Starting with this edition, The Hormuz Ledger is The Macro Ledger.

The strait was the entry point, but was never meant to be the ceiling. The supply chain cascade, the credit stress, the dedollarisation architecture, the sovereign reserve shifts — none of that lives inside one chokepoint, and never did. The analysis outgrew the name before Edition 3.

The Macro Ledger is what this publication was always going to become. Same thesis, same voice, same framework. The masthead is the only thing that changed.

We'll be here for whatever comes next.

— Diego Miranda

The story being told about this crisis is that one country has been exposed by the war and one country has been vindicated by it. The exposed country is China — energy-import-dependent, Hormuz-exposed and geopolitically isolated, the vindicated one is the United States — the new global filling station, energy-independent and the producer of last resort. That story has been priced into oil, into equities, into Treasury yields, and into the political language of an entire ceasefire negotiation.

The story is wrong, not at the edges, but at the centre.

The structural conditions this edition documents were not created by six weeks of war; they were merely exposed by it. What the consensus model has not absorbed is that the inversion is not limited to oil — every supply chain this edition revisits runs the same shape, and so does the financial architecture under all of it.

The headline narrative is one of US dominance. The structural reality, in domain after domain, is one of US dependency, and the gap between the two is what the rest of this edition measures.

The pattern is not a coincidence but the visible signature of two decades during which one country built supply chain redundancy as a strategic priority, and the other treated supply chain efficiency as a market outcome.

This is the first edition published as The Macro Ledger. The strait was always the entry point, never the ceiling, and what we documented in the strait pointed at a much larger structural shift. The rest of this edition is what that inversion looks like — in oil, in chips, in weapons, in food, and in the financial architecture downstream of all of them.

Read on.

Two claims have been made with equal confidence over the past six weeks: that the United States is energy independent and that China is the strategic casualty of the Hormuz crisis. Both are consensus, and both are wrong in the same way — each conflates a headline number with a structural position, and each one collapses under the same scrutiny.

The US oil system is a 2.2 mb/d net crude importer with refineries optimised for a grade it does not produce, drawing heavy sour from three suppliers — one of which is diversifying toward Asia, one of which is collapsing, and one of which is a ruin, whilst the Chinese oil system is a 70% import-dependent colossus that spent twenty years engineering the exact layered architecture the current crisis tests.

When the US government claims energy independence, it aggregates crude oil with refined petroleum products, and under that aggregation, yes, the US is a net exporter. However, under the physical reality, it imported 6.2 mb/d of crude oil in 2025 and exported 4.0 mb/d — a net crude deficit of approximately 2.2 mb/d. The crude exported is not the crude imported. Over 80% of US Lower 48 production is light sweet (API above 35), and over 60% of US refinery capacity is optimised for heavy sour (API at or below 27). Canada supplies 62% of crude imports for a reason: US refineries physically require the grade the US does not produce.

Counting refined products made from foreign feedstock as American exports, and netting them against crude imports to claim self-sufficiency, is an accounting artefact, not a physical reality.

There is no version of American refining that runs on domestic supply alone.

The PADD 3 Gulf Coast complex — the heart of American refining — has over 3 mb/d of coking capacity designed to convert heavy, high-sulphur crude into gasoline, diesel, and jet fuel. These facilities represent hundreds of billions of dollars of installed investment, built over decades to process grades from Venezuela, Mexico, and the Middle East. Reconfiguring a single refinery for light sweet crude costs $100 million to $1 billion; across the Gulf Coast fleet, the total retrofit would run into tens of billions across a decade. No company has announced it, no investor has funded it, and no policy has incentivised it.

The last US refinery was built in 1977, and since 2020, over 800K bpd of refinery capacity has closed. The fleet is shrinking, not expanding, and the configuration mismatch is permanent absent a multi-decade reconfiguration programme that nobody is building.

Machine-learning models from Gorozen, using subsurface geological data, estimate the average publicly traded Permian company has 3.7 years of Tier 1 drilling locations remaining: the Permian has developed nearly 60% of its Tier 1 acreage, the Eagle Ford and the Bakken reached that same threshold in 2018 — the year both basins stopped growing, the Bakken is now 66% developed, with nearly 90% of its best acreage drilled out.

