the same weapon.

The word “drone” now refers to so many different machines that using it in a sentence is about as precise as saying “vehicle.” A Predator flying at 25,000 feet for 24 hours and a $400 racing quadcopter duct-taped to a hand grenade are both “drones.” They share a category the way a cruise ship and a kayak both share the category “boat.”

The Council on Foreign Relations, in a useful piece of taxonomic frustration published last week, broke it down into five categories.3 There are tactical surveillance drones — small quadcopters, basically flying cameras. There are medium-altitude long-endurance platforms, the MALE drones, which is the category that includes Turkey’s famous Bayraktar TB2 and the American Reaper. There are one-way attack drones, also called kamikaze drones or loitering munitions, which is the category that includes Iran’s Shahed-136 and its American clone, the LUCAS. There are first-person-view drones, the FPV models that Ukraine has turned into the deadliest infantry weapon since the machine gun. And there are collaborative combat aircraft — autonomous or semi-autonomous jets designed to fly alongside manned fighters as “loyal wingmen.”

These are five different weapons with five different purposes at five different price points, and lumping them together would be like writing a single essay about “guns” that covers both a hunting rifle and a naval cannon. But the defense discourse does this constantly, because the word “drone” is exciting and the distinctions require homework. I will try to do the homework.

The place to start is how old the idea actually is. In 1917, the U.S. Army asked an inventor named Charles Kettering to build an unmanned flying bomb that could hit a target 40 miles away. Kettering — who had previously invented the electric starter for automobiles — assembled a team that included Orville Wright and built the Kettering Bug: a 600-pound biplane with a Ford engine (the engine alone cost $50), a pneumatic control system cannibalized from Kettering’s personal player piano, and 300 pounds of explosives. Total cost per unit: $400. The government ordered 20,000. The war ended before any were used.4

The idea of the expendable flying weapon is 108 years old.

The Bug’s descendants went in two directions. One became the cruise missile — the German V-1 “buzz bomb” of 1944, which terrorized London at a rate of over 100 per day and is generally considered the first combat drone, or at minimum the first operational cruise missile. The V-1 was powered by a pulsejet engine, navigated by gyrocompass, and told how far to fly by a mechanical odometer that counted down to zero. Nearly 10,000 were launched at British targets.5 The other direction became the surveillance drone — radio-controlled target aircraft used for anti-aircraft gunnery practice in the 1930s, including the British DH.82B Queen Bee, from which the term “drone” is believed to derive.6

For decades after World War II, drones remained either target practice or reconnaissance tools. The U.S. flew surveillance drones over Vietnam and China. Israel developed small tactical drones in the 1970s and 1980s that became the basis for a globally dominant export industry — by 2014, Israel accounted for over 60 percent of global drone exports.7 But the attack drone, the lineage that leads to the Shahed and the LUCAS, took a different path.

It took a German path.

In the mid-1980s, West Germany and the United States jointly developed a weapon called the Drohne Anti-Radar, or DAR, manufactured by Dornier. It was a single-use kamikaze drone designed to hunt Soviet radar installations — a delta-winged flying bomb weighing 110 kilograms, launched from a truck, navigating by GPS, carrying a fragmentation warhead. Range: approximately 600 kilometers. The Cold War ended. The project was cancelled. The design details were sold to Israel, which developed them into the IAI Harpy anti-radar drone. Iran, through channels that remain disputed, appears to have drawn on similar concepts for the Shahed family. The Shahed-136’s engine is a reverse-engineered German Limbach L550E. Its silhouette is strikingly similar to the DAR’s.8

The LUCAS drone the U.S. is firing at Iran in March 2026 is, at the level of design lineage, a great-grandchild of a 1980s West German anti-Soviet weapon, filtered through Israeli engineering and Iranian manufacturing.

What made drones explode in the 2010s and 2020s was not a single breakthrough but the convergence of several civilian technologies reaching maturity simultaneously. GPS, originally a military-only system with intentionally degraded civilian signals, was opened to full civilian accuracy in 2000. Smartphone-driven investment drove the cost of gyroscopes, accelerometers, cameras, and GPS receivers to a fraction of their military-grade equivalents. Lithium polymer batteries became cheap and energy-dense enough to power small electric motors for meaningful durations. Commercial drone companies, led by China’s DJI, created a mass market that subsidized the components militaries needed.

