The live cost of war, counted honestly — including what the headline leaves out. Now tracking: the US–Israel–Iran war · Operation Epic Fury.
What you're shown vs. what it costs
The Pentagon reports the striped slice. The clock counts the rest.
The Metric That Matters · Strait of Hormuz
The bypass is what broke this week. On Sep 10, drones launched from Iraq struck pumping infrastructure on Saudi Arabia’s East–West pipeline — Petroline — in the Riyadh and Madinah regions, injuring several people. On Sep 11 the Saudi Energy Ministry shut the line as a precaution, and it has not reopened. Planet Labs and Copernicus Sentinel-3 imagery show smoke and extensive damage south of Medina. The line runs 1,200 km from Abqaiq to Yanbu and carries up to 7 million barrels a day of design capacity after its natural-gas-liquids conversion. For six months this page has counted a crude premium that persisted because the strait was shut; the premium had been partly absorbed by three workarounds — dark transits through the Omani corridor, Petroline to the Red Sea, and the UAE line to Fujairah. The largest of the three is now offline, and Saudi Arabia cannot redirect eastern crude to Yanbu by any other means at scale. A pipeline cannot be escorted. A navy can convoy a hull; no navy can convoy 1,200 km of pump stations across open desert.
The one scheduled off-ramp did not open. On Sep 13, Iranian Foreign Minister Abbas Araghchi said Tehran and Muscat had finalised the text of a new maritime route through the strait — maps drawn, joint statement on operation agreed — while stressing the arrangement “by no means” amounts to reopening Hormuz. Hours later Oman postponed the regional meeting set for Sep 14 in Salalah at which the eight states were to be briefed. Foreign Minister Badr Albusaidi posted the flags of Bahrain, Iran, Iraq, Kuwait, Oman, Qatar, Saudi Arabia and the UAE alongside the words “in the interests of consensus.” Not one of those eight governments had confirmed it would attend. Bahrain had already announced a boycott until diplomatic ties with Tehran are restored, and said any Hormuz arrangement must carry no fees, no permits and no discrimination. Iran’s foreign ministry spokesman Esmaeil Baqaei named the requesting party: Saudi Arabia, citing the war in Yemen. No new date was given. This page books a transit arrangement when it is signed and priced. A finished document with zero signatories is not a signed arrangement, and the market agreed — Brent rose about 2% on the day the meeting was meant to happen, quoted at $106.69 and at $107.54 for the November contract, with WTI at $102.
The second chokepoint went the same way. Houthi forces took the port of Mokha on Sep 10 and, days earlier, captured Perim Island — Mayun — at the mouth of the Red Sea, splitting Bab el-Mandeb into two channels under their control. That matters to this ledger for one reason: with Hormuz constrained and Petroline shut, Yanbu on the Red Sea was the remaining Saudi outlet, and its exit now runs past ground held by Iran’s ally. Two chokepoints under the influence of one actor is not two problems. It is one problem with the redundancy removed. On Sep 13 the Houthis claimed a ballistic-missile and drone strike on a Saudi base at Sharurah and said Saudi aircraft flew 129 sorties across seven Yemeni provinces in 48 hours; Yemen’s Riyadh-backed forces said they struck Houthi positions near Mokha’s port. Inside the strait itself, an Iranian cargo vessel was struck off Qeshm Island early on Sep 13 per Iranian state media, with UKMTO logging the incident.
The counting gap is unchanged and the counters have gone quiet. CENTCOM now reports 101 commercial vessels redirected since the blockade was imposed, up one from the 100 carried here last week — a count of traffic prevented, published by the party asserting the strait is open. Against that, the trackers: Kpler’s ten-day average was running near 10 commodity vessels a day in the second week of September, with 18 crossing on Sep 8, 7 on Sep 10, and no verified 24-hour count published since. This page has no independently verified transit figure for Sep 11–14 and is not printing one. Washington’s early-September escort tally — 40 vessels carrying 18 million barrels on Sep 1, per two US officials to CNN — still carries no published vessel-class breakdown and is still quoted here as a claim. The escort counts hulls. The tracker counts cargo. Brent counts cargo.
The official series moved this week, and it moved against the market. The IEA’s September Oil Market Report put Saudi supply at 6.0 million b/d in August — down 2.3 million and a 30-year low — and cut its 2026 global supply outlook by 5.7 million b/d to 100.7 million while trimming demand by 2.5 million. EIA’s September STEO, published Sep 9, raised its 2027 US crude production forecast to 14.3 million b/d and holds a mean-reversion path for Brent into next year. Its 2Q26 Hormuz series is unchanged and remains the only calibrated barrel figure on this page: 4.9 million b/d through the strait against 21.6 million in 4Q25, with Bab el-Mandeb at 8.1 million. Mitsui O.S.K. Lines still plans for no normalization at Hormuz by year-end. Charter for a Hormuz passage is still reported above $500,000 a day; war-risk cover still prices at multiples of its 0.25% pre-war level. The US Strategic Petroleum Reserve remains below 290 million barrels, its lowest since 1982.