The child well problem compounds the geological clock. Wells drilled near existing wells — "children" — produce 5–20% less oil than parent wells, and in 2012, 30% of new wells were children, by 2022, the number was 85%; these wells are increasingly drilled into depleted pressure fields, and the productivity impairment is structural, not cyclical.

The EIA's own long-term outlook projects US production declining to 12.4–12.7 mb/d by 2050. That is the government forecaster, not the industry's critics.

Canadian heavy crude was captive to the US market for decades. Alberta was landlocked, pipelines ran south, and WCS traded at a $10–20/bbl structural discount to WTI, the Trans Mountain Expansion, operational since May 2024, tripled pipeline capacity to 890K bpd on Canada's Pacific coast, and two-thirds of TMX crude exports went to Asia-Pacific in the first eleven months of 2025. China became the second-largest destination for Canadian crude, the WCS–WTI discount narrowed from $15 to under $10 and Prime Minister Carney has committed Canada to 50 Mtpa of LNG for Asian markets by 2030 and signed a strategic energy cooperation agreement with Beijing.

The two other historical heavy-sour suppliers are collapsing. Pemex exports hit a 35 year low of 368K bpd in December 2025, down 54% YoY, and Maya crude exports are down 86% from 2020. Pemex carries $101 billion in debt and is losing 100K bpd of production per year; the IEA projects Mexico as a net oil importer by 2030. Venezuela produces ~1 mb/d against a 3.5 mb/d peak, with rehabilitation costs estimated at $58 billion — before any political settlement, which would itself take years to negotiate.

The Gulf Coast refinery fleet was built around three heavy-sour suppliers: one is diversifying toward a competing buyer with deeper pockets and a growing demand curve and the other two cannot deliver.

Every barrel Canada sends to China is a barrel the Gulf Coast does not get.

China is 70–75% import-dependent on crude and imported 11.1 mb/d in 2024. On the conventional metric — import dependency — it is more exposed than the US. That is the consensus framing.

What the consensus framing does not count is what Beijing built around the import dependency over the last twenty years. The supplier base is deliberately diversified: Russia (~20%), Saudi Arabia (~14%), Malaysia/Iran (~13%, mostly relabelled Iranian), Iraq (~11%), Brazil (~6%). No single supplier exceeds 20%, and over 20% of imports flow through sanctioned channels — Iranian, Venezuelan, and shadow-fleet Russian crude — that operate outside Western enforcement entirely. The parallel supply chain has its own logistics, insurance, and payment infrastructure, with state-administered quotas (257 Mt allocated to non-state refiners in 2026) keeping it formalised rather than black-market.

Storage is the second layer: strategic reserves stand at 1.1–1.3 billion barrels — approximately 3.2× the US SPR and equivalent to ~118 days of net import cover, in contrast, the US SPR sits at 415M barrels, 58% of its 714M capacity.

Pipelines are the third: the ESPO pipeline delivers 700K bpd of Russian crude overland to northeast China, Power of Siberia 1 reached full capacity of 38 bcm/yr in 2025, Power of Siberia 2 (50 bcm/yr) has an MOU signed as of September 2025, with first supply projected for 2030–31, the Central Asia gas pipeline adds ~36 bcm/yr from Turkmenistan and Kazakhstan. These volumes arrive without transiting a single maritime chokepoint.

Demand destruction is the fourth: China's EV fleet displaced 540K bpd of gasoline demand in 2026 — up 25% YoY from 430K in 2025. EV and LNG heavy-duty vehicles displace another ~500K bpd of diesel, the CNPC Economics Institute forecasts Chinese oil demand peaking between 2026 and 2030. The import dependency is real, but the trajectory of what is being imported is pointing down, not up.

Roughly 57% of Chinese crude imports transited Hormuz pre-crisis — genuine exposure, but not the 90% the headline seaborne share implies, the remaining 43% arrives via routes the current crisis does not directly touch.

None of this makes the Chinese architecture invulnerable. Tightened sanctions enforcement against the teapot-refinery supply chain would compress the sanctioned barrel channel, a Russian supply interruption would strip out the overland buffer. Peak demand is a forecast, not a fact, and Chinese energy forecasts have missed in both directions historically. What the architecture does is distribute failure modes across enough independent channels that no single failure is catastrophic — which is a different property from invulnerability, and a more useful one.

