IEA cuts 2026 oil-demand forecast by 940,000 b/d as fuel scarcity curbs consumption
The International Energy Agency cut its 2026 oil-demand forecast by another 940,000 barrels a day and now expects global consumption to fall by 2.5 million b/d this year. Read as a normal demand revision, that sounds bearish. The rest of the September Oil Market Report points the other way: the US-Iran conflict is still impairing Middle East supply, inventories have been drawn down and refined products remain scarce enough to force consumers to use less.
The agency also cut its 2026 supply forecast by 1.3 million b/d from August and pushed a full Middle East recovery into 2027. Observed global inventories fell another 95 million barrels in August and are down 507 million barrels since February, an average draw of roughly 2.8 million b/d. Chevron chief executive Mike Wirth described the stock releases and previously constrained supply that softened the first phase of the shock as buffers that have largely been played out.
The sign on demand is less informative than its cause. Missing crude and product barrels raise producer realizations and refinery margins while lifting fuel costs for freight, agriculture, airlines and industry. Consumption then gives way because the delivered price is clearing a shortage.
Refined products are tighter than crude
The bottleneck is most visible downstream. Gulf countries exported roughly 13 million b/d of oil in August, about half their pre-war level. Crude losses had narrowed to just under 45%, helped by bypass routes and US military escorts through the Strait of Hormuz. Refined-product and LPG exports were still almost 60%, or 3.7 million b/d, below February levels. Gulf diesel and gasoil exports averaged only about 390,000 b/d, little more than a quarter of their pre-war pace.
Russian refinery disruptions widened the deficit. The IEA put combined Gulf and Russian net diesel and gasoil exports 1.6 million b/d below February. That helps explain why middle distillates have moved more violently than crude. When the agency wrote the report, ICE Brent was around $105 a barrel, while US diesel had exceeded $200 a barrel and Atlantic Basin refining margins had reached record levels in August.
A crude-only read understates the economic shock. A refinery can have access to feedstock while the system still lacks enough finished diesel, jet fuel or LPG in the locations that need it. The useful unit is the barrel that can actually move through terminals, shipping routes and refineries to the end user.
Saudi nameplate capacity overstates the usable cushion
Saudi Arabia illustrates the distinction. The IEA listed Saudi sustainable capacity at 12.11 million b/d, August production at 5.97 million b/d and an implied target of 10.42 million b/d. In normal conditions that gap could be read as a large supply cushion. Under the agency's disruption-adjusted framework, security, terminals, export routes and shipping constrain what can actually reach the market. Its effective spare-capacity estimate for the OPEC-8 was only 70,000 b/d.
That physical constraint reconciles the apparently contradictory demand numbers. The IEA expects demand contraction to slow from 5.3 million b/d in the second quarter to 3.4 million in the third and 2 million in the fourth. The agency explicitly links the losses to missing petrochemical feedstocks, product shortages and high fuel prices. Consumption is adjusting because supply cannot clear at tolerable prices.
There is a credible bearish route from here. A durable diplomatic settlement that restores Gulf exports, followed by the IEA's projected 8 million b/d rebound in global supply in 2027, would reverse today's scarcity quickly. Crude's September 11 pullback on renewed diplomatic hopes shows how much of the price still depends on that probability.
The IEA's 70,000 b/d disruption-adjusted OPEC-8 spare-capacity estimate is the number that defines the present regime. With Gulf crude and product flows still impaired, weaker 2026 demand is part of the scarcity adjustment itself. Restored physical supply, rather than another forecast revision, would change that conclusion.