The result was that by roughly 2015, the ingredients for a functional military drone — GPS navigation, stabilization, camera, communications, airframe — could be sourced from consumer electronics catalogs for a few hundred dollars. The Bayraktar TB2 uses a Garmin navigation radio. Garmin released a statement noting that this product “is not designed or intended for military use.”9 It is there anyway. The first armed drone strike in the modern sense was in Afghanistan in 2001. By 2022, Ukraine was fighting the first full-scale drone war. By 2026, both sides of a U.S.-Iran conflict are firing kamikaze drones at each other whose design lineage traces to a West German defense contractor that ceased to exist in 2002.

The technology was always possible. What changed was the price.

And the price is where the entire structure of modern warfare comes apart. A Patriot interceptor missile costs approximately $4 million. A THAAD interceptor costs $12 to $15 million.10 An Iranian Shahed-136 drone costs somewhere between $20,000 and $50,000 to produce. The Center for Strategic and International Studies estimates the figure at roughly $35,000.11

When Iran launches a Shahed at a target defended by a Patriot battery, the defender must fire a $4 million missile to destroy a $35,000 drone. The cost ratio is approximately 114 to 1 in the attacker’s favor. If the attacker sends fifty drones, the defender has spent $200 million to neutralize $1.75 million worth of hardware. The drone doesn’t even have to hit anything. It just has to be launched.

This is not a flaw in the Patriot system. The Patriot works exactly as designed. It tracks an incoming aerial threat, computes an intercept solution, and fires a missile that destroys the threat. The problem is that the Patriot was designed to kill things that cost more than the missile you kill them with — aircraft, ballistic missiles, cruise missiles. Nobody anticipated that the incoming threat would cost less than a used Honda Civic.

The system works. The math doesn’t.

One analyst at the Middle East Institute Switzerland estimated that for every dollar Iran spends producing drones, Gulf states spend $20 to $28 on defensive fire.12 Two AN/TPY-2 radars — the sensor backbone of the THAAD system, each costing upward of $1 billion — were recently disabled by Iranian drones costing roughly $30,000 apiece. That is a cost-exchange ratio of more than 30,000 to 1.13 The European Policy Centre characterized this in a recent analysis as a problem of “mathematical inevitability.”

The Pentagon’s emerging answer is to drive the cost of American drones to $5,000 per unit through its Drone Dominance Program,14 at which point the math inverts: a single Patriot interceptor costs 800 drones, a single THAAD interceptor costs 3,000, and the entire logic of shooting expensive missiles at cheap targets collapses into absurdity. But that production target is a projection, not a fact. The U.S. produced approximately 650 Patriot interceptors last year.15 Estimates suggest Iran produces on the order of 12,000 Shaheds per month.16

The weapon at the center of that production gap — the latest node on the genealogy that runs from Dornier’s factory to the Persian Gulf — is the Shahed-136. It is 3.5 meters long with a 2.5-meter wingspan. It weighs about 200 kilograms. It carries a warhead of 30 to 50 kilograms of explosives and flies at roughly 185 kilometers per hour on a two-bladed pusher propeller.17 It looks like something a talented engineering student might build in a well-funded garage, because that is more or less what it is. It has been described as “the poor man’s cruise missile,”18 and the description is apt — it does what a cruise missile does, minus almost everything that makes a cruise missile expensive.

It navigates by GPS and inertial guidance. It can be launched from a rack on the back of a truck. It does not require a runway, a trained pilot, or any infrastructure more complicated than a GPS signal and someone willing to press a button.