The Iran war was framed, in its opening weeks, as the moment Chinese energy security would be exposed as hollow. Fifty days in, the architecture held; domestic production set records (4.32 mb/d, +1.5% YoY), refinery throughput set records (14.7 mb/d in December). The SPR was not drawn down, overland pipelines ran at capacity, sanctioned barrels continued flowing, and the yuan-denominated Hormuz toll added a new revenue line.

The same crisis exposed the US architecture's structural dependencies; Gulf Coast refiners are running short on heavy sour, TMX crude is flowing to Shanghai, not Houston, 68 empty VLCCs sailing from the Atlantic toward Texas is 5 to 6 weeks of US surplus — a photo opportunity, not an energy strategy. The Strategic Petroleum Reserve is half empty, the rig count is down 33% from its December 2022 peak and the EIA projects decline from here.

The consensus narrative has the arrow pointing the wrong way, and the question the rest of this edition asks is what happens to everything else that runs on energy and chemistry — chips, weapons, food and money — when that arrow is recognised.

Each of the sections that follows is a different answer to that same question.

The three sectors covered in this edition have been covered before — food in Edition 3, semiconductors in Edition 4, defence in Edition 5. We return to each not to catch up on news but because in each case, what has happened since is not recovery but confirmation; the biological clock on the 2026 Northern Hemisphere crop has passed while the market was still pricing fertiliser as a cyclical stress, the semiconductor supply crisis a was quadruple-input in Edition 4; it is quintuple one now, and the demand that meets those inputs is running at the steepest angle in the industry's history, and the Pentagon has started to publicly name the defence production problem, and every action it has taken since — from the automaker talks to the $1.5 trillion budget request — is downstream of the materials it does not control.

Edition 4 described a quadruple-input crisis — helium, bromine, neon, and tungsten — across the semiconductor supply chain. At the time, sulphuric acid was the meta-variable running underneath; naphtha was the petrochemical cascade building pressure on Korean and Japanese producers; and fab energy was a tail risk that had not yet metastasised. Five weeks later, all three have moved forward; China's sulphuric acid export ban takes effect 1 May, naphtha crackers in South Korea are running at 65% utilisation with 1 to 2 weeks of inventory, 3 AWS data centres in the Gulf have been physically struck, and roughly half of planned US data centre capacity for 2026 is now cancelled or delayed.

The cleanest way to see the current state of the sector is by stress-testing the stack.

Seven pressure points across the industry, from three independent geopolitical surfaces. Tungsten, sulphuric acid, and the antimony and gallium adjacent to fabrication are China-gated, whilst helium, neon, sulphur, and naphtha are Hormuz-gated and bromine is Israel-theatre-gated. Fab energy is gated by all three simultaneously — Hormuz for Taiwan's LNG supply, kinetic strikes for the Gulf data centres, and the transformer and switchgear lead-time bottleneck that was constraining US DC buildout independently of the war.

No single diplomatic outcome unblocks the stack. A Hormuz ceasefire does not restore China's acid exports or give the industry its tungsten back, a US–China trade deal does not repair Ras Laffan or reclear the bromine plant. The Diplomat framed it accurately in early April: this is a category of vulnerability that was never in any semiconductor stress test. Industry planning was built around single points of failure — Taiwan Strait scenarios, single-vendor dependencies, specific-country exposures.

No model assumed seven concurrent pressure points from three independent sources.

The demand side is running in the opposite direction, and at the steepest angle the industry has ever seen.

ASML's Q1 2026 net sales came in at €8.8 billion, above guidance midpoint, beating street estimates, gross margin 53%, at the top of the guided range and EPS €7.15 against €6.00 consensus. The full year 2026 revenue guide was raised to €36–40 billion, from €34–39 billion. CEO Christophe Fouquet attributed the raise to AI infrastructure capex creating supply constraints and pushing customers to expand orders. EUV Low NA systems are ramping to 60 in 2026 and 80 in 2027.

Memory is the sharpest signal: DRAM and HBM spot prices rose 80–90% in Q1 2026 against Q4 2025, according to Counterpoint, memory inventories fell from 17 weeks in October 2025 to under 4 weeks. SK Hynix has confirmed its 2026 DRAM, NAND, and HBM capacity is fully sold out, largely to NVIDIA, and Samsung is planning a 50% HBM capacity surge for 2026 with HBM4 mass production launching in February for NVIDIA's Rubin architecture. Samsung and SK Hynix together control 90% of global HBM; together with Micron they control 95% of global DRAM. The Korean duopoly is reallocating up to 40% of advanced-wafer capacity to AI memory, forcing commodity PC and smartphone markets into rationing. Citi projects the HBM market at $43 billion by 2027, from roughly $3 billion in 2024.

