In March, across this publication's first two editions, I argued that the market was underpricing both the duration and the severity of this war, on the grounds that no exit from it existed. That much has held. What sat underneath it has not: a ladder, escalation running in one direction, each rung leading to the next.

Five months on, I am leaving it.

Not because the March version was wrong for March — it described its period accurately, and a framework that fits its moment and then stops fitting has not failed, it has been overtaken. The problem is that a ladder running one way should by now have arrived somewhere, and this one has not. What the record shows instead is something that keeps not happening. There is a specific class of target in this war that everyone has threatened, that each has held both the capability and the stated intention to strike, and that none has struck. Once is coincidence. Across 5 months, through every phase of the war, by both sides, it stops being an absence and becomes a structure.

I am calling it the Consequence Ceiling.

What follows sets out what the ceiling bounds, why it holds, how little room is left beneath it, and what a war bounded at both ends actually implies — which is not what the register almost everyone is using would lead you to expect.

The memorandum that was supposed to end this war contained two incompatible terms and never chose between them.

Iran retained its claim to control transit through the Strait and to levy tolls on it after 60 days, whilst the United States undertook to guarantee safe passage through the same water. Both cannot be true. A guarantee of passage is an assertion of authority over a waterway; a toll is an assertion of authority over the same waterway by someone else. The document contained both and resolved neither — and that is not a drafting failure. It is what a settlement looks like when the parties want an agreement and the terms of one do not exist.

It died accordingly, and did so without anyone actually breaching it. Suspended within a day of formal signature when the Lebanon tripwire fired, it was buried 3 weeks later when Iran enforced the claim it had never surrendered — striking 3 commercial vessels in the Strait over 7–9 July, including the Qatari linked LNG carrier Al-Rekayyat. The American half unwound in the same window with the Treasury's revocation of the 60 day oil sales waiver the whole structure rested on. Each party executed the position it had written into the text. The document was the incompatibility.

The reason it could not have been drafted any other way is that neither party’s war aim is material — and neither is still open. Both resolved months ago, which is what the negotiating record keeps failing to absorb.

Iran's objective was to establish that attacking it is unaffordable, and that proposition has been demonstrated by the attacker's own institutions rather than by Iranian claim. Interceptor inventories are classified and the public estimates do not reconcile, so nobody outside the Pentagon knows the number. The arithmetic underneath it is not in dispute. More than 1,200 Patriot interceptors had been expended by late April at over $4m each, per internal Pentagon estimates and congressional officials, against production of roughly 600 a year and an Iranian ballistic missile inventory believed to exceed 1,000 — before drones are counted, and against close to 9,000 projectiles of all types fired since February. THAAD interceptors are built at 11 or 12 a year; this war has consumed more than a decade of that line.

And in the last week of July the largest planned offensive of the war was cancelled — not on diplomacy, law, allied pressure or casualties, but on inventory. A superpower cancelling its own offensive on interceptor arithmetic is Iran's war aim, proven, in the adversary's own staffing.

The American objective was to establish that coercion works, and it has resolved in the negative on the same authority. Two official US assessments reported inside a single week describe the campaign producing the inverse of both its stated aims: a senior official telling the New York Times that recent strikes have had the opposite effect to the one intended, keeping Iran cohesive and allowing its leadership to direct attention outward; and an intelligence assessment holding that the new Supreme Leader is substantially more interested in acquiring a weapon than his father was, with Iranian officials who always argued that only a weapon deters attack now believing themselves vindicated.

So the war is not being fought over an unresolved question, but over an unacknowledged answer. Such acknowledgement is the one term neither party can put into a document.

Iran cannot stop, because a demonstration only deters once it has been received. The proof exists and the strikes continue, which means the proof has not yet done the work proofs are for, and so the advertisement has to keep running. In similar fashion, the United States cannot acknowledge, because acknowledgement is not the price of a settlement but the substance of the defeat.

The clearest evidence that the object is not material is that Iran has stopped naming one. In March, Tehran published three conditions: recognition of enrichment rights, reparations, guarantees against future attack. It no longer offers terms at all. Across 4 months and five mediators — Qatar, Oman, Pakistan, Egypt, Turkey — the line has been functionally unchanged, and it is not a position but a refusal: no negotiation while strikes continue, regardless of what is on the table. Four named Iranian officials denied inside 8 days that talks were occurring. One dismissed a US backed Qatari proposal offering a 10 day halt in exchange for reopening two shipping lanes with "don't even think about it." The Speaker of Parliament's formulation was "all or none." And the one asset Tehran might have traded has been reclassified out of tradability: a deputy foreign minister now describes the Strait as part of the national security of the Islamic Republic, a capability Iran cannot under any circumstances abandon, and says it will never return to its pre-war condition. A bargaining chip can be traded; a declared national security asset cannot.