Iran has been producing these at staggering scale. A military source told Fars News Agency that Iran fired over 500 ballistic and naval missiles and almost 2,000 drones in the first week of the current war alone.19 Production estimates for Iran’s domestic capacity suggest roughly 400 Shahed-class drones per day — on the order of 12,000 per month, though the figures are contested.16 Russia, which licensed the design and produces it at the Alabuga Special Economic Zone in Tatarstan under the name Geran-2, is estimated to add 5,000 to 7,000 per month on top of that.20

The effectiveness rate for individual Shaheds is not impressive in isolation. One analysis estimated that between 11.5 and 18.7 percent of strikes hit their target in standard operations, rising to 40 to 50 percent during mass attacks.21 But effectiveness per drone is the wrong metric. When your weapon costs $35,000, you don’t need a high hit rate. You need volume. The math works even if most of them miss, because the ones that get through impose costs wildly disproportionate to the ones that don’t, and the ones that get intercepted still force the defender to burn through irreplaceable munitions.

Iran, in other words, has built a weapon whose primary function is not to destroy things but to be cheaper than the thing that destroys it. The actual explosions are a bonus.

The United States looked at this weapon, studied it, and built its own. In July 2025, Secretary of Defense Pete Hegseth signed a memo titled “Unleashing U.S. Military Drone Dominance.”22 In December 2025, the U.S. military unveiled the LUCAS — Low-cost Unmanned Combat Attack System — a drone reverse-engineered from a captured Shahed-136 and built by SpektreWorks, a defense contractor in Phoenix, Arizona.23 On February 28, 2026, LUCAS saw its first combat use against Iranian targets in Operation Epic Fury.1

The timeline from policy memo to combat deployment was eight months. The timeline from capturing an enemy weapon to firing your own version of it back at the enemy was approximately eighteen months. Admiral Brad Cooper, the CENTCOM commander, described the process with a concision that borders on poetry: “We captured it, pulled the guts out, sent it back to America, put a little ‘Made in America’ on it, brought it back here and we’re shooting it at the Iranians.”24 The Pentagon usually takes eighteen months to finalize a requirements document.

LUCAS costs $35,000 per unit — essentially the same price as the weapon it was copied from.25 A Tomahawk cruise missile, the previous workhorse of American long-range precision strike, costs approximately $2.5 million for the latest version. LUCAS does roughly the same job at 1.4 percent of the cost. It can loiter, confirm target identity, and strike when conditions are favorable. It has autonomous navigation, inertial backup systems for electronic warfare environments, and — the significant upgrade over the original Shahed — Starlink terminals that enable swarm coordination and dynamic targeting while keeping a human in the loop.26

An IRGC spokesperson responded by saying, “There is no greater honor than seeing self-proclaimed superpowers kneel before an Iranian drone and copy it.”21 He neglected to mention that the Shahed series was itself developed in part from a captured American RQ-170 surveillance drone that Iran forced down in 2011. The genealogy of this weapon is a closed loop of nations reverse-engineering each other’s designs, and the loop is tightening.

Not every branch of the drone family tree leads to the Shahed. Turkey’s Bayraktar TB2 represents the other evolutionary path — the reusable, precision drone rather than the expendable, mass-produced one. It is a MALE drone: it flies, it watches, it fires missiles, and then it comes home. The TB2 costs approximately $5 million per unit, a quarter the price of an American MQ-9 Reaper,27 and it has been exported to at least 34 countries. Turkey accounted for 65 percent of global armed UAV exports as of 2021.28

The TB2 became internationally famous in 2020 when Azerbaijan used it to devastating effect against Armenian forces in Nagorno-Karabakh.29 Ukraine used it early in the 2022 Russian invasion to strike Russian supply columns, generating a folk song in the drone’s honor. Baykar, the manufacturer — run by two brothers, one of whom is President Erdogan’s son-in-law — generated $1.8 billion in exports in 2023, with over 90 percent of revenue from international sales.30

The TB2’s proliferation story illustrates an irony worth dwelling on. The U.S. refused to sell Turkey armed drones in the 2010s, citing human rights concerns about their potential use against Kurdish groups.31 Turkey built its own. It then exported them to countries with human rights records that make Turkey’s look gentle. The U.S. export restriction did not prevent the proliferation of armed drones. It ensured that the proliferation would be led by someone else.

But the TB2 also revealed the limits of the reusable drone in high-intensity warfare. Against opponents with modern air defenses — Russia in Ukraine, for instance — it proved vulnerable. Slow, not stealthy, effective only against enemies who cannot shoot back from the sky. The lesson was that the future belonged not to expensive reusable drones doing precision work but to cheap expendable drones doing volume work. The Shahed, not the Bayraktar, turned out to be the template — and the genealogy continued accordingly.

