Crude has retreated below seventy dollars and the headlines have moved on, but the price of the one refined barrel that airplanes actually burn has not agreed that the war is over. The second half opens with a warning printed in the jet crack spread, and eleven constraints that the ceasefire did not resolve.

The Warning — Aviation Fuel Says the War Is Not Over

Begin with the one price almost nobody is watching, because it is the one that tells the truth. West Texas Intermediate (“WTI”) crude closed the first session of the second half at $68.58, and Brent at $71.57 — both back to roughly where they traded before Operation Epic Fury (“OEF”) began on February 28. The interim accord signed in Switzerland on June 19 is holding, the Strait of Hormuz is reported at approximately 85% of pre-war throughput, and the market’s attention has moved on to payrolls and the Federal Reserve. The consensus read is that the energy shock is over and the second half can be traded as a normalization.

The jet fuel market disagrees, and the publication treats that disagreement as the single most important signal on the tape entering the third quarter. The argument of this section is simple and falsifiable: the barrel of crude has been told the war is over; the barrel of jet fuel has not gotten the message; and when those two prices disagree this violently, the jet fuel price is the one carrying the information.

Why the crack spread, and not the crude price, is the honest variable

Jet fuel is a kerosene-type middle distillate refined from crude. Its price decomposes into two parts: the cost of the crude feedstock, and the refining margin — the “crack spread” — that sits on top of the feedstock cost. That decomposition is the whole analytical key. When a shock is purely a crude-supply shock, the crack spread stays near normal and the jet price tracks crude both up and down; the refining machinery is intact, so the margin does not need to widen. When a shock also damages the refining and logistics of turning crude into kerosene, the crack spread blows out and stays wide even after the crude price recovers, because the bottleneck is no longer the oil in the ground — it is the capacity to refine and move the finished product.

That second configuration is exactly what the 2026 tape shows. In recent history the jet crack has averaged around $20 per barrel. Through 2026 it has run above $50, and McKinsey’s travel practice projects the full-year average could exceed $50 — more than double the historical norm. At the height of the disruption the North West European jet crack printed above $121 per barrel, more than four times normal. The U.S. Energy Information Administration (“EIA”) measured the U.S. Gulf Coast jet crack at roughly $1.25 per gallon in the March–May window, up from $0.42 at the start of the year. Crude has round-tripped nearly all the way back to its February level. The crack has not. That gap — recovered crude, still-elevated crack — is the market saying in the clearest available language that the physical refining damage from OEF has not been repaired.

The physical evidence corroborates the spread

A wide crack spread could in principle be noise. It is not, because the physical inventory and capacity data line up behind it precisely. Europe historically sources between a quarter and 40% of its jet fuel from Persian Gulf refineries; the region imports about 30% of its consumed kerosene through Hormuz. In mid-April the head of the IEA warned Europe had “maybe six weeks of jet fuel left,” and a Goldman Sachs research note flagged European inventories dipping below the IEA’s critical 23-day coverage threshold during June, with several localized markets already under 20–23 days — the level at which physical rationing and unplannable cancellations become, in the analysts’ word, mathematically inevitable.

The airline response is the corroborating behavior. Carriers cut roughly 13,000 flights and 2 million seats from their May 2026 schedules, the largest European reductions coming from Lufthansa and Turkish Airlines per Cirium. Looking across June through September, roughly 9.3 million seats have already been removed across eleven major markets, with more trimming expected. United cut summer capacity about 2%; Delta and American about 3% each versus pre-conflict plans. Spirit Airlines exited the market entirely, removing low-cost seat supply and pushing fares higher — which is, perversely, what unhedged U.S. carriers need to survive the fuel bill. None of this is the behavior of an industry that believes the shock has passed. It is capacity rationing against a supply line the industry does not trust.

The profitability arithmetic makes the stakes concrete. The International Air Transport Association (“IATA”) expects global airline profits to halve from $45 billion in 2025 to $23 billion in 2026, with net margin falling from 4.2% to 2.0% and net profit per passenger dropping from $9.10 to $4.50. Jet fuel’s share of operating expense is set to rise to 31.4% from 25.4%. IATA put the projected 2026 average jet fuel price near $152 per barrel, almost 70% above the 2025 average, and singled out the crack spread as the specific exposure most airlines carry unhedged, because they hedge crude — the liquid market — and are left naked to the refining margin. That is the exposure the publication is short.