The marker this week is the loss of the workaround. For six months the gross-versus-net gap — disruption declared against barrels actually delivered — was held open by routes that did not pass Iranian guns. One of those routes is burning and the other is behind a captured island, and the gap is now narrowing from the net side, which is the wrong side. The product shock continues to run in parallel and independent of the strait: refinery damage across the Middle East and Russia has held US diesel at record levels since AAA’s national average cleared the June 2022 high on Sep 4, printing $5.90 on Sep 8 and still reported at a record into mid-September. Meanwhile August CPI came in at 3.4% annual with core at 0.3% monthly on Sep 11, energy carrying a disproportionate share of the gain, and markets now price roughly a 90% chance of a Federal Reserve hike on Sep 16 — the first in three years. The ten-year Treasury sits at 4.975%. That is Ledger 3 repricing in real time: the energy shock is now being financed at a cost of capital that is itself rising because of the energy shock.
$29B official floor (Pentagon, May 12) — about $24B of it repairing the 42 aircraft lost — plus disclosed burn since. AP now cites more than $37.5B as a running public estimate; no new Pentagon comptroller figure has replaced the May floor. Regional base damage still uncounted. The rate now steps with the fighting, in both directions. The model ran the July nightly campaign at ~$400M/day, stepped down through the August lull as strikes paused after Jul 29, then stepped back up to ~$400M/day on Aug 30 when the US hit Larak and Iran retaliated on Jordan. The Sep 1 wave, the Sep 5 three-tanker strike, and the Sep 8 five-tanker destruction sit inside that envelope — one night's ordnance does not re-rate the day, and the model does not step on a single night. It still does not step up for sustained blockade enforcement in the Gulf of Oman or the Red Sea front, both of which carry real cost and run whether or not anything is signed. Iran's Sep 9 restricted-zone declaration does not step the rate either: a sanctions instrument is not a disclosed military expenditure, and it belongs in Ledgers 2 and 5, not here. The active rate sits above standby and below the opening surge, and it holds until a comptroller figure exists.
Official + disclosed rateThe acute shock drove oil to a $126 peak in April and added ~+0.6pp to headline inflation (Dallas Fed). Oil round-tripped below $70 on Jul 1 (EIA), spiked on the July strikes, then compressed through the August lull toward the mid-$80s — before the Aug 30 escalation carried Brent back above $90, to ~$95 by Sep 4, to $97.50 on Sep 7, and through $100 on Sep 9 — roughly $100–$102, the first triple-digit print in over a month and about $35 above the year-ago level. The premium above the ~$68 pre-war baseline now sits near $32–$34 and is trading tanker-for-tanker, not a compressing announcement cycle. The tighter market is still products, and this week it set a record: AAA's national diesel average reached $5.85 on Sep 4, passing the $5.8159 high set in June 2022, and stood at $5.90 on Sep 8 — up from $5.37 a month earlier and $3.71 a year earlier. Refinery damage in the Gulf and the Volga takes out conversion capacity no pipeline bypass restores; the White House has convened refiners on expanding capacity, which does not change a 2026 print. A second energy channel runs alongside it: Ukrainian drone strikes on Russian refineries have pushed refined-product margins independent of the strait. The pump signal reached the register in August — Walmart naming a psychological effect from roughly $4.00/gallon gasoline — and the ledger no longer has to hold a July number for it: AAA's national regular average printed $4.17 on Sep 3 and $4.15 on Sep 7. Gasoline is not at a record (the $5.016 high stands from June 2022); diesel is. Henry Hub sits in a $2.9–3.3 range with production strong. Attribution caveat: Section 301 tariffs of 10–12.5% on roughly 60 trading partners took effect Jul 24, so recent inflation carries a second, non-war driver this ledger does not claim. Pass-through already in prices does not reverse cleanly: Moody's cites ~$132B to consumers and taxpayers, booked and capped here as an in-year figure. August CPI prints Sep 11.
