The Hormuz Shock: Can Digital Intelligence Defend a Physical Chokepoint?

Special Investigation | Energy Security & Digital Operations
By Prasad Selvaraj — The Digital Oil Field, Investigations Desk
Data current as of 24 August 2026

Six months into the worst disruption in the history of the seaborne oil trade, the industry has deployed every analytical tool it owns. This investigation tests what those tools actually changed — and finds the answer sitting in two pipelines, not two algorithms.

Note on charts: this article contains four interactive charts and a full data appendix — sourcing ledger, methodology notes and the fact-check table. Those live in the companion data edition: The Hormuz Shock — full data edition. The underlying numbers for every chart are reproduced as tables below.

Executive brief: eight findings

• Oil transit collapsed by roughly three-quarters. The US Energy Information Administration puts petroleum and other liquids moving through Hormuz at 4.9 million b/d in Q2 2026, against 21.6 million b/d in Q4 2025.
• Traffic has not recovered. Kpler tracked seven commodity vessels through the Strait on 21 August — four inbound, three outbound, and no very large crude carriers or LNG tankers at all. February's pre-crisis average was around 135 vessels a day.
• The headline numbers do not agree, and the gap is instructive. The IEA reports Hormuz flows averaging 2.7 million b/d across March–May; the EIA reports 4.9 million b/d across April–June. Different windows, different definitions, one genuinely opaque waterway.
• The two most-quoted demand forecasts point in opposite directions. In August the IEA cut its 2026 outlook to a contraction of 1.6 million b/d; OPEC, the same week, still projected growth of 580,000 b/d — a spread of well over two million barrels a day on the same year.
• This became a refined-products crisis before it became a crude crisis. The US diesel crack spread hit a record $102.20 a barrel on 17 August, with Brent itself near $93. The scarce commodity is not oil; it is diesel.
• Inventories are the shock absorber, and they are draining. Global observed stocks fell 410 million barrels between end-February and end-July — about 2.7 million b/d — even after the largest coordinated emergency release ever attempted, 400 million barrels agreed by 32 IEA members.
• Qatar has no bypass, and LNG has no pipeline alternative. No laden LNG vessel crossed the Strait between 1 March and 24 April. Roughly 20% of global LNG trade — over 10 Bcf/d — normally transits Hormuz, almost all of it Qatari and almost all of it Asia-bound.
• What actually moved barrels was steel, not software. Saudi exports from Yanbu rose from 2 to more than 5 million b/d via the East–West pipeline. No forecasting system produced a comparable result — because the binding constraint was never information.

A waterway with almost nothing in it

On Thursday 21 August, seven commodity vessels moved through the Strait of Hormuz. Four went in, three came out. Among them there was not a single very large crude carrier and not a single LNG tanker. The one notable cargo was a large gas carrier loaded with propane and butane, and it left along the Iranian side of the channel.

Seven ships. In February, before any of this began, the daily average was around 135.

The number came from Kpler's ship-tracking feed, and it arrived with a caveat that has become the defining condition of this crisis: vessels running with their transponders switched off do not appear in it. The true figure is higher than seven. Nobody outside a handful of governments can say how much higher.

That single caveat is the reason this investigation exists. Six months into the largest disruption the seaborne oil trade has ever experienced, the energy industry has pointed everything it owns at this waterway — satellite constellations, machine-learning cargo classifiers, vessel-tracking platforms, refinery optimisation models, digital twins of entire export chains. The instrumentation is extraordinary. The question is what it bought.

The answer, on the evidence assembled here, is a great deal of visibility and remarkably little supply. And the distinction between those two things turns out to be the most important lesson the industry will take from 2026.

Twenty-one miles that move the world

The Strait of Hormuz is a bend of water between Iran and Oman, twenty-one miles across at its narrowest, with a shipping channel narrower still. Everything Kuwait, Iraq, Bahrain, Qatar and Iran export by sea passes through it. So does the great majority of Saudi and Emirati crude, because both countries' primary loading terminals sit inside the Gulf.

Before February 2026, that traffic amounted to something in the region of a fifth of global oil consumption and roughly a quarter of seaborne oil trade. The exact figure depends on whose definition you use — a point this article returns to at length, because in 2026 it stopped being a pedantic distinction and became a live analytical problem.

The crisis began on 28 February, when coordinated US and Israeli strikes on Iran were followed within hours by Iranian Revolutionary Guard broadcasts warning shipping out of the Strait. Traffic fell roughly 70% almost immediately, then further. Qatar halted gas production on 2 March and declared force majeure two days later. Brent passed $100 on 8 March for the first time in four years.