The structural layer: this does not fully reverse when the shooting stops

The most important and least appreciated point is that a meaningful part of the jet-fuel constraint is structural rather than geopolitical, and therefore does not unwind on a ceasefire. Seven U.S. refineries have permanently closed or converted — Philadelphia Energy Solutions (335,000 barrels per day), LyondellBasell Houston (263,776), Phillips 66 Los Angeles (138,700), and Valero Benicia (170,000) among them — and no new U.S. refinery capacity was added in 2024, leaving 132 operable facilities as of January 2025. The Gulf Coast (Petroleum Administration for Defense District 3) produces roughly 998,000 barrels per day of jet fuel; the entire East Coast (District 1) produces about 84,000, less than 9% of Gulf output, leaving the eastern seaboard structurally dependent on Gulf shipment under Jones Act economics that add $5–6 per barrel versus foreign-flag alternatives.

The consequence is that even if crude falls further, the refining margin cannot compress the way it normally would, because physical throughput cannot rise to meet demand — the refineries that would do the compressing no longer exist. U.S. refiners did respond to the fat margins: four-week average jet fuel production surpassed 2.0 million barrels per day for the first time on record in the week ending May 1, and much of the extra output was exported to Europe and Asia at premiums. Domestic inventories rebuilt from a late-March low near 41.2 million barrels to about 45 million by late May. But that supply response is a ceiling, not a fix: it maximizes yield from existing crude runs rather than adding capacity, and it cannot serve Europe and Asia’s deficits and rebuild domestic buffers simultaneously without holding the crack wide.

How the OEF book expresses the warning

The book carries this read in two directions. On the demand side, the airline shorts are short precisely the margins the crack spread compresses: JetBlue (JBLU, $5.92) inside P24, Delta (DAL, $93.06) inside P31, and the International Consolidated Airlines Group American Depositary Receipt (ICAGY, $12.47) inside P33. Airline unit economics deteriorate directly with the crack because carriers hedge crude and wear the margin; every dollar the crack holds above normal is a dollar of unhedged cost the equity absorbs. On the supply-and-logistics side, Scorpio Tankers (STNG, $69.53) — the long leg of P33 — is a refined-product tanker owner that benefits from exactly the dislocation the crack prices: when Gulf refined-product exports are curtailed, Atlantic Basin and U.S. Gulf barrels must move longer distances to fill Europe and Asia’s deficits, and that ton-mile demand is the product-tanker bid. P33 is, in effect, long the physical relocation of jet fuel and short the airline that has to buy it — a single pair expressing both sides of the crack-spread thesis.

The broader tanker complex frames the same trade: Frontline (FRO, $34.70) on the crude side, the product carriers on the refined side. The publication’s standing position is that airline-margin compression has further to run and the product-tanker bid is more durable than the crude tape suggests, because the crack spread — not the crude price — is the variable that governs both.

Eleven Constraints the Ceasefire Did Not Resolve

The aviation-fuel warning is the sharpest instance of a general pattern: the second quarter’s supply shocks created physical bottlenecks whose lead times run far longer than any diplomatic timetable, and whose resolution the ceasefire has not even begun. What follows is the publication’s constraint-theme review, updated to the July 1 settled close. Five themes are treated in depth; six more briefly. Each closes with where it stands now and, where relevant, how the book expresses it. The arc ends with the demand sink into which every one of these constraints ultimately feeds — the memory-semiconductor complex — and then with the sovereign uncertainty none of it has resolved.

Deep Dive 1 — Helium: the OEF-Native Constraint

Helium is the constraint most directly authored by Operation Epic Fury, and the one whose supply response is measured in years rather than months. On February 28 an Iranian drone strike hit Qatar’s Ras Laffan complex, which together with Qatar’s broader output supplies roughly 30% of the world’s helium. The strike is a direct OEF read-through: not a second-order commodity effect but a primary act of the war itself, aimed at infrastructure whose loss ripples straight into the global semiconductor supply chain.

Why helium cannot be substituted or quickly replaced. Helium’s criticality is physical and non-negotiable. It cools the extreme-ultraviolet (“EUV”) lithography machines that pattern advanced logic and memory, carrying roughly six times the thermal conductivity of nitrogen, and it cannot be replicated at scale by any alternative gas. The 3-nanometer and 5-nanometer process nodes used in Nvidia’s Blackwell and Rubin accelerators consume significantly more helium per wafer than older processes, so demand intensity is rising exactly as supply is disrupted. New extraction and liquefaction infrastructure takes a minimum of two to three years to build from a standing start; SK Hynix’s chairman, Chey Tae-won, has put the AI-driven memory shortage running to 2030, which happens to coincide with the expected completion of Qatar’s facility repairs.