Modeled · in-year drag bookedThe Jul 29 meeting held at 3.50–3.75% by a 9–3 vote — Kevin Warsh's first as chair — with all three dissents arguing to hike. At Jackson Hole (Aug 27–29) Warsh read hawkish, warning that summer inflation readings did not signal an improving trend; markets moved from roughly 40% to ~57% odds of a 25bp September hike, and gold fell hard — from near $4,600 toward ~$4,440 as real yields rose. This is the rate-cut consensus of early 2026 inverting in real time: an oil-driven inflation impulse arriving into a Fed that is leaning the other way. The long end reflected it first: the 10-year touched 4.82% on Sep 2, its highest since November 2023, the 30-year 5.27%, with a buyers' strike overriding Treasury's doubled buybacks. US national debt passed $40 trillion in August. Then on Sep 3 Governor Christopher Waller said he would be inclined to hold barring surprises in the incoming inflation data, hike odds fell to ~50%, the 10-year eased, and the S&P 500 rose 1.06% to 7,747.71 — its best day since Aug 4. That rally lasted one session. August payrolls landed on Sep 4 at +162,000 against a 53,000 consensus, with unemployment steady at 4.1% and June and July revised up. Hike odds went from 49.4% to 58% in a day, the 2-year hit its highest since January 2025, and stocks gave the rally back: the Dow fell 271.86 to 53,414.25, the S&P 0.38% to 7,718.60. The selling continued through the holiday-shortened week on oil and yields — Sep 8 closed the S&P at 7,673.52 and the Nasdaq at 26,421.41 with the Dow off more than 500 points, and the 10-year topped 4.80%, its highest since October 2023, with the 30-year at 5.245%. The divergence this page flagged last week has closed, and it closed the wrong way. Equities are no longer trading a conditional hold; they are trading the same oil print the Fed is. What is left is the print itself: August CPI lands Sep 11, consensus headline steady at 3.4% and core easing to ~2.4%, with PPI expected to accelerate to 5.3% headline and 4.6% core. A 20%-plus Brent month and a record diesel average go into that number whatever anyone at the Fed prefers, and the decision follows on Sep 15–16.
Modeled · acute-phase dragLedger 1.5 · Munitions & Interceptor Asymmetry
Iran employs low-cost drones and missiles. US and partner forces counter primarily with far more expensive interceptors. The intercept usually succeeds. The economics still favor the attacker.
Defenders shifted toward cheaper effectors (guns, AIM-9X, electronic warfare, directed-energy prototypes) precisely because firing a $3M interceptor at a $30K drone is unsustainable. This narrows the per-shot ratio. It does not relieve the structural burden: every salvo still forces expenditure of limited-inventory munitions, magazine replenishment, a long logistics tail, and sustained readiness across dispersed bases. This is the multiplier standard reporting omits — and it favors the attacker over time.
Illustrative model · adds $0 to the total · confidence: low–moderateLedger 1.6 · The Asset Ledger
The $29B official figure is a total. This is the itemized invoice underneath it — every US platform reported lost or damaged, what it costs to replace, and what it cost the other side to take it off the board. The exchange rate is the point. Unit costs are public reference figures, not contract prices; where trackers and the Congressional Research Service disagree, both counts print.
| Asset | Lost / damaged | Unit reference cost | Value at risk | Taken by | Confidence |
|---|---|---|---|---|---|
| MQ-9 Reaper | 24 (CRS, May) · ~36–45 (trackers, Aug) — about a quarter of the fleet | ~$30M | $0.7B – $1.35B | SAMs, MANPADS, lost data-links, strikes on bases | CRS count official · tracker range |
| F-15E Strike Eagle | 4 destroyed | ~$90–100M | ~$0.4B | 3 to Kuwaiti friendly fire (Mar 1); 1 to a MANPADS over Iran (Apr 3) | Official |
| KC-135 Stratotanker | 7 lost or damaged — 5 hit on the ground at Prince Sultan AB; 1 crashed in Iraq, 6 crew killed | ~$40M legacy · ~$180M KC-46 replacement | $0.3B – $1.3B | Iranian missile/drone strike on parked aircraft; mishap | Official count · cost range wide |
| E-3 Sentry (AWACS) | 1 destroyed on the ground, Prince Sultan AB — parked on an unprotected taxiway | out of production; no replacement line | ~$0.3B+ · irreplaceable | Iranian missile/drone strike | Official |
| MC-130J Commando II | 2 destroyed on the ground inside Iran (self-destroyed) | ~$110M | ~$0.2B | Unable to depart during the F-15E rescue | Official |
| F-35A Lightning II | 1 damaged — first combat damage to a 5th-generation fighter | ~$80–110M | repair cost undisclosed | Iranian ground fire | Official |
| A-10C Thunderbolt II | 1 destroyed | ~$20M · no replacement | ~$0.02B | F-15E rescue operation | Official |