Ukraine, which has spent three years on the receiving end of that genealogy — Russia’s Geran-2 is the Shahed by another name — responded by becoming the world’s largest drone manufacturer. The country now produces an estimated 4 million drones per year, more than all NATO countries combined.32 The National Security and Defense Council claims a production capacity of 8 million FPV drones annually as of January 2026.33 Over 160 companies produce FPV drones in Ukraine, ranging from large factories to garage operations. A single FPV drone costs between $300 and $500.34

Three hundred dollars. For a weapon that Ukraine’s defense council says is responsible for 60 percent of Russian army losses.33

Ukraine’s drone ecosystem is a study in wartime adaptation. Before the 2022 invasion, the country had roughly ten drone manufacturers. By 2024, it had several hundred.35 The production model is distributed — small teams can assemble drones from off-the-shelf components, shift locations to avoid Russian missile strikes, and iterate on designs in weeks rather than the years a traditional defense acquisition process requires. A French general estimated that 80 percent of targets in Ukraine are now destroyed by FPV drones.36 A brigade uses several hundred per month; actual needs are estimated at around 2,500.35

The insight is that Ukraine has effectively replaced artillery — which requires heavy industry, expensive shells, and long supply chains from Western allies — with something that can be built by a small team with a 3D printer and a soldering iron. This is not a planned doctrinal evolution. It is an emergent adaptation to the specific constraint of fighting a larger enemy while dependent on foreign ammunition deliveries that arrive on someone else’s political schedule. The drone didn’t replace artillery because it’s better. It replaced artillery because it’s there.

And now the expertise is traveling the loop in a new direction. Ukraine has also become the world’s leading laboratory for electronic warfare countermeasures, fiber-optic drone control that can’t be jammed, AI-guided autonomous targeting, and interceptor drones designed to shoot down other drones. Zelensky announced in March 2026 that Ukraine was sending military specialists to the Gulf to help counter Iranian Shaheds37 — the same Shaheds that Ukrainian soldiers have been fighting for three years. The design travels one direction around the loop. The knowledge of how to defeat it travels the other.

The weapons traveling this loop are increasingly being aimed by machines. The U.S. military used Anthropic’s Claude AI model in Operation Epic Fury to assess intelligence, identify targets, and simulate battle scenarios. According to the Wall Street Journal, Claude is central to Palantir’s Maven Smart System, which provides real-time targeting for military operations; the U.S. struck over 1,000 targets in the first 24 hours.38 The speed of target generation at that scale requires machine assistance — no human intelligence team can produce that volume of vetted target packages in that timeframe.

This is where the cost logic and the AI logic converge, and where the problem becomes something worse than an accounting headache. Cheap drones mean more targets to strike and more threats to assess. More targets and threats mean more data to process. More data to process means the humans cannot keep up. The machine keeps up. And so the machine, gradually and then suddenly, becomes the system — not because anyone decided it should be, but because the volume made any alternative operationally untenable.

Israel has been further down this road than anyone else for longer. The Gospel system, deployed in Gaza since at least 2023, generates building and infrastructure targets. Lavender generates lists of suspected militants. Lavender reportedly has a 90 percent accuracy rate,40 which sounds good until you remember that 10 percent of a list of tens of thousands of people is thousands of people who are not what the machine says they are. An analyst might spend 20 seconds reviewing a Lavender-generated target before approving a strike.41 Twenty seconds to decide whether an algorithm has correctly identified a human being as someone who should be killed.

Good enough is doing a lot of work in that sentence.

The structural problem with AI targeting is not that the technology is bad. The technology is, by most accounts, impressive. The problem is that the technology is good enough to be trusted and not good enough to be trustworthy. This is the gap between design and experience rendered in its most consequential form. A system that is right 90 percent of the time will, when applied to a population of 40,000 flagged individuals, generate 4,000 errors. Each error is a person. The system does not know this. The system does not know anything. The system produces outputs, and humans — under time pressure, in a war, with institutional incentives to clear the queue — convert those outputs into ordnance.

