What have the wars of 2022–2026 taught about missile defence, and what do they mean for the design and cost of Golden Dome?
The short answer. three wars have now tested layered missile defence at a scale no exercise could. Ukraine has absorbed thousands of drones and cruise missiles and hundreds of ballistic missiles a year since 2022. Israel, the United States and Gulf partners defeated most of Iran's ballistic-missile salvos in April 2024, October 2024, June 2025 and again from February 2026. The lessons are consistent. Space-based warning and commercial space services are indispensable and, so far, resilient. Interceptor magazines are not: in the 2026 Iran war the Congressional Budget Office estimates the United States used between one-half and two-thirds of its inventory of certain missile-defence interceptors, and rebuilding stocks could take at least five years. Golden Dome's most binding constraint may be on the ground, in solid rocket motor and seeker factories, rather than in orbit.
Why these wars matter for a homeland shield
Golden Dome is designed for an attack on the United States that has never happened. The wars in Ukraine and the Middle East are the closest available evidence about how layered defences behave against real salvos: how many interceptors are fired per threat, how fast stocks are drawn down, how attackers adapt, and which parts of the architecture fail first. They are imperfect analogues. The distances are shorter, the missiles mostly theatre-range rather than intercontinental, and the defended areas much smaller than a continent. But the mechanics of saturation, cost exchange and magazine depth apply directly, and so does the role of space.
Ukraine: four years of mass salvos
Russia's long-range strike campaign against Ukraine has combined one-way attack drones, cruise missiles and ballistic missiles in mixed salvos designed to exhaust and confuse the defence. The volume has grown each year. A Ukrainian defence ministry adviser said that in January 2026 alone more than 6,000 missiles and drones were launched against Ukraine; other reporting puts Shahed-type drone launches at about 100 a day, or roughly 3,000 a month.
Ukraine's answer has been to sort threats by cost and to defeat cheap threats cheaply. Interceptor drones costing an average of up to about $3,000 each, according to the same adviser, have become the main defence against Shahed-type drones; one type alone was credited with more than 3,000 drone kills in the first four months of 2026. Mobile gun teams, electronic warfare and short-range systems take more. Expensive interceptors such as Patriot, which cost several million dollars per missile, are saved for ballistic and high-value cruise missiles.
Against ballistic missiles the record is harder. The Royal United Services Institute's dataset to 24 October 2025 counted 939 Iskander and Kinzhal missiles fired at Ukraine, of which 227 were intercepted. Russia modified the Iskander-M and Kinzhal to dive steeply or manoeuvre late in flight and, according to Ukrainian officials, to release radar decoys; Ukrainian Air Force data showed the share of ballistic missiles shot down falling from up to 37% in the summer of 2025 to 6% in September, though on fewer launches. RUSI cautions that these percentages mix missiles fired at areas without Patriot coverage with genuine failures. The direction of travel is nonetheless clear: attackers adapt their missiles to the specific defence they face, and they do it in months, not years.
Ukraine: space as a combatant's utility
The war in Ukraine is also the first in which commercial space services became part of the front line on both sides.
| Space service | How it has been used | What happened |
|---|---|---|
| Satellite broadband (Starlink) | Ukrainian command, drone control and logistics; also used illegally by Russian units | In early February 2026 SpaceX and Ukraine activated a whitelist of registered terminals; Ukrainian officials said Russian units lost access and frontline activity fell. Reuters could not verify the scale independently |
| Commercial radar and optical imagery | Ukrainian targeting and damage assessment; ICEYE radar satellites; US government-purchased imagery | In March 2025 the US paused Ukraine's access to the government imagery portal (GEGD) for about a week alongside a pause in intelligence sharing; access was restored on 11–12 March |
| Satellite navigation (GPS) | Precision munitions and drones on both sides | Persistent Russian jamming and spoofing, extending to the Baltic region, where tens of thousands of aircraft have reported GPS interference |
| Missile warning | US and allied warning of Russian launches | Not publicly described in detail; no reported loss of warning capability |
Three lessons follow for Golden Dome. First, a proliferated commercial constellation proved hard to disable by jamming or attack; Russia's reported attempts to disrupt Starlink with electronic warfare did not take it out of the war. That is the strongest real-world evidence for the Space Development Agency's proliferated design. Second, control of a commercial network is itself a lever, for the operator and for the government that licenses it, as both the Starlink whitelist and the imagery pause showed. A defence architecture that depends on one commercial operator inherits that operator's policies and its government's decisions. Third, navigation and communications satellites face constant electronic attack in war; resilience to jamming is not optional for any layer.