What followed was not a single closure but a sequence of partial reopenings, ceasefires, re-closures and negotiated exemptions — a pattern that has proven far harder for markets, and for forecasting models, to handle than a clean shutdown would have been. As of this week the Iranian parliament is advancing legislation to levy transit service fees, Tehran has warned of renewed vessel seizures, and Washington is preparing a further sanctions package. The waterway is neither open nor closed. It is negotiable, daily, per vessel.

A clean closure is a supply problem. An intermittent, discretionary, politically-priced closure is an information problem — which is precisely why the industry expected its analytics to earn their keep.

What the numbers say

The scale of the interruption is not in serious dispute, even where the decimal places are. The dashboard below collects every indicator for which this investigation could obtain a documented pre-crisis baseline and a documented recent reading on a comparable basis. Where no comparable figure exists, the cell says so rather than estimating one.

Table 1 — Hormuz Crisis Dashboard

Indicator	Baseline	Latest	Change	Source
Hormuz oil transit (petroleum & other liquids)	21.6 mb/d (4Q25)	4.9 mb/d (2Q26)	−77%	EIA
Hormuz oil transit (IEA basis)	~20 mb/d	2.7 mb/d (Mar–May)	−87%	IEA
Commodity vessel transits, daily	~135/day (Feb)	7 (21 Aug)	−95%	Kpler
LNG through Hormuz	>10 Bcf/d	0 laden (1 Mar–24 Apr)	−100%	EIA
Gulf crude production	see note	23.9 mb/d (July)	8.3 mb/d shut in	IEA
Gulf crude exports	~20 mb/d (early July)	15 mb/d (July avg)	−2.1 mb/d m/m	IEA
Global observed oil stocks	~8.31 bn bbl (implied)	<7.90 bn bbl	−410 mb	IEA
Global refinery throughput	~85.9 mb/d (implied)	80.9 mb/d (July)	−5.0 mb/d y/y	IEA
North Sea Dated / Brent	$71.13 (implied)	$96.80 (July avg)	+$25.67	IEA
US diesel crack spread	no comparable baseline	$102.20/bbl (record)	—	AGBI
TTF (Europe)	~$10.96 (derived)	$20.80 (14 Aug)	+90%	EIA / GLH
JKM (NE Asia)	~$10.61 (derived)	~$21.75 (14 Aug)	+105%	EIA / GLH
War-risk premium, Hormuz transit	0.25% hull (Feb)	1–3% (peaks 5–10%)	4–12×	Fairway

Notes. "Derived" values are back-calculated from a percentage change stated by the cited source and are not observed prints. Rows from different providers use different definitions and must not be compared across rows. Change columns are computed from the two cells shown in the same row only.

Two things stand out before any analysis begins. The first is the sheer depth of the interruption. A 77% fall in oil transit through the world's most important chokepoint, sustained across a quarter, has no modern precedent. The IEA has put cumulative supply losses above 1.3 billion barrels.

The second is subtler, and it runs through everything that follows: the two most authoritative public agencies covering this waterway publish materially different numbers for it.

Chart 1 data — oil flows through the Strait of Hormuz

Period	mb/d	Basis	Source
4Q 2025	21.6	Petroleum & other liquids	EIA
1Q 2026	not published	—	—
Mar–May 2026	2.7	IEA oil flows, 3-month avg	IEA
2Q 2026	4.9	Petroleum & other liquids	EIA
3Q 2026	not yet published	—	—

No crude-versus-products split is charted because no split on a consistent quarterly basis was obtainable for 2026; plotting one would require inventing the components.

The AIS problem: how much oil is actually moving?

Ask four credible organisations how much oil crossed the Strait of Hormuz this spring and you get four defensible answers that do not reconcile. This is not sloppiness. It is the predictable output of four different measurement systems pointed at a waterway that has become deliberately difficult to observe.

Why the numbers don't agree

Different definitions of "oil." The EIA's 21.6 million b/d baseline covers "petroleum and other liquids" — crude, condensate, natural gas liquids and refined products together. The IEA's ~20 million b/d and S&P Global's split of roughly 15 million b/d of crude plus 5 million b/d of products describe overlapping but not identical quantities. A 1.6 million b/d "discrepancy" between the EIA and the IEA is mostly a category boundary.

Different observation windows. The IEA's 2.7 million b/d is a March–May average. The EIA's 4.9 million b/d is April–June. Those windows overlap by two months but differ at both ends — and the difference matters enormously, because a US–Iran memorandum signed on 17 June produced a measurable recovery in tanker movement that lands inside the EIA's window and outside the IEA's. Neither figure is wrong. They are answers to different questions.