This is the cleanest AI demand signal visible on the tape, and also the cleanest amplifier of every pressure point above, when the industry is shipping every qualified die it can make, each material shortage translates directly into price, and there is no inventory to absorb it, and no slack production to pull forward. The supercycle is not causing the constraint; it is the reason the constraint bites immediately, rather than being deferred across build cycles.

Every axis of this stress concentrates semiconductor-adjacent production further in China: On the Hormuz axis, China has abundant domestic electricity, Russian naphtha flowing overland, and insulation from the petrochemical cascade hitting South Korea and Japan, on the China axis, Beijing controls 79% of tungsten; export controls on gallium, germanium, and antimony are active; and the sulphuric acid ban from 1 May completes the sequence. Edition 4 noted that the war was delivering China leverage it could not have obtained through trade policy alone, 5 weeks later, the leverage is showing up in the earnings — the only large region where semiconductor-adjacent packaging and process chemical production is expanding is the one region whose government is gating the inputs.

The market is pricing this crisis as a repeat of 2020–21, where a capacity shortage was solved by building more fabs. But 2020–21 was a capacity problem, and this is a materials problem. No amount of new fab construction resolves the constraint. Every fab under construction on earth — Arizona, Ohio, Dresden, Kumamoto, Hsinchu — draws from the same upstream pool.

One structural tail is under-priced across the sector: TSMC manufactures the overwhelming majority of the world's leading-edge logic on Taiwan, and Taiwan sources roughly 37% of its electric-grid fuel as LNG from the Middle East. LNG prices rose 60% post-blockade, Ras Laffan is offline through at least 2029–31, TSMC's high helium recovery rates mitigate the fab-level helium constraint significantly — but the grid running those fabs is a different problem, and that is energy security, not materials security; that distinction sits nowhere on the fab income statement, and will remain invisible in financial disclosures until it becomes unignorable, which is the pattern every grid-exposure risk follows.

The ceasefire is irrelevant to semiconductor material supply on the 6–12 month horizon that matters, the 200 helium containers are already empty, the Ras Laffan repair is measured in years, China's acid ban begins in two weeks, and the Bloomberg-reported "possible resumption later in 2026" signalling is a Beijing lever, not a planning assumption — with acid now representing 64% of Chinese smelter byproduct revenue, up from 27%, the economic default is retention, not export.

What would break the thesis is a coordinated resolution across all three geopolitical surfaces — Hormuz reopening durably, China lifting export restrictions on tungsten and acid, and Ras Laffan's repair timeline accelerating materially — inside a window short enough to matter.

Edition 5 described the quad-threat — rare earth magnets, tungsten, nitrocellulose, and sulphuric acid — as the four interlocking constraints defining the duration limit on US kinetic operations. That quad-threat framing understated the problem by more than half. The Pentagon's own critical minerals list contains 13 inputs, The DoD solicited proposals for securing all 13 on 28 February — the day before strikes began. Every one of them runs through China.

On 15 April, the Wall Street Journal reported that the Pentagon has approached General Motors, Ford, GE Aerospace, and Oshkosh about redirecting commercial manufacturing capacity toward weapons production. That same day, Defence Secretary Hegseth confirmed a $200 billion supplemental war funding request and Trump has proposed a $1.5 trillion military budget — the largest in modern history. Framework agreements were announced on 25 March with BAE Systems, Lockheed Martin, and Honeywell Aerospace to quadruple THAAD interceptor seeker output, triple PAC-3 production, and accelerate the Precision Strike Missile. Deputy Defence Secretary Stephen Feinberg now leads weekly meetings of a Munitions Acceleration Council with defence firm executives.

None of it will close the gap. The constraint on US munitions production is not money, not factory space, and not willingness, but materials.

Thirteen materials. China controls 55–100% of supply on every single one. Nine are already under formal Chinese export control or restriction and the ones that aren't — vanadium, hafnium, nickel, zirconium — are flagged by the DoD as major dependency risks regardless.