A party with demands is a party inside a bargaining process. One that has declined to state any, while refusing material offers as they arrive, is telling you that what it wants cannot be written into a clause.

Then there is the mechanism Iran built to force the ending, which is producing the opposite of the ending.

The doctrine was coherent and it was never hidden: strike the states that host American forces rather than the forces alone, make basing the United States more expensive than the protection is worth, and let the hosts do the work of pressing Washington toward a ceasefire. Iran has executed it consistently — Kuwait's power and water, the Jordanian airbases, Al Udeid, Bahrain, Oman's Duqm.

It has produced the reverse on both counts. Four senior Gulf officials told the Times of Israel that the strikes appear to have had the opposite effect; the UAE, Saudi Arabia, Bahrain and Qatar are now privately pressing for the campaign to continue rather than end, and the Associated Press independently reports Riyadh and Abu Dhabi lobbying the President to keep going. The President noticed it himself, observing that Iran was hitting countries that had been neutral, and that both they and he were surprised. The conversion has gone further than lobbying. Bahrain and Kuwait each sent fighter aircraft into Iran to strike drone and missile storage sites, under Emirati targeting and defensive air cover — the first known direct military retaliation by either state, reported by the Wall Street Journal on a single source with Kuwait publicly denying it, and worth stating with that caveat attached. On 29 July Saudi and American aircraft struck Iranian aligned militia sites across eastern Iraq in a jointly coordinated operation Riyadh confirmed. States that began this war absorbing strikes are now initiating them.

Which gives the leverage its proper name. What Iran holds is negative power: it can stop things and it cannot cause things. Chokepoint denial imposes costs on everybody and produces compliance from nobody, and it has a documented failure mode — it assembles coalitions against the party holding it rather than followers behind it. Hezbollah is degraded past 3,700 dead in Lebanon, the Iraqi militias are quiet, and the Houthis prosecute their own war against Saudi Arabia on their own timetable rather than Tehran's. Iran has proved that attacking it is unaffordable, and the proof has assembled a larger set of parties attacking it. Winning the argument and improving the position are different achievements, and only the first has occurred.

There is a second constraint underneath, independent of the first: neither delegation can commit its principal. Iran's Supreme Leader has not been seen publicly since 28 February, and the Foreign Minister states publicly that the Supreme Leader holds final authority on war and peace and that he has never met him. On the American side the lead Iran negotiator described an offer of enriched uranium for medical purposes; the President said in the same news cycle that it had perhaps not been very serious and that he was giving them nothing. One delegation cannot reach its principal; the other cannot rely on what its principal said 8 hours ago.

The extension is the finding. Every principal party is constrained, and the constraints differ in duration rather than in kind. The United States cannot stop without conceding the first unambiguous failure of an initiated coercive campaign against a mid tier adversary, Iran cannot stop because its deterrent is an ongoing demonstration rather than a completed one, China cannot hold its optimal position, which required cheap oil and American exhaustion simultaneously and can no longer have both, the Gulf states have become belligerents in a war they hold no instrument to end and Russia collects a revenue dividend while 42% of its own refining capacity sits idle and it imports petrol. So the useful question is not who is winning; it is whose constraint becomes unsustainable first — and unlike the first question, that one is measurable.

The strongest objection to all of this is present tense rather than historical. The Associated Press, citing two regional officials, reports that mediators led by Qatar and Pakistan have made progress toward returning the parties to talks and restoring the collapsed framework, one describing that progress as significant, with work under way between Iran and Oman on a vessel passage mechanism. Tehran's public denial is narrower than it sounds — no direct negotiations with the United States — and narrow denials running alongside active indirect channels are what the late stage of a mediated settlement looks like. If a mechanism emerges that assigns authority over the Strait to one party rather than to two, the reading above is wrong and the solution set was never empty.

None of which is a claim that the parties are equally placed. Iran denied the United States its war aim and the asymmetry is stark. What the frame adds is the constraint running against the party holding the advantage, which is the part that tells you when this changes rather than only who is ahead.

There is a specific set of targets in this war that every party has threatened but none has executed, and all have manoeuvred around. Abqaiq's stabilisation capacity, Ras Tanura, Kharg, South Pars, Gulf desalination at population scale… 5 months of maximal rhetoric, and not one of them has been struck.

Start with the threats that were specific enough to be checked, because a general warning proves nothing and a formula with a named trigger proves a great deal.

Trump told Iran that if Qatar's liquefied natural gas were attacked again, the United States would destroy the South Pars gas field — adding, on the record, that he did not want to authorise that level of destruction because of what it would mean for Iran's future. Ras Laffan was struck and South Pars was not destroyed.