Different vessel definitions. Kpler's seven vessels on 21 August counts commodity vessels. Windward's figures — just under 100 transits in a 24-hour period, roughly a third below baseline — count ships. Container ships, bulkers, naval escorts and support craft are ships but not cargoes of energy. Comparing the two produces nonsense, and it has been done repeatedly in general-interest coverage this year.

The measurement itself is under attack. In the 24 hours following the opening strikes, Windward identified GPS and AIS interference affecting more than 1,100 vessels, clustered across Emirati, Qatari, Omani and Iranian waters. One very large crude carrier had its AIS signal spoofed to six separate locations inside a single day — displaced onto airport runways, into the waters around the Barakah nuclear plant, and onto Iranian land coordinates near Assaluyeh. Tankers are also transiting dark by choice. Windward's own assessment is blunt: AIS has become "increasingly unreliable for monitoring traffic."

This is the crucial point for anyone who builds or buys energy analytics. Vessel-tracking platforms are extraordinary instruments, and in normal conditions their cargo inference is good enough to trade on. But their primary input is a cooperative broadcast — a signal the vessel chooses to send, which a third party can jam or falsify. Under exactly the conditions where the intelligence is most valuable, the sensor degrades. Worse, it degrades asymmetrically: the vessels most likely to go dark are the ones whose movements carry the most information.

The practical consequence is that every publicly quoted Hormuz transit figure this year is a floor, not a measurement. Kpler says so explicitly. Analysts who have treated these prints as observations rather than as lower bounds have systematically underestimated flows — and the corollary, that the gap between reported and actual is itself unknown, is not something a better model can fix. It requires a different sensor.

A model is only as honest as its worst input. In 2026 the worst input was the one everybody assumed was solid.

There is a second, quieter distortion. The interference generates false positives in compliance screening — banks, charterers and insurers receive sanction alerts triggered by spoofed positions. The cost of that noise is real, and it falls on exactly the institutions whose willingness to underwrite a voyage determines whether the voyage happens.

Chart 2 data — what is actually observable

Date	Vessels	Definition	Source
Feb 2026 (monthly avg)	135	Commodity vessels/day	S&P Global
3 Mar 2026	7	Commodity vessels	S&P Global
20 Aug 2026	14	Commodity vessels (implied)	Kpler via Al-Monitor
21 Aug 2026	7	Commodity vessels (4 in / 3 out)	Kpler via Al-Monitor

There is no continuous public daily series on a consistent definition. These are individually sourced observations, not an interpolated line. All values exclude dark transits and are lower bounds.

The crisis beyond crude

Public attention has followed the Brent price. That is the wrong instrument.

Brent traded around $93 a barrel in mid-August — elevated, but well below the $126 peak reached in March and below the $118 print of late April. By the standards of the disruption, crude has behaved almost calmly. The distress has migrated downstream.

On 17 August the US diesel crack spread — the margin between diesel and the crude it is refined from — reached a record $102.20 a barrel. The following day, northwest European diesel carried a premium of nearly $95 over crude, against a Brent price of about $93. Diesel, in other words, was worth roughly twice its feedstock.

The arithmetic behind that is not primarily about the Strait. It is about refineries. Over 20% of the Middle East's 9.6 million b/d of refining capacity is offline; Kpler estimates roughly 2.2 million b/d of downtime in August against about 400,000 b/d a year earlier. Around 700,000 b/d of diesel supply has been lost through Hormuz since March, and a further 700,000 b/d of Russian production since June following strikes on refineries there. Combined, that is about 1.4 million b/d against an internationally traded diesel market of roughly 8.5 million b/d.

Saudi Aramco's Jizan refinery, which had supplied 6–8% of EU and UK diesel imports since April, has been unable to load since 24 July because of Houthi activity on the Red Sea route. That detail deserves emphasis: Jizan exists partly because it sits outside the Gulf. The bypass route acquired its own chokepoint.

The EIA's second-quarter review shows the same story in the export data. US distillate exports averaged 1.56 million b/d, 30% above the five-year average. Jet fuel exports averaged 356,000 b/d — more than double the five-year average, with production 24% above normal as refiners tilted their yields toward the products commanding the premium. Gasoline production, by contrast, ran just 1% above average. Refiners are not running harder across the board. They are running differently, and the market is paying them handsomely to do it.

The world has been watching a crude price while a distillate shortage did the actual economic damage.

This distinction matters operationally, because crude and products fail differently. A crude shortfall is absorbed by inventories, spare production and rerouting — all of which the industry has deployed. A distillate shortfall is a conversion problem: you need the right crude slate, at the right refinery, with the right units running. No amount of crude in floating storage produces diesel if the hydrocracker that would have made it is offline in the Gulf.