| Rotary & other uncrewed — HH-60W, AH-64, MH-60S, CH-47, MQ-1, MQ-4C Triton; MH-6/AH-6 ×4, UH-60 ×2 | 1 each (CRS / trackers); ×4 self-destroyed and ×2 damaged per Iranian and open-source claims | $20M – $180M | ~$0.3B – $0.6B | Small-arms fire, mishaps, one non-hostile ditching, rescue op | Mixed · partly single-source |
| AN/FPS-132 early-warning radar | 1 knocked offline | ~$2.1B | ~$2.1B | Iranian ballistic missile | Single-source · directional |
| Regional base infrastructure | multiple bases struck across Kuwait, Bahrain, Jordan, Iraq, UAE, Saudi Arabia | — | not yet counted | Iranian missile and drone waves | Pentagon: excluded from $29B |
| Destroyed airframes, subtotal | ~42 (CRS) to ~55 (trackers) | — | ~$2B – $3B | — | Estimated |
| Tomahawk cruise missiles fired | ~850+ | ~$2M | ~$1.7B | Expended on Iranian targets | Reported |
| Patriot interceptors (PAC-3 MSE / PAC-2) | ~1,500 drawn down — stock ~2,330 → ~800 | ~$4M | ~$6B | Fired at drones and missiles costing $20K–$2M each | Pentagon memo, reported |
| THAAD interceptors | 38–80% of inventory | ~$12M | multi-billion · $35B replenishment contract signed | Fired at ballistic missiles | Analyst range |
| ATACMS / PrSM | "virtually all" of the Army's long-range stock | ~$1.5–3.5M | unquantified | Expended | Analyst statement |
| Carrier strike group, days at sea | USS Abraham Lincoln 200+ days continuous, a record; reached Thailand Sep 2 after nine months | $6M – $8M/day | ~$1.2B – $1.6B, one CSG | Presence, escort, strike | Modeled |
| Munitions and interceptors, subtotal | — | — | >$26B | — | AEI, June |
| Direct cost, independent estimates | CSIS $35.2–42.5B · AEI $38.6B (both June) | — | $35B – $43B by June | — | Analyst · higher now |
| Pentagon official, cumulative | May 12 · ~$24B of it repair/replace | — | $29B | — | Official floor |
Read the exchange column. Three of the four F-15Es cost Iran nothing — Kuwaiti friendly fire took them. The fourth cost a shoulder-fired missile. The AWACS and five tankers cost one salvo at parked aircraft on an unprotected taxiway. Two dozen to four dozen Reapers went to missiles a fraction of their price. A $2.1B radar went to a ballistic missile that cost, at most, a few million. None of these was a fair fight, and none of them was supposed to be. The defender wins the engagement; the attacker wins the invoice. Every row above is a $29B total broken into the trades that produced it, and every trade ran the same direction.
The ledger above is one war's instance of a general law: when the cost of the attack falls faster than the cost of the defense, winning every engagement is how you lose. The book states the law, derives it from the physics of cheap autonomy, and lays out the only architecture that changes the exchange rate — hardware that decides before the model does. By the desk that runs this clock. AI² Press.
Get the Book on Amazon →Ledger 4 · Opportunity & Capital Diversion
This ledger invents no new money. It takes the hard floor already counted in Ledgers 1–3 and re-expresses it as the civilian outcomes that same public capital could otherwise fund. The dollars are the same dollars. The point is the trade-off.
Illustrative translation · adds $0 to the totalLedger 5 · The Rest of the World
Every figure above this line is a cost to Americans. That is the counter's discipline and it does not change. But the strait is not an American waterway, and the bill does not stop at the US border. This ledger counts the non-US economic cost across five channels. It adds $0 to the counter above and has no live ticker — it is a cumulative band, not a clock.
| Channel | Gross | Net | Basis |
|---|---|---|---|
| Crude premium | $215–360B | $80–155B | ~103 mbpd global consumption × 193 days × $12–33/bbl above the ~$68 pre-war Brent baseline, non-US share; corroborated by an OilPrice estimate of ~$330B added to global oil/fuel/LNG import bills Mar–Aug; Sep 9 Brent near $100 widens the upper bound |
| Gas & LNG | $38–68B | $27–48B | EU TTF ~€65/MWh, +97% y/y, highest since March; Qatari LNG to Europe blocked; EU storage ~55% vs 80% Nov target; Asian JKM spillover |
| Maritime & freight | $25–50B | $18–35B | War-risk AWRP at 3–10% of hull value against 0.25% pre-war; VLCC spot to $170K–$326K/day; Cape rerouting |
| Trade friction | $10–25B | $7–15B | Delayed and diverted cargo, contract renegotiation, inventory carry, ~6,000 stranded seafarers |
| Inflation spillover | $20–45B | $12–30B | Non-US CPI pass-through; EM currency and import-cost pressure |
The Question Everyone Asks
Not yet, on the strictest reading. Almost certainly yes, on the reading most people mean. Both answers are below, with the arithmetic that separates them, because a page that counts honestly has to be willing to talk itself down as well as up.
