Read 12 of 24

February became September

Published 2 October 2026

A recovery can be real and still leave a shortage.

Late-September reporting dates the US-Israeli war on Iran to 28 February 2026. Al Jazeera's 30 September account describes seven months of hostilities, negotiations through mediators and a continuing US blockade of Iranian ships and ports. It also describes oil flows recovering while security and insurance risk remained unresolved. These are the publisher's reported events and assessments, not a declaration of a final outcome. Al Jazeera, 30 September.

The official monthly market assessment provides a dated anchor. Published on 11 September, the IEA report describes global production falling to 100.1 million barrels a day in August, with more than ten million barrels a day of Gulf output shut in. Its annual forecasts were supply falling 5.7 million barrels a day and demand falling 2.5 million in 2026. These are September forecasts built on information available then, not the final annual outcome. IEA September Oil Market Report.

Why would demand fall during a shortage? Because high prices and unavailable products suppress consumption. Factories can curtail production, transport activity can become more expensive and buyers can substitute where possible. Demand destruction is a description of consumption that does not occur under the shock, not proof that the energy system no longer needs fuel.

The report also records global observed inventories falling by 507 million barrels since February, averaging about 2.8 million barrels a day. August alone accounted for a ninety-five-million-barrel draw. The stocks were bridging the difference between ongoing needs and constrained supply. A stock draw can prevent an immediate collapse in consumption while leaving a thinner buffer for the next interruption. Source record.

Now move forward nineteen days. Al Jazeera cites Kpler estimates of 16.328 million barrels a day of Middle Eastern crude exports during September, versus 19.513 million in February. It reports about 9.719 million passing through Hormuz itself. Regional exports and strait flows are not the same measure: regional exports include alternative routes. Dividing one regional figure by another gives roughly eighty-four percent, while the article's broader recovery description uses just under eighty percent. We preserve the reported figures and avoid turning those differing formulations into false precision. Source record.

Bloomberg commentary on 30 September similarly argues that exports by regional US allies through the waterway plus bypass routes had recovered to about eighty percent of pre-war levels, while Iranian exports were heavily constrained. It is commentary with an explicit geographical scope, not a measurement saying every Hormuz product flow is eighty percent restored. Javier Blas, Bloomberg, via ET.

A report describing August's effective closure and a report describing September's partial recovery need not conflict. A report about cargo volumes and one about vessel counts may also diverge because vessel size, cargo type, tracking gaps and routing differ. The important contradiction to avoid is calling today's strait completely shut while citing recovering cargo flows as if they were the same dated observation.

Politics adds another uncertainty. Donald Trump has described an approaching end to the war, while Iranian officials reject claims that their leverage has vanished. These are positions and claims. A forecast of an agreement is not an agreement. The article reports new strikes on ships and continued elevated tanker insurance costs, which explain why commerce can resume without confidence returning fully.

For India's buyer, the practical question is not who won a press conference. It is whether the nominated cargo can be loaded, insured, financed and delivered on schedule. A more secure route can lower risk premiums. A new attack can raise them again before much actual volume is lost.

The ratio that resists a slogan

Sources: IEA baseline and dated Kpler reporting. mb/d means million barrels per day. The regional figure is not a strait-only figure, and the total-oil baseline is not a crude-only baseline. The table exists because a false comparison can tell a convincing but wrong story of closure or recovery.

The reported regional volumes imply a shortfall of 3.185 million barrels a day against the February comparator. That is arithmetic on provisional observations, not a complete estimate of the world's lost supply. It excludes other production changes, product flows and the timing of inventories. It also cannot establish how much of the shortfall will remain next month. Source record.

Information latency creates another trap. A monthly report published on 11 September can be more rigorous than a breaking story on 30 September, yet describe an earlier operational situation. The later story can capture a change before the official dataset does, while remaining provisional. A responsible account uses both for the claims each can support.

Policy and diplomacy respond to those lags. A leader may claim progress before operators trust the route. A shipowner may wait for evidence of safety. A buyer may diversify even after some cargoes resume. The resulting price premium is partly compensation for uncertainty about the next voyage, not only the amount already lost.

