Ukraine Military History Institute
The threat that matters most is the one with no warning
Rob Lee draws a line through Russia's missile arsenal that most coverage never draws. The question is not only how destructive a weapon is, but how much notice it gives.
Air-launched and bomber-launched strikes announce themselves. Whenever a MiG-31 takes off, the aircraft that carries the Kinzhal air-launched ballistic missile, the air alarm sounds in Kyiv and across the country. When the Tu-95s that carry the Kh-101 take off, the same thing happens, and everyone is ready for the strike.
Ground-launched ballistic missiles are different. Lee's list is the Iskander, the S-400 used in a ground-attack role, the North Korean KN-23 and KN-24, and he adds that Tsirkon and Onyx can still appear, since those are ground-launched too. There is very little warning, they are hard to track, and they can operate in a decentralised way.
His description of what that means in practice is stark. The warning can be seconds before impact. In Kyiv, during the strikes of the previous two months, you usually hear the first impact and then the alarm goes off.
Two minutes at best, by one analyst's own tracking
Dmytro Putiata offers a slightly different number from his own observation. What he tracks is up to two minutes before you hear the impact of the strike.
Either way, the conclusion is the same. Seconds or two minutes is not enough time for a civilian population to act on, and it is very little time for an air defence crew.
The North Korean missiles are larger, and they have improved
Putiata then turns to the North Korean weapons, and his account carries appropriate hedges.
From what he has heard, Russia has two launchers capable of firing the North Korean missiles, which are much larger and more powerful than Iskander. They are reportedly part of rocket brigades under the Moscow Military District. He supposes there will be more: perhaps up to six launchers and more than 100 missiles.
He doubts North Korean servicemen will be the ones firing them. His expectation is that it will be Russia, though he says plainly that we will see.
The accuracy story is the important part. Earlier launches were, in his words, super inefficient, landing a couple of kilometres from the target. Samuel Cook interjects that they have improved since, and Putiata agrees: after a period when they largely disappeared, they came back much better, definitely with help.
His evidence is grim. A strike in the Dnipro area killed an entire family in a house that stood perhaps a couple of hundred metres from a railway station. He assumes the station was the target. A miss of that size is a large improvement on kilometres, and it still killed the wrong people.
Why cheap drones are the proposed answer
Lee's argument for how to contest these launchers starts somewhere unexpected: with the economics of drones that are allowed to fail.
A cheap drone such as the Hornet can be sent up without a target already identified. If it finds nothing, it can simply be crashed into a tree. A loitering munition in the class of the RAM 2X Bulova will not be launched that way, because it is too expensive to send without a known target.
That difference is what makes Ukraine's middle strike campaign in the occupied areas work, and Lee says it is very much enabled by Starlink, though not entirely. Hornets fly the roads, and the effect goes beyond the targets they hit. Trucks see them overhead and are deterred from driving more than they need to. Russia has to change its air defence posture, commit resources, and run fuel trucks in convoys. Lee mentions a video posted the day before showing a fuel convoy being destroyed.
The limit is geographic. If Ukraine wants to hit something 200 to 300 kilometres into Russian territory, Starlink is not enabled there.
The Iskander problem is a targeting problem
This is where the two halves of the chapter meet.
Lee puts the Iskander's 9M723 at roughly 500 kilometres of range, and notes the same launcher can fire the 9M728 and 9M729 cruise missiles at other ranges. In practice the launchers sit within about 500 kilometres of the border. Dynamic targeting there is very difficult, because it requires flying ISR that far, something cheap that can cover the distance and also act as a strike asset.
A Starlink-connected Hornet or Blyskavka is cheap enough, and available in large enough quantities, to do that. His picture has two parts. First, when a launch happens and teams are ready, they immediately send drones up to go hunting around the launch area. Second, persistent ISR overhead makes it harder for the launchers, the TELs, to move at all. He compares it to what happened to artillery once drones were watching it.
He is careful not to overclaim. It won't be a 100% solution, but it can contest the launchers to a greater extent than now.
The ask, and the backdrop
Lee reports that President Volodymyr Zelenskyy made this ask of President Trump, in an exchange that also involved a meeting with Elon Musk: enable Starlink in that territory. He credits Mykhailo Fedorov with a relationship with Musk and Silicon Valley that helped get SpaceX to agree to cut off Russia's use of Starlink. Putiata is drier about it, asking why that relationship only produced results once Fedorov became minister of defence.
The reason the ask matters is the shortage behind it. Lee closes by noting that Ukraine lacks Patriot interceptors, and that the United States has gone through a large share of its own stock, around 65%, which he says, with some hesitation, was the figure reported the previous week.
What to watch
- Whether Starlink is enabled inside Russia's border zone. Lee's argument depends on it. If access opens even partially, launcher hunting near the border should become visible in strike reporting.
- The number of North Korean launchers in Russian service. Putiata's range was two now and perhaps up to six, with more than 100 missiles. Any confirmed count above two tests his expectation.
- Miss distances on ballistic strikes. He described an improvement from kilometres to a couple of hundred metres. Reported impacts relative to likely targets will show whether accuracy keeps improving.