On 22 July he published a mechanical formula rather than an ultimatum: from that point forward, any Iranian attack on a ship in the Strait would be answered by the destruction of one bridge or power plant, explicitly including those in or beside Tehran. Iran has struck ships since and Tehran has not been hit.

IRGC Aerospace Commander Mousavi, responding directly through Tasnim, said that if Iranian bridges or power plants were attacked, Iran would cut off electricity to America's Gulf allies. Iran has executed that against exactly one of them: Kuwait's Shuaiba power and desalination complex was struck on 17 July, the Al-Subiya power and water plant on 19 July — and Kuwait is the one Gulf state whose infrastructure Iran can destroy without producing a global energy event. Catastrophic locally, structurally invisible everywhere else. Iran's own Khatam al-Anbiya Joint Command had promised that American strikes on Iranian infrastructure would bring the destruction of every piece of infrastructure in the entire region as if it had never existed. That has not happened either.

Two instances need separating by more than mechanism — they sit a week apart, and only one of them is still moving.

The earlier one is now settled. On 23 July the President told Axios he was considering a massive attack, bigger than ever before, that he was close to a decision and that everything was set; two other US officials said the same day that no decision had been taken and no orders had gone to the military. The offensive was held, and a week on it still has not come. But the documented reason it was held was inventory rather than consequence — General Dan Caine, Chairman of the Joint Chiefs, advised that major operations, while feasible, would exhaust Central Command's interceptors — and a party that declines to take a rung is not the same object as a party that cannot presently afford to take it. Only the first is evidence of a boundary.

This week supplies the live one. On 28 July the IRGC fired ballistic missiles from Iran at US forces in Jordan, at Muwaffaq Salti and at CENTCOM's regional headquarters per an IRGC statement carried by IRNA, with Jordan's military reporting 5 intercepted and no casualties. The American answer arrived the next day and went to eastern Iraq: militia logistics and weapons sites, struck jointly with Saudi aircraft, in a theatre holding no Iranian strategic air defence to penetrate. On that same Wednesday morning the President told Fox News by telephone that "we're going to beat the fucking shit out of them," that "we'll be hitting them hard," and that "they're going to get a beating," adding that he had personally reviewed video of American personnel calling out coordinates.

Stated intent at its maximum, on the day the munitions landed somewhere else. Iran struck American forces directly; the United States answered a third country's militias, with a partner sharing the sortie load. The pause ordered in late July was a pause on strikes against Iran specifically, and it has not been broken — no American munition has been confirmed on Iranian soil since it began.

Held rather than counted, and dated precisely because the dating is the interesting part: on 28 July, the same day Katz said publicly that Washington was restraining Israel, a senior Israeli security official told Channel 15 that the United States was preparing something bigger than anything seen so far. An anonymous allied characterisation of American intentions is a class of claim that has been wrong before, and the two propositions cannot both hold. They resolve on the same event — the next confirmed American strike on Iranian soil, and what it hits.

The precedent that does the real work is March, and it matters because it is not restraint. It is a full execution followed by a withdrawal.

Iran did not hold its Gulf energy card in reserve. On 2 March Iranian drones struck Ras Laffan, which ships roughly a fifth of the world's LNG; QatarEnergy halted production and declared force majeure. In the same window Saudi Aramco shut the Ras Tanura refinery after debris from intercepted drones started a fire. After the Israeli strike on Iranian oil facilities on 7 March the IRGC vowed retaliation against Gulf energy assets and warned that a wider regional war could take crude to $200. Then, following the US and Israeli strike on the South Pars and Asaluyeh complex on 19 March — the first direct attack on Iran's upstream energy infrastructure of the war — the IRGC issued evacuation warnings naming facilities as direct and legitimate targets: the SAMREF refinery and the Jubail petrochemical complex in Saudi Arabia, the Al Hosn gas field in the UAE, the Mesaieed petrochemical complex and Ras Laffan in Qatar. Saudi defences intercepted 20 drones aimed at the Shaybah field, which produces around 1m bpd. Habshan and the Bab oil field in the UAE were engaged and intercepted, with debris fires forcing operations to halt.

Two deterrents landed, one public and one private. Trump's South Pars threat was the public one. The private one, per Reuters, was Saudi Foreign Minister Faisal bin Farhan telling Araghchi directly that continued Iranian strikes on Gulf energy infrastructure could force Riyadh to grant US forces access to Saudi military bases. Days later President Pezeshkian publicly apologised to the Gulf states.

Iran has not returned to Ras Laffan, Ras Tanura, Abqaiq or Shaybah since. It went to Kuwait instead. The card was played, it drew a response Tehran was not prepared to absorb, and the campaign moved to the one target set that imposes cost without triggering the answer.