The application column is the important one. Arsenic at 100% Chinese supply is the semiconductor dopant that GaAs and every III-V chip runs on; without it, no radar, no precision-guided munition, no advanced sensor. Samarium at ~90% goes into the permanent magnets that guide every precision weapon. Yttrium at ~90% is in radar, jet engine coatings, and precision optics. Tungsten at 79% is the armour-piercing penetrator core and the WF₆ that fabricates every chip downstream of it. Germanium at ~60% is night vision, fibre optics, and satellite solar cells. Graphite at ~65% is nuclear moderator and EV battery anode — the same input that civilian EV supply chains are competing for.

There is no substitution pathway for any of them at scale. The US has taken equity stakes in MP Materials for rare earth magnets, Lithium Americas for lithium, and Trilogy Metals for copper, and Energy Fuels is developing gadolinium and samarium processing. These are real programmes. MP Materials' target capacity at full ramp is 1,000 tonnes, per year of rare earth oxides represents 0.3% of Chinese output of approximately 300,000 tpa, and the domestic tungsten mine coming online in 2026–27 — Almonty's Sangdong operation — represents roughly 7% of global supply at full capacity.

The Pentagon knows. The day before strikes began, the DIB Consortium — 1,500 defence industrial base companies — was asked for proposals to secure the 13 minerals at $100–500 million per project. A $12 billion critical minerals stockpile was launched via the Export-Import Bank, but none of it arrives in time for this war; mining projects require 3–7 years for permitting and construction, and processing plants take years after that. Magnet factories and the certification chain between them add another layer. The DFARS prohibition on Chinese magnets and tungsten in US military equipment takes effect 1 January 2027 — industry experts have called the deadline "nearly impossible to achieve."

The war is consuming munitions at rates that do not wait. RUSI's assessment: 11,294 munitions expended in the first 16 days, at a replacement cost of $26 billion. The US consumed 82% of its JASSM-ER stockpile in six weeks — 1,875 of 2,300 missiles — against a production rate of 396 per year, the replacement timeline at peacetime production is measured in years. The Pentagon's "quadruple production" order, against existing Lockheed PrSM contracts calling for 335 units by 2029 with a 2–3 year contract-to-delivery lag, is an order to ramp a line that delivers 54 units in 2026 and 208 in 2028. Quadrupling a number that small does not solve a war that has consumed thousands.

RTX CEO Christopher Calio has said publicly the company will not scale production without additional funding and firm purchase commitments — signalling that industry does not trust the demand signal to persist beyond the current conflict. Missile manufacturers, including Anduril and Lockheed, have told the Munitions Acceleration Council that government targets are not realistic.

The World War II analogy has been widely invoked — Ford's Willow Run plant converted to B-24 bomber production in 18 months, one bomber every 63 minutes at peak, and the DPA has been mentioned in the same frame. The analogy is inverted, and the inversion is the point.

In 1942, America converted factories because it controlled its materials. In 2026, America cannot convert factories because it does not control its materials. The automaker talks with GM and Ford are a factory-retooling proposal applied to a materials-shortage problem. GM is already contracted to build infantry squad vehicles for the DoD, but the distance from light tactical vehicles to guided missiles is not a production line question, it’s about rare earths, tungsten carbide, gallium-nitride modules, specialty alloys, and a certification chain; none of which exists domestically at scale.

The war is consuming weapons made from materials controlled by the country benefiting from the war's continuation. That sentence was the previous edition’s framing, and Day 50 has not weakened it. China's domestic rare earth magnet capacity is expanding. The sulphuric acid ban from 1 May adds the processing input the remaining reserves would need. CNN reported on 11 April that US intelligence indicates China is preparing to deliver MANPADs to Iran within weeks — routed through third countries, but at the level of direct government transfer. The asymmetry is operational: the US consumed 82% of its most capable standoff missile in six weeks; China can resupply the weapon category that actually worked against US aircraft in days, at negligible unit cost.

What breaks the thesis is a material Chinese policy reversal on tungsten, magnets, acid, and the export controls across the rest of the 13 list — within a window short enough to matter to the current force structure. None of the signals as of Day 50 point that way. The default regime from 1 May forward is acid retention, rare earths are controlled rather than restored, and tungsten exports sit at near-zero volume.