Asia's exposure is not evenly distributed

The reflexive assumption is that China, as the largest buyer of Gulf crude, is the most exposed economy. The import data says otherwise.

China takes about 38% of its crude through Hormuz. India takes about 42%, cushioned considerably by Russian barrels arriving by other routes. Japan, South Korea and Taiwan take between 60% and 75%. The most exposed economies in this crisis are three advanced industrial democracies with limited domestic production and no pipeline alternatives whatsoever.

They are also, not coincidentally, among the best prepared. Japan and South Korea maintain substantial government and commercial petroleum reserves and have drawn on joint strategic stockpiles. Exposure and resilience are different variables, and conflating them produces bad risk analysis.

The genuinely acute distress sits further down the income scale. Bangladesh, Pakistan and Singapore depend on Qatar for roughly 40%, 60% and 90% of their LNG imports respectively. These are economies without indigenous gas at scale, without coal switching capacity in Singapore's case, and now bidding for spot cargoes against far better-capitalised European buyers. Several have resorted to demand management that would be politically unthinkable in Tokyo or Seoul: fuel rationing, mandated remote working, odd-even vehicle days, and enforced air-conditioning setpoints.

Chart 3 data — who depends most on energy moving through Hormuz

Economy	Crude via Hormuz	LNG from Qatar
Japan / South Korea / Taiwan	60–75%	n/p
India	42%	n/p
China	38%	~33% of LNG via Hormuz (1H25)
Singapore	n/p	~90%
Pakistan	n/p	~60%
Bangladesh	n/p	~40%

"n/p" means not published in the sources consulted; no value has been estimated to fill a gap. The two columns use different denominators and must not be read against each other. These are pre-crisis structural dependency shares, not current import shares.

Qatar, and why LNG breaks differently

Crude oil is fungible, storable and reroutable. LNG is none of those things to the same degree, and Qatar's position illustrates why with unusual clarity.

Qatar exports roughly 77 million tonnes a year of LNG, second only to the United States, from a single complex at Ras Laffan on the Gulf coast. There is no East–West pipeline equivalent. There is no Fujairah. Every molecule Qatar sells by sea leaves through the Strait of Hormuz, and the country has no meaningful alternative export route.

Between 1 March and 24 April, no laden LNG vessel crossed the Strait at all. QatarEnergy halted production on 2 March and declared force majeure on the 4th. Over 10 Bcf/d of supply — about a fifth of global LNG trade — simply stopped.

The price response was immediate but, revealingly, more moderate than the volume loss implies. By the week ending 24 April, TTF stood at $14.80/MMBtu, up 35% from pre-closure levels, and JKM at $16.02, up 51%. By 14 August those had risen further, to $20.80 and the high $21s respectively. Substantial, but not the catastrophic spike a 20% supply loss might suggest.

Three things absorbed the shock. US LNG exports ran at 17.9 Bcf/d in March, the second-highest month on record, with terminals at 94% of approved capacity. European storage entered the crisis in reasonable shape and stood at 60.4% on 14 August, down 17.7% year-on-year but not critical. And demand destruction did the rest — the rationing and switching described above.

The LNG market absorbed a fifth of its supply disappearing. It did so through flexibility that was built, financed and commissioned years before anyone modelled this scenario.

This is the first place where the article's central question gets a clean answer. Destination-flexible US cargoes, spot-tradable volumes and interconnected European storage are what softened the blow. All three are physical and contractual infrastructure. Analytics helped route the cargoes; they did not create them.

The pipelines that matter more than ever

Two pipelines carried the actual response to this crisis. Understanding what they did — and what they could not do — is the core of the argument that follows.

Saudi Arabia's East–West pipeline runs from the Eastern Province across the peninsula to Yanbu on the Red Sea, with a nameplate capacity of 7 million b/d. Exports from Yanbu rose from about 2 million b/d before the conflict to more than 5 million b/d by early June. That is roughly 3 million b/d of incremental delivered volume — the single largest supply-side response of the crisis.

The UAE's Abu Dhabi Crude Oil Pipeline runs from Habshan to Fujairah on the Gulf of Oman, outside the Strait, with a capacity of about 1.5 million b/d. Emirati exports recovered from 1.9 million b/d in March to 4.3 million b/d in June, around 85% of pre-war levels — though that total includes volumes still moving through Hormuz, not Fujairah alone.