The customer exposure. Samsung Electronics and SK Hynix sourced 64% of their helium from Qatar in 2025. Memory manufacturers entered the summer running down roughly six-month stockpiles set to expire across June and July 2026. The U.S. distributor Airgas declared force majeure on helium shipments in April. That combination — expiring buffers, force majeure, and a two-to-three-year rebuild — is what makes helium a binding constraint rather than a passing spike: the buffer runs out on a clock the ceasefire cannot reset.

Where it stands now. Ras Laffan restarted after the June 23 incident, but reporting places it near a quarter of nominal capacity. Air Products (APD, $306.40) and Linde (LIN, $533.55) — the two majors that between them control the bulk of merchant helium supply and distribution — retain absolute pricing power over semiconductor customers who have neither an alternative source nor a substitute molecule. Air Products beat on its fiscal first quarter earlier in the year and raised guidance toward a $13.00–$13.25 range on exactly this dynamic. Regional helium pricing during the acute phase ran to multiples of pre-war levels. The theme is not in the book as a discrete pair today, but it is the upstream cause of the memory re-rating that dominates the closing deep dive, and APD or LIN remain the cleanest direct expressions should the desk choose to add helium exposure ahead of the next stockpile-expiry cycle.

Deep Dive 2 — Silver: Paper versus Physical

The Silver Paper-versus-Physical (“PvP”) divergence is the publication’s own long-running thesis and the analytical bridge between the commodity constraints and the sovereign-credibility thread that closes the edition. The core claim is that the paper silver price set in the futures market has been structurally decoupling from the physical metal, a decoupling visible in three places: depleted Commodity Exchange (“COMEX”) registered inventory, persistent backwardation in the futures curve, and delivery stress around First Notice Day. Silver is simultaneously an industrial input — solar photovoltaics, electronics, and, increasingly, the same electrification build-out that drives copper — and a monetary-credibility hedge, which is exactly why it belongs in both halves of this review and why its signal is harder to dismiss than a single-use commodity’s would be.

The Q1 template and what changed. During the first quarter the market began to believe the squeeze when COMEX registered inventory dropped below roughly 40 million ounces — the level at which the deliverable float looked thin against open delivery demand. Backwardation, where spot trades above deferred futures, is the curve’s signature of physical tightness; its persistence through Q2 is the evidence the tightness was not a one-off. The London Bullion Market Association’s unencumbered metal was the complementary indicator on the physical side. Together they framed a market where paper claims on silver exceeded readily deliverable metal by a widening margin.

Where it stands now. Spot silver closed July 1 at $60.08 per ounce, having pushed through the $60 handle — a level that changes the character of the trade from “accumulate cheap insurance” to “manage timing risk on a crowded move.” The silver-miner complex still screens cheap against the metal on the desk’s June work: Fortuna (FSM, $8.35) at a forward price-to-earnings near 5.3, Coeur (CDE, $16.54) near 8.5, and Pan American (PAAS, $44.33) near 9.1 were flagged as the cheapest primaries; First Majestic (AG, $17.04) and Hecla (HL, $15.59) complete the primary set. The metal proxy iShares Silver Trust (SLV, $53.58) and the miner basket Global X Silver Miners (SIL, $77.19) frame the two cleanest non-single-name expressions. The reason the miners can still be cheap with spot above $60 is operating leverage: a primary miner’s earnings rise faster than the metal once spot clears all-in sustaining cost, so a lagging miner multiple against a fast-moving metal is precisely the setup the desk screens for.

The book’s existing exposure and the concentration caution. The book already carries one silver-adjacent line through P17 (Sibanye-Stillwater, SBSW $8.51, long / Honda short), a platinum-group-metals primary with silver and gold credits — a separate thesis, but one that is metals-cycle correlated, which means adding a primary silver miner would concentrate rather than diversify the book’s metals exposure. The critical unresolved data point before any position is what happened to COMEX registered inventory and LBMA unencumbered silver through June and into early July; those were the leading indicators during the Q1 stress, and a break above $55–$60 spot without a confirming inventory draw would raise the risk of chasing a move that has already discounted the thesis.