The deeper problem is that AI targeting compresses the kill chain — the sequence from identification to intelligence validation to legal clearance to weapons release — into a loop so tight that human oversight becomes less a check on the system than a bottleneck the system is designed to route around. The commander of Israel’s Unit 8200 described human personnel as a “bottleneck” that limits the army’s capacity during a military operation.41 The word is revealing. A bottleneck is something you widen. When the bottleneck is the human being who decides whether a target is a combatant or a father walking home from work, widening it means spending less time on the question.

In simulated war games, AI models from multiple companies chose nuclear escalation in 95 percent of scenarios.42 The simulations don’t predict real-world behavior. But they reveal something about how optimization algorithms reason under pressure: they find the solution that satisfies the objective function, and the objective function does not include the concept of restraint. Restraint is a human contribution. The question is whether the humans remain in a position to contribute it when the machine is generating targets faster than any human can review them.

The obvious solution to cheap drones is cheap drone defense. The defense establishment’s instinct, naturally, has been to pursue expensive drone defense instead. The U.S. Army has spent years prototyping directed-energy weapons — lasers that theoretically cost pennies per shot, high-power microwaves that can fry a swarm’s electronics at once. The Stryker-mounted 50-kilowatt laser. The Navy’s ODIN dazzler.43 The Air Force’s THOR microwave system. These are sophisticated programs developed by large defense contractors at significant cost, and they represent the institutional reflex of a military-industrial complex whose answer to any problem is a program of record with a prime contractor and a Congressional district.

The problem is that a $35,000 drone is being addressed with solutions that require clean-room maintenance, fail when optics get dusty, and cannot yet survive the combination of desert heat and cluttered airspace that defines every theater where drones are actually being used. After deploying four Stryker-mounted laser prototypes to the Middle East for soldier testing, the Army reported that initial feedback was “not overwhelmingly positive,”44 a phrase that puts my own dry sense of understatement to shame.

Ukraine, facing the same problem with less money and more urgency, built a network of 10,000 microphones that track drones by their acoustic signature and compute flight paths in real time. The system was originally developed by two engineers in a garage. Each microphone costs around $500.45 It does not shoot anything down. It tells you where the drone is, how fast it’s moving, and where it’s going, so that something cheaper than a Patriot missile — a machine gun, a jamming system, a fighter’s 30mm cannon — can do the rest. It is an ear, not a fist, and it works in dust.

The offense is cheap, scalable, and here. The defense is expensive, experimental, and coming. The gap between the two is measured in years, and wars are measured in days.

The roster of nations in this race grows by the quarter. The United States, Turkey, Iran, Israel, China, Ukraine, Russia, and now France, which just announced a Shahed-class drone to be manufactured by Renault on automotive production lines at 600 units per month.46 Renault. The car company. Making kamikaze drones. China’s DJI dominates the commercial drone market from a U.S. blacklist while its products are used by militaries on every continent. Ukraine produces 4 million drones a year across 160-plus manufacturers and is rapidly becoming an exporter.32 Russia mass-produces licensed Shahed copies at Alabuga alongside its own Lancet loitering munition.

Which brings us to the question that the cost numbers and production numbers and genealogy charts all eventually arrive at: what does this mean for the people underneath these things when they land.

The honest answer depends on which drone you mean, but the dishonest framing is the one that treats the question as settled. It is not. But the problem comes first, because it is the one that the precision advocates do not want to talk about.

A 2016 analysis found that drone strikes in non-battlefield settings — Pakistan, Yemen, Somalia — resulted in roughly 35 times more civilian fatalities per strike than manned airstrikes in conventional battlefields like Iraq and Afghanistan.48 The drones are not less accurate. They use the same targeting systems as manned aircraft. The difference is in how they are used.

Because drones are cheap and risk no pilot, they get used in situations where manned aircraft would never be sent. The bar for launching a strike drops. The aperture of acceptable targets widens. Signature strikes — killing people based on behavioral patterns rather than confirmed identity — become possible precisely because you have a drone overhead 24 hours a day watching people move. The technology enables patience, but the institution uses the patience to generate more targets, not fewer errors.

The weapon is designed for precision. The system that deploys the weapon is designed for throughput.