Capacity is not spare capacity. Adding nameplate figures together — 7.0 plus 1.5 — and calling the result 8.5 million b/d of bypass capability is the single most common error in coverage of this crisis. It is wrong twice over. First, the East–West line was already carrying volume before the war; its incremental contribution was about 3 million b/d, not 7. Second, reporting on the Saudi system describes roughly 5 million b/d of crude exports via Yanbu, 700,000–900,000 b/d of products and around 2 million b/d allocated to domestic refining — figures which sum to more than the stated 7 million b/d nameplate. That inconsistency exists in the public record and is flagged here rather than smoothed over. On any reading, available headroom is small.

Chart 4 data — Hormuz flows against what can actually bypass them

Measure	mb/d	Type	Source
Hormuz oil flow, 4Q 2025	21.6	Observed flow	EIA
Saudi East–West pipeline	7.0	Nameplate	Fortune
Saudi Yanbu — incremental exports	~3.0	Observed increment (2.0 → 5.0+)	IEA
UAE ADCOP to Fujairah	1.5	Nameplate, current	The National
UAE ADCOP expansion	~3.0	Nameplate, from 2027 (forecast)	The National
Iraq Kirkuk–Ceyhan	not verified	—	Excluded

These figures are not additive and must not be summed. The Iraqi Mediterranean route is referenced in general coverage but no capacity figure could be verified to primary-source standard, so it is excluded rather than estimated.

Set against 21.6 million b/d of pre-crisis transit, the entire bypass estate — nameplate, not spare — comes to something under 9 million b/d, and the volume actually added when the system was pushed to its limit was closer to 3 million.

The UAE has fast-tracked a second West–East line that will roughly double Fujairah capacity to around 3 million b/d, operational in 2027. That is the correct strategic response. It is also, precisely, a year and a half too late for 2026 — which is the whole point.

Enter the digital oil field

Everything to this point is geopolitics and steel. The question this publication exists to ask is what the industry's analytical apparatus contributed while all of it was happening.

The honest framing is this: if infrastructure cannot be built inside a crisis, can better information materially reduce the damage? The answer is yes, in specific and measurable ways — and the specificity matters far more than the enthusiasm.

Where forecasting genuinely earned its cost

Supply forecasting during this crisis has been unusually difficult, and the reason is structural rather than technical. Models trained on the behaviour of open waterways were asked to predict an intermittently negotiable one, where the governing variable was not weather or congestion but a political decision taken vessel by vessel. Machine learning handles regime continuity well. It handles regime change badly, and 2026 delivered several regime changes in six months.

Where it worked was narrower and more useful: cargo-level ETA prediction under disrupted routing, refinery feedstock gap identification, and inventory depletion projection. These are bounded problems with dense historical data and short horizons, and they are exactly the problems where an operator asks "what do I need to know in the next ten days" rather than "where will Brent be in a quarter."

The datasets required are unglamorous and mostly proprietary: berth-level terminal schedules, pipeline nominations, tank gauge telemetry, refinery unit availability, contractual delivery windows, and vessel positions. Note what dominates that list. It is internal operational data, not market data — and an operator with excellent internal telemetry and mediocre market analytics outperformed the reverse throughout this crisis.

The intelligence stack, and where it fractured

A modern energy-intelligence architecture is conventionally described as a layered stack: satellite imagery and synthetic aperture radar at the base; AIS above it; then weather, port activity, pipeline telemetry, inventory data, commodity pricing, and geopolitical intelligence; with analytics on top and human decision-makers above that.

In 2026 the fracture ran through the AIS layer, and it propagated upward. Cargo inference depends on vessel identity and position. Destination prediction depends on both. Anomaly detection depends on a stable baseline of normal behaviour. When more than 1,100 vessels are being spoofed inside a day and an unknown number are running dark by choice, every layer above AIS inherits the corruption.

Synthetic aperture radar is the correct compensating sensor. It sees through cloud and darkness, it detects hulls rather than broadcasts, and it cannot be spoofed by a transponder. It is also expensive, revisit-limited, and it tells you a ship is there without telling you what is in it. The operators who coped best this year were those who treated SAR and AIS as mutually validating rather than treating either as ground truth — and who, crucially, had established that discipline before they needed it.

The instrument that failed was not the analytics. It was the sensor everybody had quietly stopped questioning.

The digital twin of an energy crisis

The most substantive digital opportunity in a disruption of this kind is not forecasting. It is integrated re-optimisation across an asset chain that is normally managed in separate silos.

The scenario below is an illustrative operational example. It is constructed to be technically plausible and is not a description of any particular company's systems.

Consider a Gulf producer facing a sudden reduction in available tanker capacity, because war-risk premiums have priced several charterers out and two vessels have declined to transit. In a siloed organisation, that fact reaches production planning days later, via the trading desk, after loading slots have already been missed.