Which brings the question of whether this is a boundary the parties understand or merely one an analyst has drawn around 5 months of coincidence. On 28 July Israel's defence minister answered it on the record, at a televised defence conference. Israel very much wants to strike Iran's energy targets, he said, and the United States is not approving it at the moment, because of the concern that Iran would retaliate against neighbouring countries and cause a global oil crisis. The sentence most relays dropped is the one that matters: the United States has interests in Iran that go beyond Israel's own, including the Strait of Hormuz and global energy markets. He added that the IDF has been instructed to hold readiness to strike Iran alone, having done so twice already.

That is the first admission of the boundary by a party subject to it, and the reason is named.

None of which describes a war that has stopped escalating, and the strongest objection to the whole reading is that the mechanism for taking the rung is already running rather than waiting on a decision. Trump's formula and Mousavi's are now pointed at one another and neither carries a stated exit: ship struck, Iranian power plant destroyed, Gulf power destroyed. Iran has begun executing its half. Both are formulas rather than choices, which means the next escalation may not be a decision anyone is observed making — it may simply be the next turn of a loop, with each turn moving the target set one rung up. Beneath the boundary the war is being fought without restraint of any kind: over 11,000 targets struck inside Iran, roughly 3,468 Iranian dead by Tehran's own count and more than 3,700 in Lebanon with 1.3m displaced, Kuwait's water and power.

So the claim is not that escalation has ceased but narrower and stranger than that. A specific rung has been approached repeatedly, threatened explicitly, and never taken — by either side, across 5 months, through every phase of the war including the ones in which each party had both the capability and the stated intention.

Everything above is record. What follows is the account that best fits it, and that is worth more than the record precisely because it predicts things it cannot.

The mechanism is not deterrence, and calling it deterrence obscures what is actually restraining these parties. In deterrence, what stops you is what the adversary does in return; the calculation runs through their response, and here the calculation runs through your own. What restrains each party at the terminal class is not retaliation but the effect of its own strike on a system it has to carry on operating inside.

The terminal targets are all shared substrate. Gulf hydrocarbon processing does not price one economy's energy; it prices everybody's, including the buyers whose custom Iran depends on and the allies whose stability Washington is nominally defending. Destroying Abqaiq's stabilisation capacity or Ras Tanura would not transfer a cost from one belligerent to another. It would remove a facility both belligerents' positions currently rest on.

Iran's case is the sharper of the two because its dependency is singular. An Iranian strike that took Gulf processing out at scale would land its heaviest blow on the only customer Iran has — a state whose commercial exposure to the other side of this war runs an order of magnitude larger than its exposure to Iran, and whose conduct throughout has tracked the protection of a substrate rather than the choosing of a side.

The reciprocal half is Iran's own export architecture. Kharg handles roughly 90% of Iranian crude exports and sits 25km off a hostile coast; South Pars is the largest gas deposit on earth. Both are reachable, and both are intact. What that produces is not a balance of threats but a mutual hostage in which each party's instrument of coercion is also its own exposure — and the arrangement functions because Iran still has something to lose. That last clause is the load bearing one, and it is why Kharg is the most consequential open item in this war rather than merely the most dramatic. Destroying it removes the hostage and leaves the grievance intact — a state with no exports left has no remaining reason to preserve anyone else's. Occupation is not the softer version of that. Taking it is the easy part; holding it means defending a fixed position across contested water, against the same cost asymmetry that has consumed most of a superpower's interceptor stocks in 5 months.

Desalination is where the mechanism becomes fully legible.

Kuwait draws roughly 90% of its drinking water from desalination and holds no exploitable natural freshwater reserves; Saudi Arabia, the UAE, Qatar and Bahrain carry structurally similar dependencies. Destroying that capacity at scale kills, and it kills quickly — on a weekly timescale, across populations rather than balance sheets, with no strategic reserve of drinking water and no overland corridor that substitutes for one. That is the first thing to say about it and the reason it sits where it does on the list.

The second thing is who dies. The Gulf states are between 70% and 88% non-national by population, and the migrant workforce is the least insulated part of it — housed in labour accommodation, furthest from any private supply, first to lose access when a system degrades. So a strike on Gulf desalination is simultaneously a mass casualty event and a domestic emergency in a dozen states that are not party to this war, among them two nuclear powers and the mediator who brokered the memorandum. It then becomes an economic event in those same states, because the labour system that collapses is the one their remittance flows run through, and the workers who survive it go home to economies built on the money they were sending.

That is what makes the target class categorically different rather than merely severe. Refineries, terminals, bridges and grids degrade an economy over quarters, which is why both sides have been willing to strike them. Desalination transmits a Gulf catastrophe outward, in weeks, to populations and treasuries neither belligerent is at war with — including, in Iran's case, the states whose goodwill it has spent 5 months trying to keep.