Until that changes, the Pentagon's industrial mobilisation push is real in ambition and imaginary in execution. When ambition meets chemistry, chemistry wins.

Edition 3 flagged food as a pressure point — the IRGC target list included Saudi and Qatari petrochemical capacity, the Red Sea had not yet been stress-tested on grain, and GCC states were already airlifting staples and the previous edition upgraded the vector to CONFIRMED on the back of CPI gasoline at +21.2% and a 107-year low in US winter wheat plantings. This edition marks the arrival of that thesis.

The biological clock has passed: the second fertiliser application window for US corn closed in mid-April with retail nitrogen at multi-year highs. Farmers have already made their 2026 planting decisions and they have already applied — or under-applied — their fertiliser, and that crop is now being determined in the fields. No ceasefire reverses what has already been planted, and no diplomatic outcome re-applies nitrogen to acres that went in light, the food-inflation tail this generates runs through 2027 on seed-to-shelf cycles of 9–24 months.

DTN's fourth-week-of-March retail fertiliser survey is the single cleanest proof of the thesis in this edition: 5 broken records in one report. Anhydrous ammonia at $1,035/t — the first print above $1,000 since April 2023, up 20% MoM, urea at $826/t — the first print above $800 since November 2022, up 35% MoM, MAP at $906/t and DAP at $857/t, both at multi-year highs, UAN32 at $558/t, up 20% MoM and the CBOT May 2026 urea futures contract closed at $681/t on 19 March, up 68% from end-February — the wholesale spike is running ahead of retail catch-up, and the retail catch-up is what hits the farmer's cost base.

The structural driver is a dual chokepoint and nobody is pricing it as such. China's sulphuric acid export ban takes effect 1 May, layered on top of a phosphate export ban that runs through August. Three of the top 10 global urea exporters ship through Hormuz — the strait that has been effectively closed since late February, Russian ammonium nitrate exports remain suspended following the 1 April gasoline export ban (AN is a byproduct of the same refining chain), and European nitrogen production is at approximately 75% of normal on sustained gas cost pressure. In short: all 3 major nitrogen sources are simultaneously disrupted and there isn’t a fourth one.

Farmers have already voted with their drills; the USDA's 31 March Prospective Plantings report — surveyed during the first two weeks of March, with urea already up 50% from pre-war and diesel above $5/gal — shows 2026 corn acreage at 95.338 million acres, down 3% YoY and 3.45 million acres below 2025, soy acreage is up 4% to 84.7 million. The mirror is near exact and it is not a coincidence, soybeans are legumes; they fix atmospheric nitrogen through root nodules and do not need urea. When the price of nitrogen doubles in the weeks before planting decisions lock in, rational farmers do not abandon acreage — they rotate toward the crop that does not require the input whose price has doubled and 73,800 surveyed operators told the USDA that is what they intend to do.

The fertiliser disruption is expressing itself not as empty fields but as a fundamental shift in what those fields grow and the caloric implications are not neutral: corn yields approximately 180 bushels per acre at trend whilst soybeans yield approximately 50. A like-for-like swap of 3.45 million acres from corn to soybeans is a significant net reduction in total grain tonnage produced on US farmland, even though the total principal crop footprint barely moved, but what matters is not how many acres are planted, it is what grows on them.

Wheat and rice are the sharper signals, with all wheat acreage at 43.8 million acres — the lowest in the USDA's series, which extends back to 1919 — a 107-year low.

Spring wheat specifically is down 6%, with North Dakota cutting 8% from last year; these are decisions made in April and May, squarely inside the disruption window, under full knowledge of urea and diesel costs. If realised, that would be the lowest spring wheat area since 1970. Rice is also down at 18% nationally, with Arkansas — the largest long-grain producer — cutting 24% to its lowest level since 1983, Texas rice is at its lowest since the data series began in 1929 and Mississippi its lowest since 1973. Rice is both nitrogen-intensive and water-intensive, in states facing the worst drought conditions in years — Louisiana topsoil was rated 59% short to very short at the end of February, Arkansas at 53%. When both fertiliser and water are scarce, farmers do not plant rice.