An integrated twin evaluates the whole chain simultaneously. Given reduced liftings, how long until crude storage reaches tank tops, and which fields must be choked back first to protect the ones with the highest restart cost? Which of those barrels can be diverted to the bypass pipeline, and does the pipeline have hydraulic headroom at that flow rate given the pumps currently in service? If terminal loading shifts to the Red Sea coast, does the receiving refinery's crude slate still work, or does the change in gravity and sulphur force a unit reconfiguration? Which maintenance windows must move, and does deferring them create an unacceptable failure probability on equipment now running at maximum utilisation?

Those questions are answerable. They are also, individually, questions the industry has been able to answer for years. What the twin adds is answering them together, in hours rather than weeks, with the constraints of each domain visible to the others. In a crisis measured in days, that compression is worth real money.

It is worth being precise about the limit, though. Every option the twin generates is an option that already exists physically. The twin finds the best available reconfiguration of the assets you have. It does not add assets.

Why predictive maintenance became a strategic function

One consequence of this crisis has been under-discussed, and it is the most operationally actionable finding in this article.

When the Saudi East–West line went from roughly 2 million b/d of export throughput to over 5 million, and when Fujairah became the primary Emirati outlet, those systems moved from comfortable utilisation to running near their limits — and they stayed there for months. The same is true of the export terminals, loading arms, tank farms and pumping stations behind them.

Equipment operating near maximum utilisation fails differently than equipment with slack. Duty cycles lengthen, redundancy that existed on paper gets consumed, maintenance windows become nearly impossible to schedule because there is no spare train to take the load, and the consequence of any single failure escalates sharply because there is no alternative route behind it.

The business case for predictive maintenance is conventionally written in terms of avoided downtime cost at normal throughput. That understates it badly under these conditions. When a bypass pipeline is carrying volume that has nowhere else to go, the cost of an unplanned pump failure is not a maintenance event — it is a national export interruption.

The verifiable answer to "does geopolitical disruption increase the value of predictive maintenance" is therefore yes, and the mechanism is specific: disruption concentrates flow onto fewer assets, and concentration converts routine reliability engineering into critical-path risk management. Pumps, compressors, valves, loading systems and tankage on bypass routes should be re-tiered accordingly.

Cyber risk: the other chokepoint

The industry's response to a physical blockade has been to lean harder on remote operations, cloud analytics, satellite data feeds and interconnected terminal systems. Each of those is a dependency, and dependencies concentrate risk in the same way pipelines do.

The GPS and AIS interference documented across the Gulf this year is instructive precisely because it is not a cyberattack on an operator. Nobody breached a control system. An external signal environment was degraded, and every downstream system that trusted it degraded with it. That is the shape of the risk worth planning for: not a dramatic intrusion, but the quiet corruption of an input that a dozen automated processes assume is reliable.

Three exposures follow directly. Data integrity — analytics and compliance screening consuming a feed that is being actively falsified. Availability — remote operations centres and cloud-hosted optimisation depending on connectivity across a contested region. And convergence — OT and IT networks that were separated a decade ago and have been progressively bridged since, often for the sake of exactly the analytics this article describes.

The defensive posture is unglamorous and well understood: validate positional data against an independent sensor before it drives a decision; maintain a degraded mode in which operations continue without cloud analytics; keep OT segmentation intact even where it makes the data pipeline harder to build; and treat the accuracy of a third-party feed as an assumption to be tested rather than a fact to be consumed.

AI cannot build a pipeline

It is worth stating the strongest version of the optimistic case before testing it, because the optimistic case is not silly.

During this crisis, analytical systems have done genuinely valuable work. They have told operators where cargoes are when transponders lie. They have identified feedstock gaps at specific refineries before those gaps became shutdowns. They have re-sequenced tanker schedules around war-risk pricing that changes weekly. They have modelled inventory depletion closely enough to inform a 400 million barrel coordinated release. They have compressed decisions that took weeks into decisions that take hours. In a market where Brent moved an average of $4 a day through the second quarter — against $1 in the same months of 2025 — the value of deciding faster is not theoretical.

All of that is real. None of it produced a barrel.

The supply response to this crisis came from four sources, and they can be named precisely. Saudi Arabia moved roughly 3 million b/d of incremental export through a pipeline built in the 1980s. The UAE moved volume through a line commissioned in 2012. The United States exported a record 13.1 million barrels of crude and products in May, about 25% above the prior year, out of terminals financed over the preceding decade. And 32 governments released 400 million barrels of oil that had been sitting in salt caverns and tank farms, in some cases for forty years.

Every one of those is a physical asset that existed before the crisis began, built by a decision taken years or decades earlier, by people who could not have known what it would be used for.

Digital intelligence changes what you do with the infrastructure you have. It has no opinion about how much infrastructure you have.