The Iran–Israel exchanges of 2024 and 2025
Iran launched its first direct attacks on Israel in April and October 2024, each combining drones, cruise missiles and ballistic missiles. Both were largely defeated by a coalition defence: Israeli Arrow and David's Sling interceptors, US and partner aircraft and ships, and early warning from space.
The twelve-day war of June 2025 was larger. Estimates of the number of Iranian ballistic missiles fired range from about 525 (Alma Research) to 574 (the Jewish Institute for National Security of America, JINSA) and 631 (other analysts). JINSA counted 322 missiles engaged and 273 intercepted, an 85% success rate, and estimated that only 49 struck meaningful targets. The cost was in interceptors. Estimates of US THAAD use range from 100 to 150 interceptors (reported by CNN) to between 100 and 250 (JINSA), equivalent, on CSIS figures, to 20–50% of the Pentagon's total stock. About 80 SM-3 interceptors were reported fired, including in combat roles first undertaken in 2024. Israel's own Arrow stocks were reported to be under strain by the end.
The 2026 Iran war
The United States began military operations against Iran on 28 February 2026. A ceasefire took effect in April and was extended, fighting resumed in July after attacks on shipping in the Strait of Hormuz, and by late August the shooting had paused again under an extended ceasefire with mediation continuing. This study did not find an authoritative account of the conflict's status in October 2026.
Exhibit 3.2
Patriot use ran into the thousands; THAAD use was most of the pre-war stock
US interceptors expended in the 2026 Iran war (estimates) · interceptors
Patriot (PAC-2/PAC-3)
THAAD
Standard Missile (SM-3/SM-6)
each bar runs from the low to the high estimate
The war produced the largest missile-defence campaign in history. The commander of US Central Command told Congress in May 2026 that the United States and regional partners had intercepted more than 6,000 one-way attack drones and more than 1,500 ballistic missiles. The United Arab Emirates alone reported, by 10 May, a cumulative 551 ballistic missiles, 2,265 drone attacks and 29 cruise missiles directed at it; Bahrain reported stopping 70 missiles and 59 drones on a single day early in the war.
| Measure | Estimate | Source |
|---|---|---|
| Ballistic missiles intercepted by the US and partners (to May 2026) | More than 1,500 | CENTCOM testimony, cited by CRS |
| One-way attack drones intercepted (to May 2026) | More than 6,000 | CENTCOM testimony, cited by CRS |
| Patriot interceptors expended | ~1,700 (US officials, reported August 2026); 1,060–1,430 (other estimates) | Press reporting cited by CRS; Al Jazeera compilation |
| THAAD interceptors expended | ~200 (reported); 190–290, about 82% of pre-war inventory (CSIS, April 2026) | CRS; CSIS |
| Standard Missile interceptors expended | ~150 (reported); 130–250 SM-3 (other estimates) | CRS; Al Jazeera |
| Share of certain interceptor inventories used since June 2025 | One-half to two-thirds | CBO, September 2026 |
| Time to rebuild those stocks | At least five years | CBO |
| Cost of operations to 1 August 2026 | ~$38B, of which $21.7B to replace munitions and $13.1B for missile-defence interceptors | CBO |
| Supplemental request | $67.1B, including $21B for munitions (24 June 2026) | CRS |
The Department of Defense does not publish interceptor inventories, and the counts above are estimates from officials speaking to the press, think tanks and the CBO. They should be read as ranges. The direction is not in doubt: a regional war against a middle power consumed a large fraction of the United States' most capable interceptors in a few months.
