Ukraine Military History Institute
The assault failed, and that is not the interesting part
Rob Lee describes a recent Russian armoured assault that did not work. The tanks never reached the forward line of troops. They were stopped well before it.
What makes it worth studying is who called it sophisticated. Lee spoke to a soldier from that direction two weeks earlier, someone fighting since 2014 whom he calls a really serious guy. His verdict was that it was the most sophisticated assault he has seen by Russia.
Lee's summary is the line to hold onto: Russia is experimenting, it is learning, and it tried a whole set of innovations at the same time. A failure that tests several fixes at once still tells the attacker which of them worked.
Tanks that took 10 to 20 FPVs each
Part of the package was protection on the tanks themselves. Lee is careful not to oversell it. The active protection systems were an early experiment, they had blind spots he declines to locate, and Ukrainian FPV crews worked out how to deal with them.
The cost of dealing with them is the point. Lee says the tanks were taking 10 to 20 FPVs each before they were stopped.
Scale that up and the arithmetic turns uncomfortable. In a battalion-sized armoured assault, if every tank absorbs that many drones and some of the other components succeed, Lee says it becomes a big problem. He notes that Avdiivka was the last time battalion-sized mechanised assaults were seen. If Russia returns to them having fixed some of the other weaknesses, that becomes a challenge.
Mine rollers that drove through 16 mines
The second component was breaching. The tanks carried mine rollers fitted with some kind of demagnetiser, aimed at the PTM-3, the remotely delivered anti-tank mine. Dmytro Putiata cuts in to note it is the most common one.
According to Lee, at least one tank went through 16 of them.
The detail he finds telling is what Russia skipped. Russian units normally carry out engineer reconnaissance, checking routes for mines in advance. In this operation they did not. Lee's explanation is that they expected the rollers to work well enough that mines would not be the main obstacle.
That points at a Ukrainian dependency. The PTM-3 is one of the main ways Ukraine mines roads, and Russia has answers to it, including spotting the mines from UAVs. Lee raises the possibility that a different anti-tank mine is needed, delivered differently, perhaps something like an EFP.
The jammer was the most interesting piece
Alongside the tanks, Russia brought forward a Starlink jammer. Lee will not say which system, only that it was a problem, and he calls it maybe the most interesting part of the operation.
His broader judgement is measured. Russia has identified how Ukraine defends against its assaults. Its execution is often flawed, but its concepts are frequently not wrong. It tends to find the real problems, and it does produce solutions.
Russia writes it down. Ukraine, mostly, does not
Putiata turns this into a point about institutions, and it is the sharpest moment in the discussion.
He says Russia tested the Arena-M, saw a lot of problems and knows how to counter them. Then he describes the paperwork behind that. Russian units prepare reports of around 150 pages. He recently saw one from Rubikon on how to deploy unmanned forces, troops with little ammunition and ISR. That, he says, is exactly what every army should do.
Then the comparison. From his own experience at the headquarters of a brigade in Ukraine's Unmanned Systems Forces, he has never seen anything like it on the Ukrainian side. No proper analysis, no documentation, no procedure.
He does not say Ukraine lacks information. It has a lot of data from both sides, it sees Russian patterns, and it knows its own advantages and weaknesses. What is missing are the analytical centres that would gather that data and turn it into procedures and documents. Russia has them. He calls that a very big advantage for the enemy, one Ukraine does not have right now.
Russia's own figure: 80% of armour losses come from UAVs
Lee adds a number from the Russian side. Russia determined internally that 80% of its armour losses come from UAVs.
He uses it to push back on a 60/40 split that came up earlier in the conversation. Samuel Cook agrees the real share is higher and suggests the lower figure probably came from someone with an interest in artillery. Lee thinks the UAV share of casualties is above 80%.
He adds the caveats himself. It depends on the sector. Holding a city, glide bombs account for more. If you count TBIs, they are the majority of casualties, but men are often not rotated off for them, which raises the question of which casualties are being counted at all. Still, when he asked every brigade he spoke to last year, the answer was 75% or higher almost every time.
Why the problem is hard, and what Russia decided it hinges on
Lee lays out what an attacker without air superiority is up against. Ukraine has FPVs, heavy bombers, middle strike capability, artillery and ATGMs, very good situational awareness through systems such as Delta, and a lot of ISR drones. Russia has not achieved air superiority, so it is trying to work the problem from the ground.
The same assault showed how. Rubikon was operating in that direction, including some of the Lancets. Russia had run a drone line initiative there, with separate echelons of fires and of responsibility. Artillery was tasked to hunt UAV teams.
That last choice follows from a conclusion Lee says Russia reached last year: that UAV teams are the centre of gravity of the military. Disrupt them and manoeuvre becomes possible again. Fail to, and nothing else is likely to succeed.
What to watch
- Battalion-sized armoured assaults returning. Lee names Avdiivka as the last time they were seen. A repeat would suggest Russia believes its fixes now work.
- A change in how Ukraine mines roads. If the PTM-3 keeps being rolled through and spotted from the air, a different mine or delivery method should appear.
- Ukrainian after-action analysis becoming institutional. Putiata's gap is specific: analytical centres that turn data into procedures. Whether such a centre appears is checkable.