| Bucket | Range | Confidence |
|---|---|---|
| United States — direct, indirect, cost of capital (the counter above) | ~$236B | Modeled |
| Rest of world — economic (Ledger 5, gross) | $300–500B | Modeled |
| Israel — direct military, air defense, interceptor expenditure | $20–45B | Estimated · no official figure |
| Iran — destroyed materiel plus output contraction | $60–150B | Estimated · no disclosure |
| Gulf states — infrastructure damage, defense surge, LNG disruption | $25–60B | Estimated · partial disclosure |
| Running gross, all parties, day 198 | $641B – $991B | Modeled · sum of the rows above |
| US long tail not yet booked — veterans' lifetime care, munitions replenishment, base repair, debt service | $275–775B | Projected · decades |
| All-in, all parties, including lifetime | $0.92T – $1.77T | Projected |
The high end is now at the line. Re-added from the component rows above, the running all-parties gross on day 198 is $641B – $991B. The top of that band sits roughly nine billion dollars below a trillion — two to three days of accrual. It has not crossed. It will have crossed by the time most readers see a second reading of this page, and when it does, this line will say so plainly rather than quietly. The correction booked last week stands: an earlier reading printed $705B–$1.07T and claimed the high end had already crossed; that total did not reconcile with its own inputs, the inputs won, and the band was brought down. The number is arriving at the mark it was wrongly said to have passed, five weeks later and by arithmetic instead of assertion.
The countdown shortened again, and the reason is specific. At an all-parties gross accrual of roughly $3.3–3.8B a day — US direct burn, plus a crude premium now running about $35 above the pre-war baseline with Brent at $107.54, plus freight and gas — the high end of the band crosses a trillion within days, the central reading around the end of October, and the low end at the turn of the year. That last figure is a walk-back from “before year-end”: at 94 to 109 days out the low end now lands on or just past 31 December, and the honest statement is the turn of the year rather than inside it. The countdown runs whether or not anyone signs anything, because the largest single channel is a price differential that persists while the strait stays functionally shut — and this week that differential acquired a second driver that has nothing to do with the strait. Petroline’s closure removes bypass capacity, and removed bypass capacity widens the premium in exactly the way a Kharg strike would. The asymmetric risk this page has flagged for months as hypothetical partly materialised on Sep 11, from a direction nobody was escorting.
The one reading on which a trillion has already been passed is the lifetime one, and that is not a new claim. The drivers table below has carried $500B–$1T as the long-run all-in US figure for months. Add the rest of the world and the lifetime total clears a trillion comfortably. What changed this week is not the arithmetic on the lifetime line. It is that the running line is now close enough to it that the distinction between the two readings stops mattering to anyone but a bookkeeper.
~$400M/day active rate, Ledger 1 and everything downstream of it comes down with it. As of Sep 14, none has triggered, and one moved backwards. Brent at $107.54 sits about $35 above the $73 trigger, further away than at any prior reading of this page. Trigger two went into reverse. Last week this block reported that Iran was close to a temporary safe-passage arrangement with Oman and said the page would book it when it was signed and priced, not when it was reported. That was the right rule. Araghchi confirmed on Sep 13 that the Iran–Oman route text is finalised — maps and joint statement included — and on the same evening Oman postponed the meeting at which the other seven states were to be briefed, with none of the eight having confirmed attendance, Bahrain boycotting outright, and Saudi Arabia named by Tehran as the party that asked for the delay. A finished document with no counterparty is further from a priced transit regime than a negotiation in progress was. War-risk cover still prices at 3–10% of hull value. No comptroller figure exists; the modeled $400M/day active rate is carried unchanged despite a week of tanker-for-tanker at scale, because a disclosed rate and an inferred one are different things and this page does not promote the second into the first. The fourth uncertainty is now resolved and it resolved against the official series. EIA’s September STEO published on Sep 9 and holds a mean-reversion path — it raised the 2027 US crude production forecast to 14.3 million b/d and keeps Brent well below spot for the second half — while the IEA cut 2026 global supply by 5.7 million b/d to 100.7 million, put Saudi output at a 30-year low of 6.0 million b/d in August, and cut demand by 2.5 million. Spot is $107. Two official agencies now disagree with each other and both disagree with the tape, and that gap — not any line in Ledgers 1 through 3 — remains the largest forward uncertainty in this model. The counter does not move on hope in either direction.The Cost That Does Not Reverse · Human Cost
The dollar total above is the quoted number — it falls when oil falls. This is the number that does not reverse. Killed and wounded are tracked separately by the reporting bodies; wounded is shown only where an official body discloses it. Iran releases no reliable count, so the documented floor understates the real toll — almost certainly by tens of thousands.