Deep Dive 3 — Copper and the Smelting-Capacity Gap: the Integrating Constraint

Copper is the theme that ties the entire constraint arc to its root cause — electricity — and it is where the book carries its most explicit metals exposure. The distinctive feature since late February has been the copper paradox: record or near-record prices coexisting with elevated visible inventories. Copper closed July 1 at $6.12 per pound on the COMEX front month. In an ordinary market, high inventories cap price; here they have not, because the visible surplus is “earmarked” — committed to sovereigns building strategic stockpiles and to hyperscalers securing supply for data-center build-outs — rather than freely floating and available to clear the market.

The refining bottleneck beneath the price. Underneath the paradox sits a smelting-and-refining concentration that no mine expansion fixes quickly. China controls an estimated 97% of global copper smelting and refining capacity per industry reporting, and is the top refiner for 19 of 20 strategic minerals, at roughly 70% average market share. Treatment and refining charges — the fees smelters earn to process concentrate — collapsed to record lows across 2023–2025; some benchmarks hit roughly negative $26 per ton in early 2025, meaning smelters were paying miners for the right to process, and the 2026 benchmark settled near $0 per ton. That collapse is the market pricing a shortage of concentrate against a glut of smelting capacity concentrated in one country — a structural fragility, not a cyclical one.

The demand vector and the supply lag. The International Energy Agency projects copper demand rising roughly 50% from 2025 to 2040, against a potential 30% supply shortfall by 2035 from the current mining pipeline. Average copper ore grades have declined 40% since 1990, energy consumption per ton of production has risen 16% over the past decade, and the timeline from discovery to production now averages 12–15 years, up from 7–10 in the 1990s. A shortage with a fifteen-year lead time on new supply is the definition of a structural constraint.

Where the book stands. Two copper pairs are live. P15 (Freeport-McMoRan, FCX $60.53, long / Aptiv short) is working as intended — a U.S.-exposed miner against an auto-parts short that also captures the demand-destruction read. P30 (Southern Copper, SCCO $168.80, long / Teck Resources short, TECK $59.35) is the problem child: SCCO carried crowded short interest near 11.6% and Teck’s revenue growth (reported above 70% year-on-year) has outrun the long leg, leaving the pair marked negative and flagged as a candidate for rotation. The copper-miner basket Global X Copper Miners (COPX, $75.29) and the metal proxy United States Copper Index Fund (CPER, $37.21) frame the cleaner expressions if P30 is restructured — for instance, replacing the crowded SCCO long with the diversified COPX basket, or expressing the metal directly through CPER against an equity short.

Deep Dive 4 — UK Gilts and the LDI Transmission into U.S. Treasuries

The strongest theme in the sovereign cluster is the United Kingdom long-end, because it is the clearest live demonstration of the two macro frameworks the publication has developed — the Bond Vigilante Framework and the Global Sovereign Credibility Event — and because its plumbing transmits directly into the U.S. long end that the whole edition ultimately circles back to. UK 30-year gilt yields pushed to a 28-year high during the second quarter, printing intraday near 5.78% in May, the highest since 1998, up roughly 33 basis points year-on-year. That move was not idiosyncratic: it was part of the coordinated multi-sovereign backup the publication dated May 18 as the Global Sovereign Credibility Event, when the Japanese 30-year government bond hit an all-time high (its series dating to 1999), the 10-year JGB reached levels unseen since 1997, the German Bund 10-year touched 2011 highs, and the gilt 30-year printed its highest since 1998 — all on the same session, with G7 finance ministers convening in response.

The Bond Vigilante lens. The framework the publication applies here has a specific historical anchor. A bond-vigilante episode is the bond market refusing to ratify a central-bank stance it judges not credible against the fiscal and inflation math — forcing yields higher until policy or fiscal posture bends. The canonical instance is 1987, when the U.S. 30-year ran from roughly 7.5% to 10.2% between spring and October before the October 19 crash. The publication carries that episode as more than an analogy: it is lived desk experience from the Lehman Brothers mortgage-and-asset-backed trading desk of that era, which is why the framework is applied with the caution of someone who watched a vigilante move end in a crash rather than a soft landing. The 2026 configuration — hot inflation prints, a new Fed chair, and fiscal arithmetic no major sovereign can comfortably fund — is the modern rhyme.

