AI-95+ against the 2024 norm

43%

146,4 of 336,5 kt/week · confirmed only: 50%

AI-92 against the 2024 norm

67%

265,9 of 394,2 kt/week · confirmed only: 62%

Capacity out of action

67%

AI-95+, against the undamaged fleet

Retail outside Moscow

24%

45,9 of 192,7 kt/week

Strikes with damage

in the record, since 1 June 2026

Heat map of the refineries

where the damage sits

■ grey = running, but within reach■ green = beyond reach, further than 3000 km■ red = damaged, saturation by the share out of actiontwo red circles = at a standstillblast with a number = damaging hitsouter arc = output above the plant's own norm

Green marks the four plants further than 3000 km from the launch area: they have not been hit and there is no reason to expect them to be, which is why their capacity stands as its own group, beyond reach, on every chart here. The other grey cells are running, but they can be reached.

Make no motor gasoline and are excluded from the map: Tuapse, Novoshakhtinsk, Ilsky, Afipsky, Yaysky, Mari, Krasnodarskiy. Yaysky and Mari carry no motor gasoline either, and no damage is recorded for them.

Where it comes from and who gets it, AI-95+

who is short

The space of scenarios

where this goes

solid blue = a pause in strikes, then the record's temposolid grey = the same from S0, confirmed onlydashed = the ±40% corridorgreen = plus the scaled-up importgrey dashes = if the strikes stopcoloured steps = consumption cut by force by the Kremlinblack dashes = thresholds below which somebody goes unfed

Pressure 40% stronger is one and the same outcome from any of these: 40% more strikes; the same number of strikes but 40% more effective; or an equivalent slowing of repair, say because more high-octane equipment ends up under attack. Pressure 40% weaker is the mirror: fewer strikes, lower effectiveness, better air defence and jamming, or faster repair. Combinations land in the same place: 20% more strikes plus 10% on effectiveness plus 10% on repair times give roughly the same line, and a reader who wants 20% rather than 40% takes the middle between the solid line and the dashes.

The steps above bring down only the part of demand that is not unconditional: rationing does not touch the army, the state services or the food deliveries.

The same picture for AI-92

the second deficit circuit · substitution is limited

blue = output at the current tempodashed = the ±40% corridorblack dashes = thresholds derived from the AI-95 sharescoloured steps = consumption cut by force by the Kremlin

Two remarks that matter

The charts do not model the consumption that disappears through long queues, higher prices or the plain absence of fuel at the pump. That forced removal of consumption is exactly what the regime will carry out without announcing it, which amounts to admitting that it cannot close the gap shown here and to moving that gap into economic and political costs instead.

The charts do not model the switch from AI-95+ to AI-92, the permitted K2 or Euro-2 fuel included, because AI-92 is itself in serious deficit. In theory some 30 to 40% of normal AI-95+ demand could move across, which is another 101 to 135 kt a week, or 26 to 34% on top of the normal AI-92 requirement that is already not being met. A clean substitution is therefore impossible: it only moves the shortage between grades, adds price pressure to both, and turns access to fuel into an ever harder fight through queues, the black market, corruption and the forcing out of the weak and of those far from the centre and from the regime.