| Toll | Killed | Wounded | Source · as of | Confidence |
|---|---|---|---|---|
| US service members | 17 | 430+ | CENTCOM / AP (Jul 21) | Official |
| — US injured this month (since Jul 7) | — | ~100 | Pentagon / Parnell (Jul 20) | Official |
| Seafarers / mariners | 20 killed + 1 port worker | 35 | Hormuz crisis compiled tally (Sep 9); 1 still missing | Compiled |
| — Saudi tanker attacked in the strait (Sep 1) | 2 Filipino sailors | undisclosed | Saudi Foreign Ministry (Sep 2) | Official · single government |
| — M/T Hercules Star, anchorage off Dubai (Sep 9) | 1 crew, 1 missing | undisclosed | Charterer statement (Sep 9) | Official · single operator |
| Commercial vessel incidents | 52 reported | — | IMO, via UANI (Jul 10) | Compiled |
| — Petroline pump stations, Riyadh & Madinah regions (Sep 10) | none reported | “multiple” | Saudi Energy Ministry (Sep 11); no number published | Official · unquantified |
| Saudi Arabia — Houthi strikes on southern cities & Sharurah base | not yet counted | 73 reported injured (Sep 8) | Riyadh statements; Houthi claims (Sep 13) | New front · unreconciled |
| Yemen — Saudi strikes on Hodeidah | not yet counted | not yet counted | al-Masirah (Jul 25); Houthis claim 129 sorties across 7 provinces in 48h (Sep 13); Riyadh silent | New front |
| All fronts (compiled) | floor only | undisclosed | Iran & others undisclosed | Undercounted |
Three US service members were killed in the week before — two at Jordan's Muwaffaq Salti Air Base on Jul 17, one in northern Iraq on Jul 18 — the first American combat deaths since March. The Pentagon now says nearly 100 service members have been injured this month alone, since Jul 7 (Parnell, Jul 20); it reports 96% returned to duty, mostly minor concussions, though independent all-injury tallies including ballistic-strike TBIs run higher. The maritime toll is climbing on its own: across the Hormuz crisis, roughly 17 merchant ships damaged, 7 abandoned, 2 captured, a tugboat sunk, and 12 seafarers killed or missing, against 52 commercial-vessel incidents reported to the IMO across the Gulf, the strait and the Gulf of Oman. The lull broke on Aug 30: the US strike on Larak Island drew Iranian-claimed casualties, Iran's retaliatory missile-and-drone strike on two US bases in Jordan was met by eight interceptions, and a supertanker struck two mines in the southern strait and caught fire. On Sep 1 Riyadh says an Iranian attack on a Saudi tanker in the strait killed two Filipino sailors — the first named maritime deaths of the new round — and the IRGC claims two more tankers were disabled by mines, which CENTCOM denies. No CENTCOM combat-death update has been published since Jul 21, so the killed row is held at its last named figure rather than extrapolated. Compiled wounded tallies from the Defense Casualty Analysis System now print higher than the July 430+ disclosure (trackers cite 794 across Epic Fury and overseas operations combined as of early September) and are marked as compiled, not substituted into the official row. The compiled Hormuz-crisis maritime toll has since risen to 20 seafarers and one port worker killed, 35 injured and one still missing, and on Sep 9 one crew member was killed and another went missing aboard the products tanker Hercules Star at an anchorage off Dubai. Since then the injury ledger has widened onto land and away from hulls: the Sep 10 drone strikes on Petroline’s pump stations in the Riyadh and Madinah regions injured several people by Riyadh’s own account, with no number published; Houthi strikes on Saudi southern cities were reported to have injured 73 in early September; and on Sep 13 an Iranian cargo vessel was struck off Qeshm Island inside the strait, with no casualty figure released by either side. The Yemeni front, the Sep 8 Jordan salvo, the Sep 9 Jazan refinery strike and the Sharurah base attack still have no reconciled public count, and the 129 Saudi sorties the Houthis say were flown across seven provinces in 48 hours have produced no casualty disclosure from any party. No CENTCOM combat-death update has been published since Jul 21 — now eight weeks. The killed row is held at its last named figure rather than extrapolated, and the length of that silence is itself worth printing. The wounded outlast the war longest of all. Every figure here is institutional, sourced, and dated.
The drivers, line by line
The labeling is the point — official numbers are a floor, the rest are models with real error bars.
