Wednesday delivered Kevin Warsh’s first FOMC decision as Fed Chair, and it delivered a fractured Fed. The Committee held the federal funds rate at 3.50–3.75% on a 9–3 vote, with three regional presidents dissenting on hawkish grounds — the first three-vote dissent at an FOMC meeting in more than three decades. Bailey’s Bank of England followed Thursday with a 6–3 hold at 3.75%, more hawkish than the market’s 7–2 expectation. Lagarde’s ECB last week gave the same pattern in a different register: "some colleagues asked themselves" whether to hike. Three of the world’s largest central bank committees held their positions in nine days, and each held with a visibly hawkish minority. The unified central-bank posture that carried the market through 2020–2025 has, on this evidence, fractured across three currencies within a fortnight.
What broke the consensus is the split mandate this publication has documented since June. Thursday morning delivered the tell in three data points on the same tape. Advance Q2 US GDP printed at 1.5% — below the consensus, dragged by federal spending cuts and inventory drawdowns; core PCE cooled to 3.3% from May’s 3.4%; and by Friday, Brent oil had settled at $90.12, up roughly 20% on the month, after Iranian attacks on tankers in Hormuz and a Ukrainian strike on Russia’s Volgograd refinery. Demand rolling over. Supply-side pressure intensifying. The Currie mechanism from Issue 29 arrived in the tape at precisely the moment the Fed had to decide, and it arrived on both sides of the mandate simultaneously. No monetary tool reconciles both.
The bond market’s response was more coherent than the committees’ own. The US 30-year yield spiked nine basis points to 5.19% on Fed day — a bear-steepening as the front end priced "held for now" while the long end priced "higher for longer, whether they act or not." UK Gilts held above 5.0% for the second consecutive week, their longest run at that level since 2008. Bunds and JGBs remained elevated. What the sovereign long end is saying, unanimously, is what the committees cannot say jointly: rates cannot stay this low against these fiscal deficits and these supply shocks. Term premium is expanding globally, and it is expanding fastest at the thirty-year point where the fiscal problem is priced most directly. When committees can no longer decide together, markets price policy themselves. That is what happened this week.
The specific mechanism deserves attention. Japan intervened in the yen market Thursday to prop up a currency at forty-year lows, sending the yen up 3.3% intraday to the 158 handle. Ueda’s BoJ voted 8–1 to hold at 1% Friday morning. Friday afternoon, US Treasury Secretary Bessent was photographed at a Camp David cabinet meeting with a notepad reading "To Do: Buy Japanese Yen $5–10 bil," and the Treasury informed banks through the New York Fed to "stand ready for future action." The yen closed the week at 157.40, its strongest since May, and Bessent’s subsequent statement referred to "close coordination" with Japanese authorities. What connects the intervention to Treasury yields is arithmetic: sustained yen defence requires selling dollar assets, and Japan holds roughly $1.1 trillion of US Treasuries as the natural source. There are two readings of the coordination available. The first is incidental — the intervention is defensive, the Treasury sell-off a side-effect that the Fed absorbs. The second is deliberate — coordinated FX as an alternative tightening tool that raises US long-end yields without a formal Fed move, achieving de facto restrictive policy while preserving Fed optionality and letting a fractured Committee avoid the vote it cannot yet take. This publication does not need to choose between them. Both readings point to the same instrument implication: US 30-year yields at 5.19% now reflect pressure that the Fed’s dot plot does not measure and the FOMC statement did not name.
Big Tech refined Issue 29’s argument in an important direction, and the refinement matters. Microsoft rose roughly 8% next day after defending $41 billion of Azure capex against Alphabet’s Wednesday problem. Meta jumped roughly 11% on 19% ad revenue growth carrying $125–$145 billion of full-year capex; the CFO explicitly declined to guide 2027. Amazon added 8–10% after AWS grew 37% at 39.4% operating margins, defending $220 billion of capital spending. Apple, notably, sold 3–4% — not on capex but on iPhone execution and margin pressure from memory costs. What the tape said was clear enough. The market was not repricing all AI capital expenditure. It was repricing capex that outpaced operating cash flow — Alphabet’s specific negative-free-cash-flow problem. Meta, Microsoft, and Amazon all demonstrated that capex funded by strong demand can carry secular multiples. Alphabet did not. The valuation clock’s strike from Issue 29 was, in the tape’s clarification, more selective than blanket — which is the more sophisticated repricing, and the harder one to fade.