Precision advocates will point out, fairly, that a Predator or Reaper can loiter over a target for hours, confirm identity, wait for civilians to leave, and strike with a relatively small munition. Compared to carpet bombing, or a ground invasion with artillery, or a B-52 dropping unguided ordnance, the precision drone is obviously less destructive. Former Defense Secretary Robert Gates argued in 2013 that you can more easily limit collateral damage with a drone than with any other weapon, including precision-guided bombs from manned aircraft.47 This is true as far as it goes. It does not go as far as the 35x differential suggests it should.

The new generation of cheap kamikaze drones — the Shaheds, the FPVs, the LUCAS — presents a different civilian problem entirely. These are not precision weapons in any meaningful sense. A Shahed-136 navigating by GPS to a set of coordinates does not loiter, does not confirm identity, and does not distinguish between a military radar installation and the elementary school next door. In the current Iran war, 180 civilians were killed in a single strike on a school in Minab.2 The UAE reported that out of 941 Iranian drones detected since the war began, 65 fell within its territory, damaging ports, airports, hotels, and data centers.49 In Bahrain, 32 civilians including children were injured in a drone attack on the town of Sitra.50

The mass-production drone does not make war more precise. It makes war cheaper. These are different things, and confusing them is how you end up with a discourse in which the word “drone” simultaneously means “the most precise weapon in our arsenal” and “a GPS-guided lawn mower engine that just hit a school.”

Everyone in this system is behaving rationally. Iran builds cheap drones because they impose disproportionate costs. The U.S. intercepts them with expensive missiles because the protected assets are worth more than the missiles. The U.S. also builds its own cheap drones because offensive cost-imposition is cheaper than defensive interception. The defense industry develops lasers because the interceptor math is unsustainable. The lasers don’t work reliably yet because the physics of focusing a beam through dust and humidity turns out to be harder than the physics of strapping a GPS to a lawn mower engine.

Nobody designed this system. Every participant is responding logically to the incentives created by every other participant’s previous logical response. The aggregate result is a world where a car company in France is manufacturing kamikaze drones, two engineers in a Ukrainian garage have built a nationally significant air defense network out of microphones, and the United States is firing copies of Iranian weapons at Iran while Iran fires the originals at American bases defended by missiles that cost more than most houses.

The blueprint said precision would replace mass. The reality said mass is back, and it brought a price tag that makes precision look like a luxury. The drone war is not a revolution in military affairs. It is a reversion to the oldest principle in warfare — that quantity has a quality all its own — executed at a price point that puts it within reach of anyone with a truck, a GPS module, and a grievance.

The cost of entry just dropped below the cost of defense.