| Driver | Figure | Confidence |
|---|---|---|
| Pentagon cumulative direct (May 12) | $29B | Official |
| — of which: hardware repair / replace | ~$24B | Official |
| US aircraft lost or damaged | 42 (CRS, May 13) · trackers count ~55 by August · see the Asset Ledger | CRS · trackers |
| Regional base damage | not yet counted | Pending |
| Energy / inflation drag to households | ~$100–132B | Modeled |
| Brent crude | $106.69 spot / $107.54 Nov contract (Sep 14, +2.0%/+2.8%) · four-month high · rose on the day the Salalah talks were to be held | ICE / NYMEX · sources differ by contract |
| Brent monthly range | ~$84–108 (Aug lull → Sep 14; $101 Sep 9, $105.16 Sep 10, $104.61 settle Sep 11, $106.18–108.46 Sep 14) | ICE |
| Brent 2026 peak / low | $126 (Apr) / <$70 (Jul 1) | ICE/EIA |
| WTI | $100.00–$102.52 (Sep 14, +1.95–2.47%) · Brent–WTI spread holding near $5 | NYMEX |
| US avg gasoline | $4.15/gal (AAA, Sep 7) · $4.17 Sep 3 · below the $5.016 record of Jun 2022 | AAA |
| US diesel / heating oil | diesel $5.90/gal (Sep 8) — record $5.85 set Sep 4, past the $5.8159 high of Jun 2022; $5.37 a month ago, $3.71 a year ago · heating oil +~98% y/y | AAA / OilPrice |
| Henry Hub gas | ~$2.9–3.3 · US insulated, production strong | EIA |
| Gold (safe-haven read) | ~$4,282–4,300/oz (Sep 14, −1.5%) · lowest since August, fourth straight weekly loss · falling into a widening war because the 10-yr at 4.975% costs more to hold it than the geopolitics is worth | LBMA/CFD · directional |
| EU gas (context, not a US cost line) | ~€65/MWh (+~97% y/y) · highest since March; Qatari LNG to Europe blocked | Trading Economics |
| Route deal status | No successor to the June 17 MOU. Iran–Oman route text finalised (Araghchi, Sep 13) — maps and joint statement — but the Sep 14 Salalah meeting to brief the region was postponed with no new date, none of the eight flagged states had confirmed attendance, Bahrain is boycotting, and Tehran says Riyadh asked for the delay. Araghchi: the text is “by no means” a reopening. Zero signatories | Escalated |
| Strait status | Iran: closed except pre-approved; US: blockade + tanker-for-tanker; JMIC: severe; MOL: no 2026 normalization; Derya struck near Kharg Sep 8; Kharg itself unstruck | Contested |
| Hormuz transits, daily | No verified count since Sep 10. Last prints: Kpler 10-day avg ~10 (lowest since May), 18 on Sep 8, 7 on Sep 10 · Energy Sec. Wright claims 9 mb/d getting through · CENTCOM: 101 hulls redirected (Sep 14) | Stale · trackers vs official claims, unreconciled |
| US Strategic Petroleum Reserve | below 290M bbl — lowest since 1982 | EIA / reported |
| Supertanker mine strike | hit two naval mines in the southern strait and caught fire (Aug 30) | New |
| Dark crude transits since Jul 7 | 72 of 84 crude tankers (86%) sailed dark via the Omani lane | Kpler |
| Dark-fleet Gulf crude exports | ~4.8 mbpd, from 1.6 in May · ME exports rebounded ~15 mbpd via workarounds (US Energy Dept.) | Vortexa / Kpler / DOE · directional |
| Maritime incident (latest) | Sep 13: Iranian cargo vessel struck off Qeshm Island inside the strait (Iranian state media / UKMTO). Sep 10–11: drones from Iraq hit Petroline pump stations, Riyadh & Madinah regions, several injured. Sep 9: several vessels on fire per UKMTO; Panama-flagged tanker with 2M bbl Iraqi fuel oil hit; Hercules Star crew death off Dubai | New |
| Oman oil spill (uncounted cost) | ~500 sq mi from a sanctioned Russian-crude tanker; beaches reached; Tehran seeks compensation | New · unpriced |
| India Russian-crude arbitrage | crude imports 4.7 mbpd Aug (from 5.05 Jul); refined-product exports at a 3-yr low as refiners run inward; discount compressing as China outbids and Novorossiysk loadings fall | Kpler / ThePrint / OilPrice |
| US campaign posture | Tanker-for-tanker at scale: Sep 1 two → Sep 5 three → Sep 8 five (largest single day); blockade in force; Rubio warns Iran will keep losing tankers; Qatar urging reopen | CENTCOM / State |
| New fronts | Perim Island (Mayun) captured by Houthi forces, splitting Bab el-Mandeb; Mokha taken Sep 10; Sharurah base struck Sep 13; Yanbu & Jizan struck; Hodeidah struck; CPC halted. Two chokepoints now under the influence of one actor | Escalating |
| Bypass SHUT (East–West Petroline) | 0 flowing since Sep 11. 1,200 km Abqaiq→Yanbu, up to 7 mbpd design capacity; shut by the Saudi Energy Ministry as a precaution after the Sep 10 drone strikes. Planet Labs / Copernicus imagery show extensive damage south of Medina. No restart notice. Saudi Arabia cannot redirect eastern crude to Yanbu by other means at scale | Official · Saudi Energy Ministry |
| US service members injured, this month | ~100 since Jul 7 | Pentagon |