The transmission mechanism into U.S. Treasuries. The reason a gilt move is a U.S. problem runs through United Kingdom liability-driven investment (“LDI”) pension structures. UK defined-benefit pensions hedge their long-dated liabilities using leveraged gilt and derivative positions operating on a collateral-buffer regime rebuilt after the 2022 mini-budget crisis, when a gilt spike forced a Bank of England intervention. When gilt yields spike, LDI funds face collateral calls and are forced to sell duration to raise cash. Because many hedge through cross-border duration carry and U.S. Treasury futures proxies, a gilt shock transmits into the U.S. long end through three legs: direct cross-border selling of Treasuries, hedge-ratio mechanics that force proxy selling, and sentiment contagion. The sector runs to thousands of schemes with the largest managers controlling the bulk of assets; the January 2025 cash-call episode was the reminder that the machinery is still loaded even after the post-2022 buffer rebuild.

Where it stands now. The U.S. 10-year par yield backed up to 4.48% into the July 1 close and the 30-year to 4.97%, with the sterling proxy Invesco CurrencyShares British Pound (FXB) at $127.56. The publication’s reading is that the long end is structurally vulnerable to a coordinated repricing precisely because no major sovereign can fiscally accommodate one, and the UK is the most probable trigger given LDI mechanics layered on top of its fiscal arithmetic. This is not a pair in the book; it is the macro overlay that sets the discount rate applied to every long-duration position the book holds — and it is the thread the closing sovereign section picks back up.

Deep Dive 5 — Yen/Dollar and the First Prints of Demand Destruction

The yen/dollar theme is where the supply shock stops being a supply story and becomes a demand-destruction story — the consumer-and-corporate-end consequence that the rest of the review only implies. The anchor is Honda’s first annual loss in nearly seventy years, reported in mid-May: an operating loss near ¥414 billion against a ¥1.2 trillion prior-year profit, accompanied by a writedown of roughly ¥2.5 trillion ($15.7 billion) spread over two years, of which ¥313 billion was explicitly quantified as Middle East cost pass-through. Honda held its dividend at ¥70 and its shares actually rose on the print, cushioned by the motorcycle business and the dividend pledge — but the record itself is the signal: a seventy-year loss threshold was crossed. Ford ($19.5 billion) and Stellantis (€26 billion) logged comparable writedowns in the same window; Honda’s was the one that broke the multi-decade record.

Why the yen carries the theme. A weak yen amplifies imported-energy and imported-input costs for a Japanese manufacturer that sells globally but pays for energy in a depreciating currency — the exact squeeze that turns an energy shock into a corporate loss. And the Bank of Japan sits at the center of the same sovereign-credibility question as the gilt story: the 30-year JGB printed an all-time high in the May 18 event and the 10-year reached levels unseen since 1997, which constrains how far the BoJ can lean against yen weakness without detonating its own long end. The yen is therefore not just an FX cross; it is the pressure gauge on a central bank trapped between a weakening currency and an unanchored long bond. The yen proxy Invesco CurrencyShares Japanese Yen (FXY) closed at $56.43.

How the book expresses it. Honda’s American Depositary Receipt (HMC, $27.26) is live as the short leg of P17 against Sibanye-Stillwater. That pairing is deliberate and does double duty: it expresses the demand-destruction read directly — short a yen-exposed auto manufacturer absorbing energy pass-through — while the long leg captures the metals-cycle and silver-credit thesis from Deep Dive 2. A single book line thus stitches the yen/demand-destruction theme to the silver PvP theme, which is why P17 recurs across this review.

The broader read. Demand destruction printing at the retail and corporate end — Honda’s loss, the airline capacity cuts documented in the aviation warning, the diesel and heating-oil stress across the northern-tier states earlier in the quarter — is the mechanism by which a supply shock becomes a stagflation regime rather than a one-off spike. Supply-side inflation that destroys demand does not resolve cleanly; it grinds. The yen is the cleanest single FX read on that transition, and Honda’s seventy-year record is its most legible corporate marker.

Six More Constraints, in Brief

Fertilizer — Nitrogen and Ammonia

The Middle East supplies roughly 30% of globally traded ammonia and 35% of globally traded urea; the Hormuz closure halted those flows and QatarEnergy suspended natural-gas-derived urea, methanol, and polymer output. Roughly 20% of European ammonia and 25% of European urea capacity was curtailed, and India — dependent on the region for about half its liquefied natural gas and a fifth of its urea — saw regional imports fall more than 51% in March. New nitrogen capacity takes three to four years to build. CF Industries (CF, $108.16) is the cleanest primary: 96% five-year average utilization, a 2025 profit near $1.46 billion, and roughly 10% of its share count retired through buybacks. It spiked to $136 in March on the disruption and has settled back. Peers: Mosaic (MOS, $21.30), Nutrien (NTR, $63.57), Intrepid (IPI, $33.51). Watch: the European natural-gas-to-ammonia spread and CF’s next utilization print.