  1. Military Times, “US confirms first combat use of LUCAS one-way attack drone in Iran strikes,” Feb. 28, 2026. Link ↵
  2. Wikipedia, “2026 Iran war,” citing U.S. Department of Defense casualty announcements and Iranian civilian casualty figures including 180 killed in the Minab school airstrike. Link ↵
  3. Michael C. Horowitz and Lauren Kahn, “The New Era of Drone Warfare Takes Root in Iran,” Council on Foreign Relations, Mar. 9, 2026. Link ↵
  4. War History Online, “The Kettering Bug: America’s First Foray Into Unmanned Drone Technology Started During WWI.” Link; History Collection, “The Story of the Kettering Bug, the World’s First Aerial Drone.” Link ↵
  5. Robin Radar Systems, “Evolution of the Drone Threat: Part 1,” Jul. 2024. Link ↵
  6. Imperial War Museums, “A Brief History of Drones.” Link ↵
  7. Wikipedia, “Unmanned aerial vehicle,” citing SIPRI data on Israeli drone exports 2010–2014. Link ↵
  8. Defense Express, “First Shahed-136 Prototype was Created in Germany in the 1980s, and It was Called DAR.” Link; Wikipedia, “Dornier DAR.” Link; sUAS News, “Die Drohne Antiradar — the original Shahed drone,” Jul. 2025. Link ↵
  9. MIT Technology Review, “Mass-market military drones have changed the way wars are fought,” Jan. 2023. Link ↵
  10. European Policy Centre, “The new economics of warfare,” 2026, citing Patriot at ~$4 million and THAAD at $12–$15 million per interceptor. Link ↵
  11. CNBC, “Iran’s Shahed-136 drone: How ‘the poor man’s cruise missile’ is shaping Tehran’s retaliation,” Mar. 5, 2026, citing CSIS analyst Patrycja Bazylczyk. Link ↵
  12. Defense News, “Iran can still fire drones and missiles — experts weigh the implications on the war,” Mar. 6, 2026, quoting Francesco Schiavi, Middle East Institute Switzerland. Link ↵
  13. European Policy Centre, “The new economics of warfare,” 2026, on AN/TPY-2 radars disabled by drones and the 30,000-to-1 cost ratio. Link ↵
  14. Drone Warfare, “LUCAS: America’s $35,000 Expendable Strike Drone,” citing the Drone Dominance Program at $1.1 billion through the “One Big Beautiful Bill Act of 2025” and Phase 1 targeting 30,000 drones at $5,000 per unit. Link ↵
  15. Norsk Luftvern, “The Drone Defense Economics Crisis: When $3M Missiles Target $38K Drones,” Jun. 2025, citing global Patriot production at ~650 units annually. Link ↵
  16. Al Habtoor Research Centre, “The Missile and Drone Dilemma,” Mar. 2026, estimating Iranian production at ~400 Shahed-class drones per day (~12,000/month) and Russian Alabuga production at 5,000–6,890/month. Link ↵
  17. Wikipedia, “HESA Shahed 136,” citing dimensions of 3.5 m length, 2.5 m wingspan, ~200 kg weight, 30–50 kg warhead, 185+ km/h speed. Link ↵
  18. CNBC, “Iran’s Shahed-136 drone,” Mar. 5, 2026, using the phrase “the poor man’s cruise missile.” Link ↵
  19. Wikipedia, “2026 Iran war,” citing Fars News Agency military source on 500+ missiles and ~2,000 drones as of Mar. 5, 2026. Link ↵
  20. GlobalSecurity.org, “LUCAS — Low-Cost Uncrewed Combat Attack System,” citing Russian Geran-2 production at Alabuga SEZ at $30,000–$80,000 per unit. Link ↵
  21. SOAA, “The Rise of America’s LUCAS Drone: When Imitation Becomes Advantage,” Feb. 2026, citing Shahed effectiveness at 11.5–18.7% in standard operations, 40–50% in mass attacks, and quoting IRGC spokesperson Abolfazl Shekarchi. Link ↵
  22. Drone Warfare, “LUCAS,” citing Hegseth memo “Unleashing U.S. Military Drone Dominance,” Jul. 10, 2025, under Executive Order 14307. Link ↵
  23. Wikipedia, “Low-cost Uncrewed Combat Attack System,” on SpektreWorks as developer and Dec. 2025 deployment. Link ↵
  24. The War Zone, “LUCAS Kamikaze Drones Lauded As ‘Indispensable’ By U.S. Admiral In Charge Of Iran War,” Mar. 6, 2026, quoting Adm. Brad Cooper. Link ↵
  25. CENTCOM spokesperson Capt. Tim Hawkins, quoted in The War Zone, Dec. 2025, confirming $35,000 per-unit cost. Link ↵
  26. The War Zone, “U.S. Military Has Used Long-Range Kamikaze Drones In Combat For The First Time,” Feb. 2026, on Starlink terminals and swarm coordination capabilities. Link ↵