| Fed funds rate | 3.50–3.75% · Aug CPI landed Sep 11 at 3.4% headline / 0.3% core monthly, energy carrying a disproportionate share · hike odds ~86–90% for the Sep 16 decision, which would be the first hike in three years · Goldman flipped to a September hike call · ECB already raised 25 bp on Sep 10 | FOMC / CME FedWatch |
| Treasury yields | 10-yr 4.975% (Sep 11), approaching 5% · 30-yr 5.354% · 5-yr 4.791% · 2-yr 4.63% · this is Ledger 3 repricing in real time | Treasury |
| US national debt | passed $40 trillion (Aug) | Treasury |
| Equities (Sep 8 close) | S&P 7,673.52 (−0.58%) · Nasdaq 26,421.41 (−0.32%) · Dow −511 pts · three sessions gave back the Sep 3 rally (S&P 7,747.71) · Sep 9 futures lower on oil | Exchanges |
| New US tariffs (confounds inflation attribution) | 10–12.5% on ~60 partners, eff. Jul 24 | Section 301 |
| Cost-of-capital drag, annual | ~$48B | Modeled |
| War-risk premium, share of hull value | 3–10% vs 0.25% pre-war (last print July) · ~6,000 seafarers stranded · Hormuz-passage charter reported >$500K/day · MOL: no normalization in 2026 | IMO / brokers / Baltic |
| Long-run all-in, US only (lifetime) | $500B–$1T | Projected |
| Rest-of-world economic cost (Ledger 5) | $300–500B gross · $135–255B net | Modeled |
| Hormuz throughput, calibrated (official) | 4.9 mb/d crude + liquids in 2Q26 vs 21.6 mb/d in 4Q25 · Bab el-Mandeb 8.1 mb/d from 5.4 · shut-ins ~5.5 mb/d (Jul) | EIA STEO (Aug 11) · Sep edition due Sep 9, not yet reflected |
| Iran restricted / exclusion zone | Announced Sep 6–9 (Rezaei, SNSC): runs from the US blockade line into the Gulf; non-coordinating vessels sanctioned and cut from insurance and support services. As of Sep 14 no coordinates have been published, and the Salalah meeting that would have briefed the region was postponed with no new date | New · unpriced |
| All parties, running gross, day 198 | $641B–$991B (re-added Sep 14 from component rows; high end now ~$9B below the trillion mark) | Modeled |
This is a transparent client-side model, not a live government feed. It computes elapsed time from Feb 28, 2026 and applies published daily-burn and modeled drag rates. It shows its own math so you can disagree with it.
$1.88B/day days 1–6 (Pentagon briefing), ~$500M/day sustained to the Apr 8 ceasefire, ~$95M/day standby through Jul 8, ~$400M/day for the July nightly strike campaign, then a step down to ~$150M/day for the August lull as strikes paused after Jul 29 (blockade enforcement only), and a step back up to ~$400M/day from Aug 30 when the US struck Larak and Iran retaliated on Jordan. The rate steps with the fighting in both directions but never on inference — a threatened blockade or a sanctions announcement is not a disclosed military rate. Conservative: it does not step up for the expanded target set, Gulf of Oman blockade enforcement, or the Red Sea front. Reproduces ~$50B, consistent with the $29B May 12 official floor plus disclosed and modeled burn since.~$132B band as an in-year figure already worked into prices. Oil's return toward pre-war levels does not unwind costs already passed through.~$48B/yr from the rate curve repriced during the acute phase. What has changed since is direction, not magnitude: the early-2026 rate-cut consensus has inverted, with Warsh hawkish at Jackson Hole and markets pricing ~57% odds of a September hike. Represents in-year drag booked and held at the band, not re-tallied on each repricing.Today's model output, stated once. On day 198 the three ledgers produce Ledger 1 $56.1B + Ledger 2 $132B (capped) + Ledger 3 $48B = ~$236.1B. Every US figure quoted in prose on this page is that number or derived from it. Note that only Ledger 1 accrues: Ledgers 2 and 3 are booked caps, so the per-second burn shown at the top is the Ledger 1 marginal rate alone, not a rate across the combined total.
To make it truly live: replace the rate constants at the top of the script with a fetch to a data file or API you control (Pentagon comptroller, CSIS, Penn Wharton, Dallas Fed, Treasury). The display layer swaps the source in without touching the UI.
Sources: Pentagon comptroller testimony, CRS, CSIS, Penn Wharton Budget Model, Dallas Fed, Moody's Analytics, US Treasury, EIA/ICE, Trading Economics. Human-cost figures: US Central Command, CRS, and national health authorities — official figures labeled; estimates and ranges labeled as such.
Further Reading · The Framework Underneath
This clock measures a single war. The book is the framework beneath it: why a $30K drone can dictate the spending of a $3M interceptor, why a force can win every engagement and still lose the war on the spreadsheet, and what deterministic control means when machines enter the loop. The pattern this page tracks in real time, argued in full.
Get the Book on Amazon → The complete AI² library →