Russia has been firing fewer ballistic missiles since August. That is the worrying part.

Since August, Russia has significantly reduced its use of ballistic missiles against Ukraine and has increasingly attacked with jet-powered drones instead.¹ That is the observation. Everything else in this column is what people have built on top of it, and I want the seams to stay visible.

The construction, as Ukrainska Pravda relays Bloomberg: the reduction could indicate that Russia is stockpiling those missiles for winter strikes on electricity and heating infrastructure, in the colder months that typically begin in late October or early November.¹ Bloomberg’s stated grounds are that ballistic missiles cause some of the most severe damage to Ukrainian infrastructure, energy facilities included, and that a shortage of interceptors for Patriot systems remains one of the main risks going into winter.¹

Note the difference between the report’s headline and its body. The headline says Russia begins stockpiling. The body says the reduced use could indicate that it is.¹ Those are not the same claim. I am working from the second, because it is the one the report itself stands behind.

Strip it down and the argument has three parts, only one of which is observed.

Observed: fewer ballistic missiles since August, more jet-powered drones.¹

Assumed: that the missiles not fired still exist. A reduction in use is equally consistent with a reduction in supply — production trouble, an inventory lower than assumed, losses in storage. Nothing in the report I read distinguishes a stockpile from a shortage, and those two explanations point in opposite directions for anyone planning around them.

Inferred: intent, and a date. The idea that the saving is deliberate and aimed at the heating season is a reading of purpose. Ukrainska Pravda’s summary does not set out what Bloomberg based it on, and I am not going to invent a basis for it. What the report does supply is a record of past behaviour: according to figures cited by Energy Minister Denys Shmyhal in February 2026, Russia had by then attacked Ukraine’s energy system 5,796 times since the start of the full-scale war, with more than 60 large-scale attacks on energy facilities.¹ That is a Ukrainian government count, seven months old, and I record it as both.

A record of nearly 5,800 attacks on one system is a strong basis for expecting another campaign against it.¹ It is a weaker basis for the specific claim that this particular lull is that campaign being loaded. The first is a forecast from consistent behaviour. The second is a causal story about a gap.

Because the two readings demand opposite responses, and both cost money.

If the missiles are being saved, then the quiet period is the most valuable time Ukraine and its suppliers will get — the window in which interceptors should be moved, grid repairs finished, and substations hardened before the first mass launch. On that reading, the Patriot interceptor shortage the report names is not a background concern but the binding constraint on how the winter goes.¹

If instead the reduction reflects a supply problem, the same actions are still sensible but the urgency is different, and treating a shortage as a stockpile means spending a scarce interceptor budget against a threat that has partly solved itself.

I cannot tell you which it is, and neither, on the evidence in front of me, can the report. What I can say is that the planning assumption and the finding are being conflated in the headline, and that this is the ordinary way a hedged assessment hardens into a fact as it travels.

Jet-powered drones are not a de-escalation. They are what has been arriving in the meantime.¹ Whatever the ballistic missile figures mean, the substitution is a change of tool, not a pause. And energy infrastructure is being struck in the other direction too: this desk has covered Moscow saying it downed 450 drones on election day and the Kapotnya refinery fire.² A refinery there, an energy system here. It is the same reasoning applied by different hands: hit the thing that keeps things running, and let the season do the rest. That comparison is mine.

What would settle the question? Any of these would move me. A resumption of ballistic strikes at previous rates, which would suggest there was never a stockpile to build. Evidence of production capacity or of new deliveries. A first mass launch at the grid in the window the report names, late October into November, which would make the stockpile reading look right in retrospect — though only in retrospect, and by then it would no longer be a warning.

Here is the useful residue, and it does not depend on which reading is correct.

There is a quiet spell in one specific weapon class, acknowledged and dated from August.¹ There is a named shortage of the interceptors that would answer that weapon class.¹ There is a seven-month-old official count of nearly 5,800 attacks on Ukraine’s energy system, and a report that names that system as the expected target of winter strikes.¹ And there is a season with a start date, late October or early November, which is the deadline every one of those facts is measured against.¹

That is enough to act on without believing the headline. The lull is real and the interceptor shortage is named; whether the lull is a stockpile being built is the part nobody has shown.¹ Which of the two readings a defender should plan against is a judgment about costs I have not been given the figures to make, and I have said above that both readings cost money. What I will say is that the reading changes the answer, that the report’s own body says only “could indicate,” and that a hedged assessment with its hedges intact is more use to anyone deciding than a firm one that has quietly dropped them.¹