  27. Grey Dynamics, “Bayraktar TB-2: Turkey’s Rise to Drone Superpower,” Nov. 2025, citing TB2 at ~$5 million vs. MQ-9 Reaper at ~$20 million. Link ↵
  28. TRT World, “Turkish drone Bayraktar TB2 reaches milestone with 1 million flight hours,” Dec. 2024, citing CNAS report on 65% of global UAV exports. Link ↵
  29. Wikipedia, “Baykar Bayraktar TB2,” on Nagorno-Karabakh deployment and combat record. Link ↵
  30. TRT World, Dec. 2024, citing Baykar’s $1.8 billion in 2023 exports with 90%+ from international sales. Link ↵
  31. ProPublica, “Bayraktar TB2 Drone Sales from Turkey Growing Despite Western Laws,” Jul. 2022, on U.S. export refusal and subsequent Turkish indigenous development. Link ↵
  32. Bloomberg, Nov. 11, 2025, reporting Ukraine producing ~4 million drones per year; Army Recognition, “Ukraine Emerges as World Leader in Drone Technology,” Nov. 2025. Link ↵
  33. National Security and Defense Council of Ukraine, “Results of Ukraine’s Defense Industry in 2025: FPV Drones,” Jan. 23, 2026, claiming 8 million FPV annual capacity and 60% of Russian losses attributable to FPV drones. Link ↵
  34. Kyiv Post, “Ukraine Drone Production Tops 2.5 Million a Year,” Feb. 2025, citing $300–$500 per FPV drone. Link ↵
  35. OSW Centre for Eastern Studies, “Game of drones: the production and use of Ukrainian battlefield unmanned aerial vehicles,” Oct. 2025, on pre-war vs. wartime manufacturer counts and brigade-level drone consumption. Link ↵
  36. OSW Centre for Eastern Studies, Oct. 2025, citing French general’s estimate that FPV drones destroy 80% of targets. Link ↵
  37. The Hill, “America’s military has an Iranian drone problem,” Mar. 7, 2026, reporting Zelensky’s announcement of Ukrainian specialists deploying to the Gulf. Link ↵
  38. Responsible Statecraft, “US used ‘Claude’ to strike over 1000 targets in first 24 hours of war,” Mar. 2026, citing Claude’s role in Palantir’s Maven Smart System. Link; Washington Post, “Anthropic’s AI tool Claude central to U.S. campaign in Iran,” Mar. 4, 2026, on 1,000+ targets in first 24 hours. Link ↵
  39. [Consolidated into fn 38.] ↵
  40. Wikipedia, “Artificial intelligence arms race,” citing Lavender’s 90% accuracy rate per Unit 8200 reporting. Link ↵
  41. Responsible Statecraft, “Israel using secret AI tech to target Palestinians,” Apr. 2024, on 20-second review times for Lavender-generated targets. Link; TIME, “Israel’s Use of AI in Gaza May Be Setting a New Warfare Norm,” Dec. 2024. Link; TechPolicy.Press on Unit 8200 commander describing human personnel as a “bottleneck.” Link ↵
  42. Rest of World, “In Iran war, AI and drones are outpacing global rules of war,” Mar. 2026, citing study where AI models chose nuclear weapons in 95% of simulated war-game scenarios. Link ↵
  43. Army Recognition, “U.S. Navy Destroyer Deploys ODIN Laser Directed-Energy Weapon System During Operation Epic Fury,” Feb. 2026. Link ↵
  44. Army Recognition, “U.S. Army Accelerates Laser and Microwave Weapons to Defeat Mass Drone Attacks,” Mar. 2026, on DE M-SHORAD prototypes and “not overwhelmingly positive” feedback. Link; Defense News, “Army readies to launch 2026 competition for counter-drone laser weapon,” Aug. 2025. Link ↵
  45. Wikipedia, “HESA Shahed 136,” on Ukraine’s acoustic microphone network of ~10,000 units at $400–$500 each, originally developed by two engineers. Link ↵
  46. Drone Warfare, “LUCAS,” citing France’s Chorus program announced Jan. 2026, with Renault manufacturing at 600 units/month. Link ↵
  47. Foreign Policy, “Drones Kill More Civilians Than Pilots Do,” Apr. 2016, quoting Gates (2013) on drones limiting collateral damage. Link ↵
  48. Foreign Policy, “Drones Kill More Civilians Than Pilots Do,” Apr. 2016, on the 35x civilian fatality differential between drone strikes in non-battlefield settings and manned airstrikes in conventional theaters. Link ↵
  49. CNBC, “Iran’s Shahed-136 drone,” Mar. 5, 2026, citing UAE Ministry of Defence on 941 drones detected, 65 impacting UAE territory. Link ↵
  50. Al Jazeera, “‘Reprehensible’: New wave of Iranian missiles, drones target Gulf nations,” Mar. 9, 2026, on Bahrain civilian injuries in Sitra. Link ↵

Comments

Post a Comment

Popular Posts