Both belligerents have already touched the category at its edges. American strikes reached a water station at Mahshahr in Khuzestan; Iranian strikes took Kuwait's Shuaiba power and desalination complex on 17 July and the Al-Subiya power and water plant 2 days later. On the same day as Shuaiba, Iran's own Energy Ministry acknowledged attacks on Iranian power infrastructure for the first time and asked southern provinces to conserve electricity in extreme summer heat. The floor is being approached from both directions by parties who each now know what it feels like from underneath.

The rival explanation deserves stating in its strongest form, because it would leave the pattern standing and this account in ruins: nobody has struck the terminal class because nobody has been able to.

On the Iranian side that does not survive March. Iran reached Ras Laffan on 2 March and shut a fifth of the world's LNG supply for a period; debris from its drones started the fire that closed Ras Tanura in the same window; it put 20 drones over Shaybah and reached Habshan and the Bab field. Reach was demonstrated, at the exact facilities in question, and then withdrawn from after Riyadh's private warning and Washington's public one. A party that cannot strike does not apologise for having struck.

On the American side reach was never the question. The United States and Israel hit South Pars and Asaluyeh on 19 March and again on 11 July, and neither strike destroyed it, against a standing presidential threat to do exactly that. But here the rival is genuinely live, because inventory has become a constraint in a way it was not in March. So the evidence differs in kind: Iran's restraint is established behaviourally, by a campaign played and abandoned, while America's rests on testimony. Behaviour is the stronger evidence, and the asymmetry is not being averaged away.

The boundary bounds the tail. The question it raises is how much room remains underneath, and the answer is less than the geography suggests, because the region's redundancy was built in sequence and is failing in the same order it was built.

Each layer was designed as the answer to the failure of the one before it. Hormuz was the route. The Petroline was Hormuz's insurance. Bab el-Mandeb is the Petroline's exit. 5 months in, all three have been struck, gated or embargoed — and the mechanism running through them is that a bypass only matters once it is loaded, and a loaded bypass is worth attacking.

Start with the layer that was never redundant to begin with. The Strait runs at roughly 20 to 34 crossings a day against 120 to 147 before the war, hitting its lowest daily count since early May in the week to 22 July. It is not physically shut. Iran administers a declared closure through a named state regulator, the Persian Gulf Strait Authority, which issues permits and levies tolls, and enforces a closure it cannot physically impose by making the open lane lethal — striking the Cyprus flagged GFS Galaxy on 11 July on the very southern corridor Oman was negotiating to reopen. Since 14 July an American naval blockade of Iranian ports has been overlaid on the same water, all flags, more than 20 warships, under a unilateral claim to be the guardian of Hormuz. Two states assert incompatible jurisdiction over one waterway and the International Maritime Organisation has ruled that neither has a legal basis for mandatory tolls.

The exits from that arrangement closed rather than opened in the last fortnight. Tehran rejected an Omani proposal to let a third country de-mine the Strait, which matters more than any fee question because it means no reopening can be executed by anyone other than Iran, on Iran's schedule. It then ruled out Oman's plan for regional joint management with voluntary fees altogether. And the reclassification of the waterway itself as a national security asset rather than a tradable one, set out above, removes the last structure a negotiated reopening could have been built on.

The mines compound this in a way no negotiation reaches. Iran laid them in a covert pattern designed to be unfindable by adversary clearance assets, and by construction cannot reliably locate them itself — so reopening requires either extended sweeps under war risk conditions or Iranian sweeps that admit the pattern and still miss some of it. The same property now cuts against Iran's own assertions: the IRGC claims the southern Omani route is mined, and a state that cannot find its own mines is a state whose mining claims are unverifiable by its own testimony.

The second layer was built for precisely this contingency. The Petroline runs 1,201km from Abqaiq to Yanbu and was constructed during the Iran–Iraq war as explicit insurance against a hostile power closing Hormuz. It has been carrying up to 4m bpd westward against roughly 973,000 bpd a year ago — the relief valve absorbing the closure, and the infrastructure every projection of Gulf exports bypassing Hormuz depends on. On 27 July drones struck Abqaiq and the nearby East-West Pumping Station; Aramco suspended operations at Abqaiq and began emergency flaring at several production sites. The Houthi military spokesman stated that targets for the supply and transport of crude from eastern Saudi Arabia to Yanbu had been struck, and separately named the pipeline itself. Origin, midpoint, and a claim on the terminus.