This is not an argument against analytics. It is an argument about sequencing, and it has a sharp practical edge for capital allocation. An operator choosing between a data platform and a pipeline debottlenecking project has been given an unusually clear natural experiment this year, and the answer is not the one a technology vendor would prefer.

But the honest conclusion is more interesting than a simple ranking, and the evidence supports it in both directions.

Consider what the physical response actually required. Saudi Arabia did not simply open a valve. Tripling Yanbu throughput meant reallocating production across fields, re-sequencing the pipeline's pumping schedule, restaging tankage, rebuilding a loading programme at a terminal that had been running at a third of the volume, and re-matching crude grades to buyers who had been receiving different barrels. The physical capacity made the response possible. Coordination made it fast.

The counterfactual is worth sitting with: an operator with the same pipeline and materially worse operational visibility would have taken longer to reach the same throughput, and in a crisis measured in weeks, "slower" and "smaller" are the same thing.

So the finding is not that physical infrastructure beats digital intelligence. It is a relationship with a clear direction of dependency:

• Physical sets the ceiling. No algorithm moves a barrel through a pipeline that does not exist, or into a tank that is full. The maximum possible response is fixed entirely by steel, geography and spare capacity — all of which have lead times measured in years.
• Digital determines how much of it you reach. Between the theoretical maximum and what an organisation actually achieves under stress sits coordination speed, visibility and decision quality. That gap is where analytics pays — and in a fast crisis it is a wide gap.
• The order cannot be reversed. Excellent analytics on thin infrastructure produces a very well-informed account of a shortage. Adequate analytics on redundant infrastructure produces barrels. Both are worth having; only one of them is optional in the short run.

Is infrastructure redundancy ultimately more valuable than digital intelligence during an extreme supply shock? On the 2026 evidence: yes, decisively, but the comparison is somewhat false. Redundancy is what you buy in the decade before a crisis. Intelligence is what you deploy in the week of one. An operator that neglects the first will not be rescued by the second — and an operator that has the first but neglects the second will realise less of it than its competitors.

The money, and who is paying it

The costs of this crisis have not landed where the headlines suggest.

War-risk insurance is the clearest transmission mechanism. Before the crisis, war-risk cover on a Hormuz transit ran at about 0.25% of hull value — roughly $375,000 for a $150 million tanker. Within a week of escalation, quoted rates moved into the 1–3% range, or $1.5 million to $4.5 million per transit, with peaks quoted at 5–10% for vessels with higher risk profiles. That is a four- to twelve-fold increase in a fixed per-voyage cost, and it is charged whether or not anything happens.

The effect of that pricing is not primarily financial. It is selective. A premium of several million dollars per transit is survivable for a large cargo of high-value product and prohibitive for a marginal one, which means insurance pricing has quietly become an allocation mechanism deciding which cargoes move. Container shipping shows the same pass-through in a more visible form: Hapag-Lloyd applied a war-risk surcharge of up to $3,500 per container on Gulf-touching shipments.

Refining margins are where the money has actually been made. Atlantic Basin margins reached all-time highs in July. Distillate and jet fuel crack spreads ran at more than double year-earlier levels through the second quarter, and gasoline margins were up 60%. A refiner with secure feedstock access and the right conversion units has had an extraordinary year — which is why the strategically valuable asset in 2026 turned out to be a hydrocracker outside the Gulf, not a barrel of crude inside it.

And the household cost is arriving on a lag. US pump prices are running nearly a dollar a gallon above a year ago. That is the diesel crack and the freight premium and the insurance surcharge, reaching the end of the chain.

Seven indicators to watch

Indicator	Latest	Why it matters
1. Daily commodity transits	7 (21 Aug)	The cleanest high-frequency signal, provided it is read as a floor. Recovery above 30–40/day would be the first credible normalisation signal.
2. VLCC and LNG carrier presence	0 (21 Aug)	More informative than the headline count. Large vessels return only when insurers and owners genuinely price the risk down.
3. Diesel crack spread	$102.20 (record)	The real distress gauge in this crisis. It moved before crude and further than crude.
4. Global observed oil stocks	<7.90 bn bbl	The buffer. At the July draw rate the system has months, not years, and emergency stocks now need replacing.
5. Yanbu export volumes	>5 mb/d	The best measure of how hard the physical bypass is being pushed, and how little headroom remains.
6. War-risk premium, % of hull	1–3%	A market-priced probability of attack, updating faster than any official assessment.
7. JKM–TTF spread	~$0.95 (derived)	Determines whether flexible US cargoes sail east or west — the mechanism actually rationing global LNG.