The month's average does not protect tomorrow's ship

Fresh reporting on 30 September adds a warning against reading a recovery average as a safety certificate. Lloyd's List identifies three tankers, Mersin Prosperity, Sinbad and Al Ruwais, in reports of attacks during Hormuz transits. The specialist publication says UKMTO issued alerts and describes the vessels as moving with AIS signals switched off. This file has inspected Lloyd's List's published account, not the original UKMTO alerts or an independent record of each incident. The vessel names and circumstances remain attributed reporting. Lloyd's List, 30 September.

The same account says most of the limited recent tracked transits used the Larak Island detour, and that no traceable crossings were recorded on Tuesday. Its own reference to vessels with AIS switched off is a reason to preserve the word traceable. No recorded crossing in that dataset is not proof that no ship physically crossed. A tracking system can reveal a dangerous pattern without becoming an all-seeing census.

Nor does the headline's description of failed ceasefire talks settle every diplomatic question. Negotiation status, ship movements and completed export cargoes are three different observations. An agreement announced in one room may take time to change an insurer's terms or a shipowner's willingness to sail. Conversely, an escorted cargo can move without a comprehensive political settlement. The story must resist turning either fact into the other.

For India, that creates two clocks. A monthly sourcing mix says what the country has been buying. A refinery's next loading window asks whether a specific cargo will arrive in time. A broad regional recovery can coexist with an expensive or dangerous final crossing. Security is not obtained by winning an argument over the best monthly percentage; it is obtained by making the next necessary delivery less vulnerable.

The deal can founder on which step comes first

The late-September diplomatic record is more specific than a blanket statement that talks either succeeded or ended. On 26 September, Reuters reported Iran's seven-day proposal transmitted through Qatari mediators. It would pair reopened Hormuz passage with a halt in fighting and demands including the lifting of the US blockade, sanctions relief and access to frozen funds. Reuters explicitly said it could not immediately confirm the Wall Street Journal's report that Trump had rejected the proposal. Reuters, 26 September.

By 30 September, Reuters reporting republished by Business Recorder said Iran had received a US response. It described Qatari mediators shuttling between envoys in New York, and an official briefed on the talks saying the sides largely agreed on needed steps but differed on sequence. The source was unnamed, and the article did not establish a final signed settlement. It also described Iran's plan as restoring a June interim agreement signed by Trump and President Masoud Pezeshkian that had later collapsed. Reuters via Business Recorder, 30 September.

Sequence is not a procedural nuisance. If one party opens a route first, it may surrender leverage before receiving the promised concession. If the other removes a blockade first, it may surrender its own leverage. Each wants evidence that the next step will follow. The difficulty is making reciprocation credible when the previous arrangement has already failed.

That explains the commercial delay beneath a diplomatic headline. A shipowner does not merely need a list of political intentions. The owner needs an insurable voyage, a crew willing and able to sail, and confidence that permissions remain valid during the crossing. A refinery needs the agreement to last through its loading and delivery windows, not only through the press conference.

Semafor's 30 September account says the White House was lowering expectations of a breakthrough while the sides continued speaking through mediators. It quotes Trump's pessimism about whether Iran would give up. This is reported political positioning, not independent proof of the opponent's condition or a forecast of the negotiations' outcome. Semafor's expectation-setting report.

The oil market receives these messages before the infrastructure receives repairs. Prices can respond quickly to the chance of a settlement; actual production, loading and insurance can respond later. A premature peace trade can fail if the sequence stalls. A permanently catastrophic narrative can also fail if usable flows recover. The sensible story stays close to the observable steps between a promise and a delivered cargo.

Oil prices are therefore part scarcity, part timing and part uncertainty. The recovery in crude did not repair every refinery or restore every diesel cargo. That is the second shortage hiding behind the first.

Get the next file

One email when a new file is published. No paid service needed.

Email Subscribe

You are subscribed.