The third layer is the Petroline's exit, and it reopened as a war of its own. On 16 July the Houthi leader declared all Saudi oil facilities and critical infrastructure targets, framing it as airports for airports, ports for ports, and blockade for blockade. On 20 July the movement declared a maritime embargo against Saudi Arabia with Bab el-Mandeb closed to Saudi vessels specifically. Riyadh struck Hodeidah on 24 July; the Houthis struck Aramco's Jazan complex on 25 July as explicit retaliation, and Reuters has since confirmed via an industry consultancy note that Aramco shut the 400,000 bpd refinery on 27 July with damage to its gasification and power block, tentative restart 15 August.

Bab el-Mandeb is not the only way out of the Red Sea, and the alternative is what makes the closure bite. Ship tracking has recorded at least six to eight vessels reversing course rather than test the embargo, among them two VLCCs loaded at Yanbu — one carrying 700,000 barrels for India, the other 2m barrels for China — both turning north toward Suez. Asian refiners led by Hyundai Oilbank are now taking Yanbu crude out through Suez and SUMED and around the Cape, turning a 24 day voyage into as much as 54. The physical penalty is heavier than the arithmetic suggests, because a fully laden VLCC cannot transit Suez on its draft: the cargo has to be part discharged into the SUMED pipeline and reloaded on the Mediterranean side, or carried in smaller hulls from the outset. The bypass has an exit. It costs a month and a cargo transfer, which is the point at which redundancy stops being redundancy and becomes a surcharge.

While the routes were closing, both belligerents crossed the same threshold in opposite directions — from degrading military capability to degrading the economic foundations that make a position sustainable.

The American side reached Iranian oil, gas and water infrastructure across Khuzestan, including the Abadan refinery and a water station at Mahshahr, then Hormozgan's bridges, railway junctions and telecom towers, and Chabahar's maritime surveillance tower. On 19 July it struck the Darkhovin nuclear power plant, which is a marker of its own: a 300MW civilian pressurised water reactor under construction since 2022, under IAEA safeguards, with no direct proliferation relevance, and therefore categorically unlike the March strikes on Fordow, Natanz and Isfahan. The IAEA has opened an investigation. The Iranian side took Kuwait's Shuaiba power and desalination complex, the Al-Subiya power and water plant, the KNPC North Pier crude export installation at Mina Al Ahmadi with injuries confirmed and fire visible on Sentinel-2 imagery, an American naval fuel support pier, an army barracks, and the operations of Kuwait International Airport.

Then on 29 July the geography extended to a coast that had not been touched. Explosions struck the Damietta gas terminal on Egypt's Mediterranean shore during cargo discharge, with Reuters geolocating circulating video and confirming both vessel identities — a US chartered floating storage and regasification unit, and a second carrier the fire spread to. Nobody has claimed it. Egypt's petroleum ministry confirmed a fire without naming a cause and some Egyptian security sources attributed it to an engine room malfunction, while Egyptian authorities simultaneously opened an investigation into the trajectory and origin of a projectile, which is not the response an engine room fire calls for. Recorded alongside that and not joined to it: Iranian state television displayed Damietta on a target map 2 days beforehand.

Whatever struck it, the target logic is new. Every prior energy strike in this war has been against export capacity. Damietta's regasification unit is import capacity, serving domestic demand against declining Egyptian production.

Set against the record, the closures are better explained as a route being taken apart than as an architecture being overrun. That reading fails if subsequent strikes cluster at one node rather than distributing along the route, which is what to watch rather than any aggregate damage total.

Which is the measurement. Three maritime axes, 4 decades of physical redundancy, every layer of it now struck, gated, embargoed or rerouted at cost — and both belligerents past the threshold into each other's economic foundations. The rung nobody takes is not somewhere near the top of the ladder. It is the one immediately above where both parties are already standing.

The scale is not in dispute and it should be conceded before anything else, because it is the strongest part of the case this section argues against.

Offensive military operations are being conducted across four continents by more than 20 parties, with a distinct second category alongside them: states whose territory has been used and struck, and which have absorbed casualties without conducting strikes of their own. North Korea has deployed an estimated 11,000 to 15,000 troops to Europe and supplied at least 40 ballistic missiles used in Ukraine. Twenty plus parties, four continents, 5 months.

That is a world war, and there is no reason to avoid the term. What there is reason to avoid is the assumption that the term means 1914.

The category has always been applied backwards. The conflict of 1914 to 1918 was the Great War to the people fighting it, occasionally the World War, and became the First World War only in retrospect — which means the numbering was a later editorial decision rather than a contemporary description. Churchill, who had by then led Britain through the second one, called the Seven Years' War "a world war—the first in history." He was not being loose. That war ran across Europe, North America, the West Indies, South America, West Africa, India and the Philippines, and counts of its belligerents run to 46. Historians have argued the same case for the Nine Years' War and for the War of the Spanish Succession, and the argument is not eccentric — it turns on whether a world war is defined by its scale and its stakes or by the particular machinery of the twentieth century's two.