Under-application on what does get planted is the invisible parallel move, Farmdoc Daily's recalibration of the MRTN (maximum return to nitrogen) model at the new nitrogen-to-corn price ratio implies systematic under application across Midwest corn country. Yield impact from a 10–15% N reduction or a 2–4 week application delay runs 10–25% for corn. That damage is the consequence of decisions that have already been made in the field, invisible until the USDA's July, August, and September yield reports publish.

Feed cost is the transmission mechanism and the US beef herd is where it lands.

Live cattle futures are at 250¢/lb as of April 2026 — an all-time high, 163% above the 2020 COVID trough and 52% above the 2015 cycle peak. The chart above is useful for two reasons: first, the structural run-up predates the war, US cattle inventory has been compressing for years, and the 2022 fertiliser spike produced a documented livestock cascade where ranchers culled breeding stock to avoid carrying loss making animals through winter, contracting the herd by 13.3%. A heifer culled in 2022 produces no calf in 2023, no yearling in 2024, and no finished animal in 2025, which is why the price line kept climbing even after feed costs eased. 2022 is still visible in the 2026 price.

Second, the 2022 shock is the comparison the current one needs to beat. The anhydrous ammonia price reached approximately $1,500/t at the 2022 peak; the current $1,035/t looks like a recovery on a chart alone, but the current price sits against a different rate environment — 2022 farmers locked in at pre-hike lending rates; the 2026 farmer is refinancing operating loans at 8–9%. It also sits against a different diesel price — the March CPI gasoline print at +21.2% MoM was the largest in the BLS series since 1967, and it sits against a herd that is already at multi-decade lows before the 2026 cull decision arrives.

The current cattle price reflects only the first-order transmission of feed cost expectations, and the breeding-stock cull that compounds it has not yet shown up, and historically does not show up until 6–12 months after the feed shock.

April CPI releases 12 May, and it will be the first full war-month reading. Pre-war USDA food inflation forecast for 2026 was 3.1%, the February print (all-food +3.1% YoY, food-at-home +2.4%) still reflected pre-war inventory, the March print (headline +3.3%, driven by gasoline +21.2%) was the first fuel-cost feed-through without fertiliser pass-through, April will be the first month in which elevated fertiliser and disrupted logistics both flow through to processor contracts.

PepsiCo's Q1 earnings on 16 April were the first executive confirmation that the food-processor cohort is repositioning: EPS $1.61 adjusted beat consensus; net sales $19.44B, up 8.5% YoY; critically, the first North American food volume growth in over two years — on a strategy of price cuts of up to 15% on Doritos, Lay's, Tostitos, and Cheetos. Ramon Laguarta flagged the global economy as "volatile and uncertain" because of the Middle East war, the executive playbook has shifted from pricing to volume defence. Private-label pressure from retailers is expected to eat into branded CPG margins, and Frito-Lay is closing plants in Florida, California, and New York.

What happens to CPG margins when input costs accelerate and pricing power collapses simultaneously is the Q2 earnings story across the sector.

IFPRI's analytical threshold is three months of war — after which planting decisions in the Southern Hemisphere mirror the constraints the Northern Hemisphere just priced in. The threshold crossing is imminent, and it will not produce an announcement, but a second round of fertiliser-shaped planting decisions from farmers in Argentina, Brazil, Australia, and South Africa, on a global nitrogen market that has not recovered from the first round. CSIS projects the food price peak at four months from the start of the conflict — late June — against an assumption of Hormuz normalisation that has not occurred.

The WFP +45 million acute hunger threshold triggers if the war runs past June, on top of 300 million people already food-insecure. Smallholders in Malawi, which sources 61.6% of its fertiliser from the Gulf, cannot outbid defence contractors or semiconductor fabs for sulphuric acid allocation.

The seven-way acid competition documented earlier has a margin hierarchy: metals outbid fertiliser and semiconductors outbid metals, leaving food at the bottom.

Every component of the next financial inflection point is now measurable: Japan's industrial production contracted 2.0% month-on-month in March, their PPI accelerated with oil and coal products up 7.7% and the BoJ meets on 27–28 April with markets pricing a 71% probability of a hike — into manufacturing that is contracting and inflation that is accelerating. The largest foreign holder of US Treasuries is being forced to choose which part of its mandate to sacrifice, in a market where Gold has overtaken Treasuries in central bank reserves for the first time since 1996 and capital has already found three alternative destinations.

The argument presented next is that the lock is structural, the loop is mechanical, and the data is now fully in the public domain.