The four clocks, reassessed. Valuation: selectively struck at Alphabet, not the complex. Liquidity: SPCX consolidates in the $110–$117 range as the August 6 lock-up wall approaches, with 911.5 million shares (roughly $116 billion) becoming tradeable in a single event; the FINRA leverage architecture is unchanged. Inventory: striking on a floor rather than a spike, with Brent up 20% in July and no exit in the Middle East. Private credit: still contained at the retail investor level, but Apollo’s daily-pricing rollout completes September 30, bringing external mark validation to $830 billion of credit assets — which is the mechanism by which the fourth clock moves from ticking behind quarterly caps to striking against a public price. Three-plus clocks in motion; the fourth’s sound has begun to travel. The tactical sequence remains flexible. The strategic direction, on this week’s evidence, is confirmed and refined: central banks cannot reconcile the split mandate; markets are pricing what committees cannot say; and rotation is now nuanced rather than blanket, but proceeding.
Bailey’s Bank of England held Bank Rate at 3.75% on Thursday by a 6–3 vote, more hawkish than the 7–2 the market had priced. The three dissenters wanted to hike; the six majority favoured continued patience. The statement pointed directly at the Middle East as the mechanism: "higher energy prices are likely to push inflation higher later this year," with risks "tilted to the upside." Bailey’s press conference echoed Lagarde’s of a week earlier and Warsh’s of the day before — the shape of the argument having become international. If Wednesday’s Fed vote was the first three-vote dissent since 1993, Thursday’s BoE vote was the sharpest hawkish tilt since March’s post-war shock, and both were held against a demand-destruction data pattern the majority could not quite bring itself to fade. The 10-year Gilt yield eased into the decision to just below 5.0%, ending the week near the same level it began — which is to say, near a two-month high, and above 5% for the second consecutive week for the first time since 2008.
What the register is now pricing, however, has evolved. Burnham’s first two weeks in office have neither produced the "public control of key sectors" referent flagged in Issue 29 nor the fiscal-envelope arithmetic to bound it. Healey’s Treasury team has begun quiet consultations on the Autumn Budget; the OBR verdict is the document that reconciles everything else. Meanwhile, oil prices near the Bloomberg-flagged $100 level and the highest 30-year Gilt yield since 1998 are combining in the way this publication first outlined at the leadership transition: sterling under pressure, mortgage swap rates elevated, roughly nine million UK households facing renewed fixed-rate repricing. The 10-year Gilt at 5.0% is now the highest sustained sovereign borrowing cost in the G-10, and it is being carried by an administration whose first policy speech dated the country’s wrong turns to the 1980s. The register is pricing not what the Chancellor has said, but what he has not yet had to. The Autumn Budget is the document. Its ordering will read as its message.
The market’s response to Big Tech earnings was the week’s most instructive tape, because it separated the two theses this publication has been arguing between since June. Meta reported first on Wednesday after the close: revenue $60.8 billion, ad impressions +19%, ad pricing +12%, full-year capex guide held at $125–$145 billion, the CFO explicitly declining to guide 2027 capex. Stock +11%. Microsoft reported the same evening: Azure capex defended against demand and supply constraints, FY2027 capex projected at $255–$260 billion, useful-life extension on data-centre buildings from 15 to 25 years and lease reclassifications lowering the calendar-2026 headline by roughly $15 billion. Stock +8% next day, crossing the 200-day moving average for the first time since the May sell-off. Amazon Thursday: revenue $200.6 billion (first-ever above that threshold), AWS $42.2 billion (+37%, fastest growth in 18 quarters), AWS operating margin 39.4%, $220 billion in capex defended by the demand growth beneath it. Stock +8–10%. Apple sold 3–4% to roughly $319 despite a record June-quarter $109.4 billion revenue, on iPhone execution and margin pressure from memory costs — sold, notably, for reasons that have nothing to do with capex.