So the useful question is not whether this is a world war. It is which kind, and the answer matters because the two kinds end differently.

What made 1914 what it was is not the number of belligerents but the structure connecting them: formal mutual defence obligations that converted a local trigger into a general mobilisation, because each signatory was contractually bound to the next. The alliance architecture was the transmission mechanism, and the belligerent count was its output. On that measure 1914 and 1939 are the anomaly in the series rather than its template — the two members organised by binding blocs, out of a set that mostly was not. Which makes the treaty record the thing to check, and checking it produces an answer almost nobody asserts.

There is exactly one formalised mutual defence obligation operating in this entire conflict structure, and it does not run between the parties everyone names.

The Russia and North Korea Comprehensive Strategic Partnership Treaty of June 2024 contains a genuine mutual defence clause at Article 4, requiring military assistance by all available means if either party is attacked. Russia invoked it, constructing the Kursk incursion as an armed invasion of Russian territory, and deployed North Korean troops under it — a fact Putin has confirmed publicly and by name. Moscow also holds an equivalent provision with Belarus.

The Russia and Iran Comprehensive Strategic Partnership Treaty deliberately contains no such clause. That absence was not an oversight and it was announced in advance: Iran's own ambassador to Moscow stated before signature that Tehran was not interested in joining any bloc. So Moscow spends Article 4 on the partner supplying manpower and withholds it from the partner supplying oil price leverage — which tells you what the relationship is, and it is not an alliance. The framing of this war as the United States, Ukraine and Israel against Russia and Iran is contradicted by the only documents that would make it true. The sole formalised bloc in the structure is Russia, North Korea and Belarus, with Iran outside it as a transactional supplier.

China is outside everything, and its conduct is the clearest single case that the bloc reading fails.

Beijing vetoed a Bahraini Security Council resolution to protect Hormuz shipping and condemned Iranian attacks on Gulf states' sovereignty in the same period. It runs roughly $300bn in trade with the Gulf states against something between $10bn and $40bn with Iran, and remains the largest buyer of Saudi crude on earth. What it has purchased from the Iranian side of the war is not partnership but indemnity: Chinese ships are exempted from the transit fee scheme the Houthis are preparing for Bab el-Mandeb, and Beijing held direct talks with the movement to secure unmolested passage for its tankers through the southern Red Sea. Reuters reports a contract for up to 400 shoulder fired air defence launchers routed through a Hong Kong intermediary, flown from Urumqi and transiting Pakistan, avoiding the Gulf entirely.

The deniability is not incidental to the transaction. It is part of what is being bought: materiel a partner would ship openly is instead moving overland through a third country under a Hong Kong company's signature, and Beijing has purchased alongside the launchers the ability to say it sold nothing.

The military content is worth stating precisely, because it is being read as escalation and it is closer to the opposite. Infrared systems at six kilometres patch a low altitude and counter drone gap. They do nothing against standoff munitions or high altitude platforms and they do not restore Iran's strategic air defence, which was dismantled in the war's opening hours. Iran has been manufacturing this exact family for 33 years — the Misagh-1 is a clone of the Chinese QW-1, in production since 1993 — so what is being sold is a later generation of a line Iran already builds, by the supplier that gave it the technology.

The condition that would settle this differently is already reported and not yet confirmed. Reuters separately has Iran close to a deal for anti-ship cruise missiles — a different class of weapon entirely, one that reaches shipping rather than airspace, which is the instrument this whole war runs through. No commercial logic explains that sale: a state protecting its own tankers' passage does not arm the party threatening them. If that transfer is confirmed, the reading above fails.

The strongest objection to all of this is that formal bloc structure is a lagging indicator rather than a precondition, and the historical case for it is exact. Fighting in the Ohio Valley began in 1754 and the Diplomatic Revolution that produced the two European coalitions came in 1756. Bloc formation trailed theatre linkage by 2 years, and on that precedent the absence of a treaty in July 2026 says nothing about the existence of one by 2028.

Two current items point the same way and both deserve full weight. Zelenskyy states that Russian satellite passes over Gulf states and American facilities correlate with Iranian strikes in preparation and for damage assessment afterwards, with the imagery appearing subsequently in Iran, naming four airbases inside Russian areas of interest on 19 and 20 July alone. And the Secretary of War told a Senate hearing there was evidence of Chinese or Russian assistance to Iran, while asking that the detail remain classified.