Israel's experience was similar. A RUSI analysis estimated in March 2026 that Israel had used about four-fifths of its pre-war Arrow stocks after roughly four weeks; the Israeli military rejected reports of a shortage. Ukraine, which had become the world's most experienced operator of cheap counter-drone systems, sent drone units to five Gulf states at their request, reportedly in exchange for Patriot interceptors.
What the wars teach
| Lesson | Evidence | Implication for Golden Dome |
|---|---|---|
| Magazines run out before sensors do | Half to two-thirds of certain US interceptor inventories used since June 2025 (CBO); Arrow stocks strained in 2025 and 2026 | Interceptor production capacity, not satellites, is the first bottleneck; funds for ground interceptors compete with space layers |
| Cheap threats must be defeated cheaply | Ukraine's interceptor drones at up to ~$3,000 versus Shahed at ~$35,000–50,000 and Patriot at ~$4M | Directed energy, guns and electronic warfare belong in the architecture to protect the interceptor inventory |
| Attackers adapt in months | Iskander-M and Kinzhal manoeuvres and decoys; interception share fell sharply in late 2025 | Discrimination and fire-control tracking from space matter more than raw detection |
| Warning from space works and is relied upon | US warning supported Israel and Gulf partners; SBIRS warned US forces in Iraq of Iran's missile strike in January 2020 | The warning and tracking layers are battle-proven in principle; their expansion is the least controversial part of Golden Dome |
| Proliferated commercial constellations are resilient | Starlink stayed in service despite jamming attempts | Supports the proliferated LEO design of the tracking, transport and backbone layers |
| Dependence on one operator is a political risk | Starlink whitelist; one-week US imagery pause for Ukraine | Concentration on a single supplier for the backbone and moving-target layers (Chapter 19) is a governance issue as well as a commercial one |
| Defence of a small area is hard; of a continent, harder | Israel and the Gulf states defend areas a fraction the size of the United States, at close range | The salvo arithmetic of Chapter 15 is optimistic if anything; coverage choices will be explicit |
The cost-exchange problem
The wars put numbers on a familiar asymmetry. Interceptors costing between hundreds of thousands and tens of millions of dollars are fired at missiles and drones that cost from tens of thousands to a few million. At the scale of the 2026 war, the CBO's estimate of $13.1 billion to replace missile-defence interceptors is about three-quarters of Golden Dome's entire fiscal 2027 request. Each $1 billion of replacement interceptors is money that cannot buy satellites unless budgets grow.
Exhibit 3.3
A Patriot costs about a hundred times a Shahed drone; an interceptor drone costs less than the drone it kills
Cost per shot, attacker and defender (approximate) · US$ per unit, log scale
For Golden Dome the implication is that the cheapest part of the architecture, sensing and data, makes the expensive part more efficient. Better tracking from space lets the defence fire fewer interceptors per threat, wait to see whether the first shot worked, and ignore objects that will miss or are decoys. A one-shot reduction in the interceptors fired against each incoming missile, from three to two, reduces interceptor consumption by a third.
Production, not invention
The industrial response to the wars began before Golden Dome. Lockheed Martin delivered 620 PAC-3 MSE interceptors in 2025, and in January 2026 signed a framework agreement with the Department to raise annual capacity from about 600 to 2,000 over seven years, following a $9.8 billion multi-year Army contract in September 2025 for nearly 2,000 interceptors. L3Harris signed a propulsion agreement in July 2026 that it says will allow it to nearly triple production of motors for the interceptor. These are the same solid rocket motor and seeker supply chains that a space-based interceptor would need (Chapter 21), and the wars have given them a guaranteed customer for years.
What would change this view
A durable end to the Iran conflict and a sustained rise in interceptor production toward the 2,000-a-year PAC-3 target would ease the magazine constraint by the end of the decade. A renewed large-scale war, or a Russian or Chinese demonstration of a new class of manoeuvring missile against a US-built defence, would sharpen it and strengthen the case for spending on ground interceptors and directed energy before space-based interceptors.