Why does the United States now want to defend its whole territory against missiles, and why does that push the defence into space?
The short answer. the threat has changed shape faster than the defence. For thirty years American homeland missile defence was sized against a handful of long-range missiles from a regional adversary, and ground-based interceptors in Alaska and California were adequate to that job. China and Russia have since fielded weapons that fly lower, manoeuvre and arrive from directions the existing sensors were not built to watch, and they have done so in numbers that change the arithmetic. Seeing those weapons early, and following them all the way, can only be done reliably from orbit.
Four kinds of threat
Golden Dome's executive order lists ballistic, hypersonic and cruise missiles and "other advanced aerial attacks". In practice four threat classes drive the architecture.
| Threat | How it flies | Why it is hard | What it demands of the defence |
|---|---|---|---|
| Intercontinental ballistic missiles | Boost (3–5 minutes), midcourse in space (about 20 minutes), re-entry (under a minute) | Decoys and multiple warheads in midcourse; very short re-entry window | Early warning, discrimination of warheads from decoys, many interceptors |
| Hypersonic glide vehicles | Boosted by a rocket, then glide and manoeuvre in the upper atmosphere | Too low for most ground radars until late; too dim and fast for legacy GEO sensors to track continuously | Continuous tracking from low orbit, fire-control quality data passed to interceptors in seconds |
| Fractional orbital and depressed trajectories | Partial orbit or flattened arc; can arrive over the South Pole | Arrives from directions where warning radars face the wrong way | Global, not regional, sensor coverage |
| Cruise missiles, drones and aircraft | Low, slow (or fast) and small radar cross-section | Hidden by the curvature of the Earth from ground radars | Moving-target indication from above: airborne or space-based |
The first class is the one existing homeland defence was built for. The second and third explain the space-based tracking layer. The fourth explains why a missile-defence programme now funds satellites that track aircraft.
Why space
A ground-based radar sees to the horizon, a few hundred kilometres for a low-flying target and a few thousand for a ballistic missile high in its arc. To follow a glide vehicle across an ocean, or a missile launched from the interior of a large country, a defender would need radars along the whole path, many of them on foreign soil or at sea. A satellite in low Earth orbit sees a footprint thousands of kilometres across and, in a constellation, can hand a track from one satellite to the next without a gap. Geostationary missile-warning satellites have done the first part of this job since the 1970s: they see the hot plume of a rocket in boost. What they cannot do is keep a dim, manoeuvring glide vehicle in view after the rocket burns out. That is the gap the Space Development Agency's tracking layer and the Missile Defense Agency's hypersonic tracking sensor were created to close.
Three properties make space the natural home for the sensing layer:
- Coverage. A proliferated low-orbit constellation covers the whole globe, including the polar approaches that ground radars face away from.
- Persistence. With enough satellites, a target is never out of view of at least two sensors, which is what a fire-control track needs.
- Resilience by numbers. Hundreds of relatively cheap satellites are harder to disable than a handful of expensive ones; losing one degrades the network rather than blinding it.
The same properties make space an attractive place for interceptors in principle: a weapon already in orbit can reach a missile during boost, before it releases decoys. Whether that is affordable is the subject of Chapter 15.
The salvo problem
Missile defence is a contest of numbers as much as technology. Each interceptor has a probability of killing its target below one, so defenders fire several at each threat. Attackers can add decoys and warheads more cheaply than defenders can add interceptors. The traditional homeland system was sized for a limited attack and accepted that a large attack from Russia or China would be deterred by nuclear retaliation rather than defeated.
Golden Dome's ambition, as written in the executive order, is broader: defence of the homeland against attacks by "peer, near-peer and rogue adversaries". That ambition is what turns a programme in the tens of billions into one in the hundreds of billions or more. The cost estimates in Chapter 6 differ above all in how large an attack the architecture is assumed to defeat.
Lessons from Ukraine and the Middle East
Recent wars have shown the threat in practice. Russia's campaign against Ukraine combines thousands of drones a month with cruise and ballistic missiles in mixed salvos, and Iran's attacks on Israel in 2024 and 2025 and the US war with Iran from February 2026 involved more than 1,500 ballistic missiles intercepted by the United States and its partners. Both theatres point in the same direction as Golden Dome's design: early warning and tracking from multiple sensors, cheap defeat of cheap threats, and deep magazines. They also show that interceptor stockpiles, not sensors, run out first. Chapter 3 sets out the evidence.
The adversaries' side
Russia declared its Avangard glide vehicle operational in 2019 and has fielded air-launched and sea-launched hypersonic missiles. China displayed the DF-17 glide-vehicle missile in 2019, and in 2021 tested a system that combined a fractional orbit with a glide vehicle. China's nuclear arsenal is growing quickly, and North Korea continues to test solid-fuel long-range missiles. Chapters 27 and 28 cover the adversaries' counterspace capabilities, which matter because the sensors themselves are targets.
The numbers behind the threat
The threat is growing in quantity as well as variety, which is what moves the debate from a limited defence to a larger one.
| Adversary | Development relevant to homeland defence | Implication |
|---|---|---|
| China | The US Department of Defense assessed that China's operational nuclear warheads exceeded 600 by 2024 and are on course to exceed 1,000 by 2030; new silo fields for solid-fuel ICBMs; DF-17 glide vehicle; a 2021 test combining a fractional orbit with a glide vehicle | A peer-scale arsenal that no plausible defence could stop in full; a reason for deterrence to remain primary |
| Russia | Avangard glide vehicle declared operational in 2019; Sarmat heavy ICBM; air- and sea-launched hypersonic and cruise missiles; nuclear-armed long-range cruise missiles | Multiple approach routes, including over the Arctic, and cruise missiles that low-altitude defence must address |
| North Korea | Solid-fuel ICBMs (Hwasong-18 and Hwasong-19) tested from 2023; continued growth of the arsenal | The limited, regional threat for which homeland defence was originally sized is itself becoming harder, because solid fuel shortens boost phase |
| Others | Iran's ballistic-missile and space-launch programmes | A potential future long-range threat |
Two trends matter most for the space layers. Solid fuel shortens boost phase, which makes boost-phase intercept harder and drives up the number of space-based interceptors needed (Chapter 15). And glide vehicles and cruise missiles move the problem from detection, which existing satellites handle, to continuous tracking at low altitude, which they do not.
Why "peer" changes everything
The difference between defending against North Korea and defending against China is not a matter of degree. Against a few tens of long-range missiles, a layered defence of ground interceptors cued by better sensors is a plausible goal. Against hundreds of missiles with decoys and glide vehicles, any defence can be saturated by an attacker willing to spend on more missiles, which cost less than the interceptors needed to stop them. The executive order's reference to peer adversaries is therefore the single most consequential line for the programme's size, and the reason estimates diverge.
What the threat does not settle
The threat case explains why the United States wants a space-based sensing layer and why it is interested in space-based interceptors. It does not settle how large either should be. A sensing layer sized to give fire-control tracks on a few dozen glide vehicles is a constellation of a few hundred satellites; it is largely funded. An interceptor layer sized to stop a large salvo in boost phase is a constellation of thousands; it is not. Most of the disagreement about Golden Dome's cost is a disagreement about the second, not the first.
What would change this view
A credible arms-control agreement limiting hypersonic or fractional-orbit systems would weaken the case for the most expensive layers. Conversely, an adversary demonstration of large-salvo glide-vehicle attacks or of effective counterspace weapons against low-orbit constellations would strengthen the case for more sensors and for interceptors that do not depend on them.