Water Desalination and Reverse Osmosis

The March 29 Kuwait co-generation strike exposed the Gulf’s desalination vulnerability: co-generation plants produce power and water from the same gas turbine, so a single strike destroys both, and the gas turbine is the binding constraint on restoration with 24-to-48-month replacement lead times. Energy Recovery (ERII, $8.90) holds roughly 80% global share in the pressure-exchanger energy-recovery devices that reverse-osmosis desalination requires, with six-to-eighteen-month replacement lead times of its own; it is the long leg of P5. Flowserve (FLS, $72.38) and Xylem (XYL, $117.27, long leg of P4) round out the water-infrastructure expression. Watch: Gulf reconstruction contract awards and ERII’s book-to-bill.

Grid-Scale Gas Turbines

The gas turbine is the constraint that sits behind both the desalination rebuild and the artificial-intelligence data-center buildout — the same 24-to-48-month lead-time equipment feeds both, and they compete for the same order slots. GE Vernova (GEV, $1134.35) is the pure-play and the long leg of P26 against the energy-sector basket; the parent-scale comparison is GE Aerospace (GE, $374.94). The demand driver is unambiguous: every megawatt of new baseload for a hyperscaler or a desalination train needs turbine capacity that is sold out years forward. Watch: GEV’s late-July earnings and the forward-order backlog disclosure.

Grain-Oriented Electrical Steel (GOES)

Grain-oriented electrical steel is the specialty steel used in transformer cores, and Cleveland-Cliffs (CLF, $9.42) is the sole domestic U.S. producer. Transformer lead times have blown out from roughly 12 months pre-2022 to 60-plus months, driven by the same AI-capex-meets-physical-supply dynamic. The theme is expressed as the three-pair GOES cluster from Tranche 5: the monopolist (CLF long / Nucor short, NUE $219.02, P25), and the global non-Chinese alternative (POSCO, PKX $50.01, long / the steel ETF short, P38). ArcelorMittal (MT, $59.26) and Steel Dynamics (STLD, $221.76) frame the peer set; United States Steel has been acquired by Nippon Steel and is no longer independently listed. Watch: transformer order lead times and any GOES capacity announcement, which CLF’s monopoly makes unlikely near-term.

Austrian Century Bonds — Duration Math Made Visible

The Austrian century bonds are carried not as a position but as the cleanest illustration of what the sovereign repricing does to long-duration capital. The 2117 and 2120 issues (the latter with a modified duration near 36) lose enormous price for each basis point of yield backup; the 2120 bond’s move from its issue yield to the current ~3.26% band is a multi-decade capital loss made visible in a single instrument. They are the teaching example for why the gilt and JGB prints in the deep-dive sovereign section matter: duration bites hardest at the very long end, and the century bonds are the purest available measure of the bite. Watch: the 2120 issue’s yield against the Bund 30-year as the cleanest read on eurozone long-duration stress.

Platinum-Group and Specialty Metals

The platinum-group-metals (“PGM”) strand runs through Sibanye-Stillwater (SBSW, $8.51), the long leg of P17 — a PGM primary with silver and gold credits that links the specialty-metals theme to both the silver PvP thesis and the yen/Honda demand-destruction pair on the other side of that same position. PGMs sit at the intersection of auto-catalyst demand (pressured by the same auto-sector weakness Honda embodies) and constrained primary supply concentrated in a few producers. It is the smallest of the metals threads but the one that stitches three others together in a single book line. Watch: the platinum-palladium ratio and Sibanye’s cost-curve position as the auto cycle turns.

The Demand Sink — Where Every Constraint Terminates: Memory and the Semiconductor Complex

The eleven constraints above are supply-side stories. They matter because of a single demand sink into which all of them ultimately drain: the artificial-intelligence build-out, and at its physical center, the memory-semiconductor complex. This is the theme the publication has returned to more than any other across the quarter — the “bottleneck migrated from crude to memory” spine of late June — and it is the correct note to end the constraint arc on, because it is where the constraints are priced.