The bigger lesson

Is the future of energy security digital or physical?

The 2026 evidence gives an answer that is less balanced than the diplomatic version of it. Physical infrastructure determined what was possible. Digital intelligence determined how quickly and completely the possible was reached. Those are not equal contributions, and pretending otherwise misreads the year.

Three million barrels a day moved because a pipeline crossed Saudi Arabia. Four hundred million barrels became available because governments had spent four decades filling caverns. A fifth of the world's LNG supply vanished and the price roughly doubled rather than quintupling, because American export terminals and European storage had been built for other reasons entirely.

Against that, the analytical apparatus delivered speed, visibility and coordination — real contributions, worth real money in a market moving $4 a day, and the difference between a good response and a slow one. But it delivered them within limits it had no ability to move.

There is a final observation, and it is the one that should trouble the industry most. The single most consequential analytical failure of this crisis was not a model that predicted badly. It was a sensor everybody trusted that turned out to be adversarial. AIS did not break because the mathematics was wrong. It broke because it was designed as a cooperative safety system in a world that had stopped cooperating.

That is worth carrying forward, because it generalises. Every layer of the modern energy intelligence stack rests on an assumption about a data source that was true when the system was designed. Under stress, those assumptions are exactly what fail — and no amount of sophistication above a corrupted input recovers the signal.

Build the pipeline. Then build the twin. And check, regularly, whether the instruments are still telling the truth.

References

1. US Energy Information Administration — Short-Term Energy Outlook, August 2026; chokepoint transit data as reported by Rigzone, 12 August 2026.
2. US Energy Information Administration — Short-Term Energy Outlook: Global oil markets, August 2026.
3. International Energy Agency — Oil Market Report, August 2026.
4. International Energy Agency — How global oil supplies have readjusted to help fill the huge gap left by the Strait of Hormuz shock.
5. Al-Monitor — Ships passing through Hormuz hover in single digits, data shows, August 2026, citing Kpler.
6. International Energy Agency — IEA Member countries to carry out largest ever oil stock release, 11 March 2026.
7. US Energy Information Administration — International LNG prices rise amid Strait of Hormuz closure, April 2026.
8. OPEC — Monthly Oil Market Report, August 2026, as reported by Reuters via Investing.com, 12 August 2026.
9. S&P Global Commodity Insights — Factbox: Hormuz oil flows still at a standstill despite US insurance pledge, 4 March 2026.
10. AGBI — Gulf refinery disruption sends diesel margins to record highs, August 2026.
11. Global LNG Hub — Natural gas prices weekly update — JKM, TTF and Henry Hub, 17 August 2026.
12. Fairway — War Risk Insurance 2026: Why Hormuz Transits Now Cost $1.5M–$4.5M More Per Voyage.
13. Al Jazeera — Iranian parliament advances plans for Hormuz service fees, 24 August 2026.
14. US Energy Information Administration — Petroleum markets responded to disruptions in the Middle East in the second quarter, 2026.
15. Windward — Widespread GPS Jamming Hits 1,000-plus Ships in the Middle East, 2026.
16. Gulf International Forum — Hormuz Disruptions and Asia's Energy Resilience, 2026.
17. CSIS — The Battle for Hormuz Will Reshape the Global LNG Market, 2026.
18. The National — UAE's West-East pipeline expansion to become operational in 2027, 15 May 2026.
19. Fortune — Saudi Arabia East-West oil pipeline: Strait of Hormuz bypass, 28 March 2026.
20. CNN — US gas prices up nearly a dollar from a year ago as Hormuz traffic remains low, 21 August 2026.

Data notes

Units. Oil in million barrels per day (mb/d) as published. LNG in Bcf/d or mtpa as published — no conversion between the two has been performed. Derived values. Four figures are back-calculated rather than printed and are labelled at the point of use: pre-crisis TTF and JKM, from stated percentage changes; the 20 August vessel count of 14, from "a 50% decrease from the previous day"; and the JKM–TTF spread, by subtraction. Missing data. No value in this article has been interpolated, smoothed or estimated to complete a series. Conflicting sources. Where two credible bodies disagree, both figures are printed with their definitions and windows shown. Paywalled data. Kpler and Vortexa proprietary datasets were not accessible; all Kpler figures are drawn from public reporting that cites Kpler. Not attempted. A days-of-inventory-cover estimate was considered and dropped, because the IEA, EIA and OPEC imply materially different supply deficits and the resulting range was too wide to publish as a single figure.

This is a fast-moving situation. Every figure carries a date and a source; readers should verify against the primary source before relying on any number for a commercial decision. Forward-looking figures are labelled as forecasts and are not observations. Corrections and source challenges are welcome.