Against which stands the sharpest available counter instance, and it happened in public. When a Ukrainian strike in the Caspian killed an Iranian sailor on 25 July, Moscow called it an assault on Iran and Araghchi accepted Kyiv's assurance that it had been unintentional, declined escalation, and converted the matter into a compensation claim. The one party of the four with something to gain from a wider war was the one attempting to widen it, and the party whose citizen had died refused in public. That is not how members of a bloc behave.

So the register is wrong rather than the scale. The comparison class that fits is not 1914 or 1939 but the century before them — the Nine Years' War, the Spanish and Austrian Successions, the Seven Years' War. Those were genuinely global conflicts fought across multiple continents simultaneously, and they were fought through chartered companies, privateers, subsidised auxiliaries and colonial proxies as much as through national armies.

That is the mechanism now operating. A private military company holds a $700m 5 year contract in the Democratic Republic of the Congo with contractors on the South Kivu front line, Colombian mercenary recruitment networks supplying Sudan's Rapid Support Forces have been sanctioned by the US Treasury, Russia's Africa Corps fights directly in Mali, and the connection between the African wars and this one is not ideological or alliance based — it is a corridor. The Emirati base ring runs Mayun and Socotra, Mokha, Bosaso and Berbera; the RSF supply chain runs through those same nodes; and that is the same water the Houthis blockaded on 20 July, beside which Jazan sits. Sudan's war predates this one by 3 years and is not a front of it. What links them is a shared logistics geography and a shared contracting market, which is exactly how the eighteenth century wars linked.

The UAE arms the RSF in Sudan while arming the Congolese government against Rwanda backed M23, partly for gold export rights. The inconsistency is the finding: corridor and resource acquisition, theatre by theatre, with no requirement that the positions cohere.

Which is why the label carries a forecast. The 1914 register predicts chain ganging, general mobilisation, and a war that resolves through the collapse of one bloc. The eighteenth century one predicts something quite different and considerably worse for anyone waiting on an ending: wars that connect through corridors rather than commitments, that run for years without a decisive front, in which the parties trade with each other throughout, and that terminate in partial negotiated settlements which redistribute positions rather than in anybody's capitulation.

The absence of a bloc is not reassurance. It is the reason there is no single settlement available to reach.

Two claims have been made here and they fail differently, which matters because a framework with one undifferentiated escape route is not a framework.

The first is a claim about a pattern, and it breaks on a list: Abqaiq's stabilisation capacity, Ras Tanura, Kharg, South Pars, Gulf desalination at population scale. If any one of those is struck and destroyed, the boundary described here did not exist, and checking that requires nothing from me.

Abqaiq has to be dealt with directly, because it was struck on 27 July and the exclusion will look convenient. The terminal class names stabilisation capacity — the process that makes Gulf sour crude shippable — rather than the site. What was hit were tanks, most likely LPG, with relief valves opened to divert gas to the flare: controlled pressure management rather than an uncontained plant fire. Thermal detection supports the distinction, at roughly 70MW against 173MW and 159MW for the Jazan tank fires 3 days earlier. The definition that excludes it was written within days of the strike rather than before it. It is disclosed because someone will check, and because a list assembled after the inconvenient event is worth less than one before it.

The second claim concerns why the pattern holds, and it fails on evidence. If the lack of strikes is ever proven to be a matter of capability rather than strategic choice, the underlying logic falls apart — even if not a single target on that list is ever hit.

What is most likely to break the first claim is not a decision. It is exhaustion.

The magazine constraint does not restrain this war so much as change its shape. A defender running short of interceptors has two options and no third: accept the hits, which is now confirmed American practice through selective non-interception, or destroy the source. Caine advised that major operations would exhaust Central Command's inventory, and Admiral Brad Cooper at CENTCOM offered a binary — halt, or return to full scale war — with nothing between them, which is an option set narrowing toward the high yield end. More than $90bn in awards inside a month, including production begun ahead of appropriation, confirms the constraint is real and does nothing about it this year; tripled interceptor capacity arrives at the end of the decade.

There is a second way it goes, and it requires no decision at all. A boundary held by mutual vulnerability rather than by agreement holds only for as long as every party reads it the same way. These threats are credible as threats and are not credible as intentions, and each side is relying on the other to know the difference — which is a functioning equilibrium until somebody does not.

While this edition was being written, none of it had resolved. No American munition has been confirmed on Iranian soil since the pause began, and the answer to Iran's 28 July salvo into Jordan went to militias in a third country. Israel's defence minister says Washington is restraining him; an anonymous Israeli official says Washington is preparing something larger. Mediators are reported to be making progress on a vessel passage mechanism Tehran denies negotiating. Each of those resolves before this framework has been properly tested, and the resolutions are the test.

The claim was never that this ends well. It is that 5 months of threats have not produced the ending they described, that the reason is structural, and that structure can be checked against a list rather than taken on trust.