The through-line. Helium cools the EUV machines that pattern the memory. Grain-oriented electrical steel and gas turbines supply and deliver the electricity that powers the fabs and the data centers that consume the memory. Copper wires all of it. Every constraint in this review is, at one remove, a tax on the same build-out — and the memory makers are where that build-out shows up as revenue and as scarcity rent. During the quarter Micron crossed a trillion-dollar market capitalization and SK Hynix crossed a trillion in the Asian session; the memory re-rating was the single largest driver of the book’s gains before the June 30 harvest.

Where it stands at the July 1 close. Micron (MU) closed at $1032.28, a record, having run through the quarter on high-bandwidth-memory (“HBM”) demand for AI accelerators. The logic and accelerator complex around it: Nvidia (NVDA) $197.58, Advanced Micro Devices (AMD) $540.88, Broadcom (AVGO) $369.34, Marvell (MRVL) $272.05. The storage adjacency that the same AI-data-center demand pulled higher: Western Digital (WDC) $598.37 and Seagate (STX) $915.19, both among the quarter’s standout winners as hyperscaler storage orders surged. And the laggard the book is short: Intel (INTC) $127.02, which ran hard into quarter-end and is the short leg of the newly opened P39 (long Broadcom / short Intel) — the clean relative-value expression that replaced the structurally-impaired SOXS constructions at the June 30 close.

The tension the second half inherits. The memory complex is priced for the constraints to bind — for helium, power, and packaging to keep supply scarce while AI demand compounds. That is a coherent thesis and it has paid. The risk is the mirror image of the aviation-fuel warning: just as the crude tape can lie about the war being over, the memory tape can lie about the build-out being infinite. When a demand sink is priced for permanent scarcity, the danger is not that the constraints resolve — they will not, on these lead times — but that the demand itself proves cyclical. The June 30 harvest of the most extended names (Corning, Generac, Micron) was the book’s answer to that tension: bank the scarcity rent while it is unrealized, keep the clean relative-value expression (P39), and let the constraints do the rest.

The Unresolved Thread — Sovereign Credibility and What the Second Half Cannot Yet Price

The constraints are physical and their lead times are known. The uncertainty that the second half genuinely cannot price is not physical but institutional: whether the sovereign long end holds, and whether central-bank credibility survives the fiscal arithmetic now visible on every major curve.

The Global Sovereign Credibility Event of May 18 remains the defining instance — Japanese, German, and UK long yields printing at multi-decade or all-time highs on the same session, the G7 finance ministers convening, and the reading that no major economy can fiscally accommodate a coordinated repricing. The gilt/LDI transmission mechanism in Deep Dive 4 is the plumbing through which such a repricing reaches the U.S. long end; the Austrian century bonds are the measure of what it does to duration; the yen and the Bank of Japan are the Asian face of the same question.

Layered on top is the domestic institutional question the quarter added: the Supreme Court’s June 29 ruling in Trump v. Cook, which preserved the President’s inability to remove a Federal Reserve governor except for serious misconduct while stripping Humphrey’s Executor protections from other independent agencies. Federal Reserve independence, in the words quoted in the June 30 edition, “lives on for another day, but is not as robust as it was.” With Chair Warsh’s first dot plot behind us and the funds rate held at 3.50–3.75%, the market is pricing a Fed that is independent but visibly closer to the political boundary than it was a quarter ago.

The synthesis. Put the pieces together and the second half opens with a coherent, uncomfortable picture. Supply constraints that will not resolve on any diplomatic timetable (the eleven themes). A demand sink priced for those constraints to bind permanently (memory). A refined-product market still flashing that the war is not over (aviation fuel). And a sovereign long end that no major government can afford to see reprice, defended by central banks whose credibility is being tested in the courts as well as the markets. None of these resolve in the third quarter. The book’s posture — realized register banked at +$487,933 after the quarter-end harvest, a leaner 24-pair active book, and the clean P39 semiconductor expression — is built for exactly this: keep the constraint exposure where it is cleanest, bank the scarcity rent where it is most extended, and watch the crack spread, the registered ounces, and the long-end print for the signal that any one of these threads has finally moved.

The truth, as a certain television series liked to say, is out there. This quarter, more of it than usual is hiding in the prices almost nobody quotes — the jet crack, the COMEX register, the 30-year gilt. The publication’s job in the second half is to keep quoting them.