The tape’s verdict is analytically clean. What Alphabet’s Wednesday reaction seven days ago tested was not "AI capital expenditure" in the abstract. It was capex that outpaced operating cash flow — negative FCF for the first time since the 2004 IPO, a specific balance-sheet condition. Meta, Microsoft, and Amazon all show that when capex is funded by revenue growth compounding at these rates, and when operating margins are expanding rather than compressing, the market is willing to hold the multiples. When it is not — Alphabet last week, TSMC on July 16 — the multiples come down. The rotation this publication has been positioning around is not "into physical economy and out of AI"; it is into cash flows the market can price. Some of those cash flows are in AI infrastructure (Microsoft, Amazon), some in AI power demand (GE Vernova, Northrop, Lockheed), some in the disinflation trade the sovereign long end is preparing (long-dated Gilts). The concentrated growth premium is being unpicked selectively, name by name, and the winners of the AI capex complex are separating cleanly from the losers.
SPCX itself finished the week consolidating in a tight $110–$117 range after touching a fresh 52-week low of $109.53 on Monday. The August 6 lock-up expiry — releasing 911.5 million shares, roughly $116 billion of supply, in a single event — is now four trading days away, with Q2 earnings (the first-ever for the newly-public company) landing on August 4 in between. Consensus expects $6.75 billion in Q2 revenue and a loss per share of $0.26. Short interest against the stock has passed $26 billion; Musk has personally warned short sellers. This publication’s call from Issues 22 through 24 that the IPO was structurally mispriced — 4% float, MSCI mechanical inclusion, retail scarcity mechanics — has now been carried to the point where the reversion arithmetic is arriving in scheduled tranches rather than in argument. August 6 is the wall. September 10 is the next release. June 12, 2027 is when Musk’s own block unlocks.
The Middle East continued to escalate through the week, and the escalation continued to arrive at the refined-product layer rather than the raw-crude layer where headline prices are measured. WTI closed Friday around $80; Brent settled near $89, up roughly 20% on the month. Iranian Revolutionary Guard forces struck two tankers transiting Hormuz under US escort; four other tankers turned back. Ukraine’s campaign against Russian refining added Volgograd to the target list. Chevron’s Wirth told CNBC the threat had "expanded beyond Hormuz"; Exxon’s Woods conceded that the artery of Middle East supply had to reopen but did not say when. Saudi Arabia’s talks with 43 countries about a maritime coalition against Iran-backed Houthi disruptions in the Red Sea are the sovereign equivalent of asking who else will help. The Currie floor is being raised, and the July CPI print, when it arrives, will be measured entirely under this configuration.
| Gold | $4,065 | Support around 3,950; MACD remains bullish |
| Copper | 13,800 | Harsh correction then bullish momentum; MACD remains bullish — price recovered, outlook uncertain |
| WTI | $80 | Hormuz premium returns; refined-product spread remains the substantive tell |
| Brent | $89 | Up roughly 20% on the month; Ukrainian strike on Volgograd refinery adds to the Currie floor |
| Carbon | 81.25 | Still selling off; high volatility — support at the 200-day moving average |
| UST 10Y | 4.72% | 2Y at 4.26% — higher on Middle East fears and the Fed’s 9–3 vote split; 30Y touched 5.19% Fed day, higher-rates bias holds |
| UK Gilts | 5.04% | Second consecutive week above 5.0% (longest since 2008); BoE 6–3 hold Thursday eased slightly; new administration, political uncertainty — at the mercy of investors’ perception of value |
| Bund 10Y | 3.20% | Higher — ECB verbal hawkishness confirmed; markets primed for September projections meeting |
| JGB 10Y | 2.80% | BoJ 8–1 hold at 1% Friday; FX intervention with US Treasury coordination Thursday, carry trade under pressure — repatriation dynamic and Treasury coordination now the operating concern |