Inflation Cooled, and the Long End Got More Expensive

This report is based on intraday data as of 12:19 PM ET and does not reflect closing prices. Markets may have moved since publication.

Reviewing Yesterday’s Call

Yesterday’s brief nominated today’s June core PCE as the judge: if inflation reaccelerated while the front end still refused to price a hike, the “anti-inflation credibility discount” would graduate from a one-day event into a regime. Today delivered the opposite input — core PCE eased from 3.4% to 3.3%, and headline PCE fell 0.1% month-over-month, its first monthly decline in six years — and the long end rose anyway. So the test needs rewriting. The question is not “does gold rise when inflation runs hot,” it is “does the long end fall when inflation runs cool.” It did not.

Two corrections to yesterday’s figures are also owed. The VIX actually closed 7/29 at 20.66 (+11.9%), not 17.65 (-3.1%), and the Nasdaq 100 ETF actually closed down roughly 2.0%, not up. That directly overturns the first of three reasons yesterday’s brief gave for calling the gold move monetary rather than fear-driven: read against the correct closing data, 7/29 was a day of rising volatility and falling equities, and there genuinely was fear mixed into that day’s precious metals bid. Today’s version of the test is far cleaner: the VIX fell 11.7% to 18.25, the Nasdaq 100 ETF rose 3.1%, copper gained about 2.5% — and the dollar index broke below 100.

Today’s Core Judgment

Cool data bought a weaker dollar, higher equities and lower volatility. The one thing it could not buy was the long end: 5s, 10s and 30s rose 1.8, 3.7 and 6.3bp respectively, and the 30-year traded at 5.21%, above its 52-week closing high set on May 19. This is the same sentence as yesterday’s 9-3 vote, spoken by the opposite data.

Macro and Geopolitical Deep Dive

Calling this a GDP miss is almost the wrong description. Q2 real GDP grew at a 1.5% annual rate, below FactSet’s +2.1% and the Dow Jones survey’s +1.8%. But none of the drags sit in private domestic demand: net exports took off 1.01pp (Q1: 0.37pp), private inventory change another 0.67pp, and government spending fell 0.8% (Q1: +4.4%). Meanwhile consumer spending jumped from +0.5% to +3.2%, and the BEA’s “core GDP” — real final sales to private domestic purchasers — accelerated from +1.7% to +3.9%. US private domestic demand accelerated in Q2. The deceleration happened in the accounting.

At least two of those drags are self-reversing. Imports rose 11.5% against exports of just 4.5% — and with 100% pharmaceutical tariffs landing 7/31 and 50% tariffs on Canadian goods landing 8/19, pulling shipments forward ahead of the effective dates is rational behavior. That kind of import surge is a timing shift, not demand leaking abroad, so Q3 net exports should improve mechanically (the 8/4 June trade release is the first check). The government line deserves more attention still: the BEA’s technical note explains that the drop in federal nondefense consumption expenditures “primarily reflect[s] sales of crude oil from the Strategic Petroleum Reserve.” Reserve sales are netted out of government consumption in the national accounts, and because the crude itself shows up in other components there is no direct effect on total GDP. Put plainly: the policy currently suppressing the oil price is also making the sentence “government spending is contracting” read better than the underlying fiscal impulse warrants.

The sharpest detail in the whole release is inside the investment line: equipment +15.2%, nonresidential structures −5.0% — and that is the tenth consecutive quarterly contraction in structures. This capex cycle is buying machines, not buildings. Put that next to Microsoft’s decision to extend the depreciable life of data centers and office buildings from 15 to 25 years starting in FY27, while reclassifying more future leases from finance to operating leases (which are not counted as capex), and one plausible reading emerges: the shell around the compute is increasingly being rented rather than built, so it appears neither in structures investment nor in full in the reported capex figure. That is a hypothesis, not a conclusion — but it does explain how AI investment can be unprecedented in scale while nonresidential construction shrinks for ten straight quarters. Separately, residential investment rose 1.5%, its first increase in six quarters, and it did so with long-end yields at cycle highs. Worth tracking, not worth extrapolating.

The gap between nominal and real is today’s most direct read on Fiscal Dominance. Nominal GDP grew 7.9% against real growth of 1.5%, implying a deflator of roughly +6.3%. The 30-year Treasury yields 5.21%. Nominal growth still exceeds the long-term cost of borrowing by about 2.7 percentage points, so the debt arithmetic still works — but it works because the deflator is running above 6%, not because real growth is. That is the quantitative form of the claim that inflation has become a fiscal instrument, and it is why one soft month of core PCE buys the long end nothing. The long end is pricing that deflator, not that core.

Devil’s Advocate: saying “inflation cooled” is a selective statement today. Monthly core PCE did fall to 3.3% with a +0.1% monthly print, and goods prices rose 0.7% MoM against services at 0.3%. But inside the same GDP release, the quarterly PCE price index rose from +4.6% to +5.1% and the gross domestic purchases price index rose from +3.6% to +5.7%. The quarterly measures accelerated while the monthly core eased. So the long-end move could just be reading the quarterly deflator, plus supply positioning ahead of next week’s quarterly refunding statement, plus month-end index extension flows. All three are real. I still lean my own way for exactly one reason: curve shape. 5s rose only 1.8bp, 10s 3.7bp, 30s 6.3bp — monotonically increasing in maturity. If the market were repricing the policy path, the front end would have moved most. It didn’t. This is term premium, not policy expectations.

Two days, two central banks, the same three-vote hawkish minority, the same oil shock, and both majorities choosing to look through it. Yesterday the FOMC held 9-3 with all three dissenters calling for a hike. Today the Bank of England held Bank Rate at 3.75% by 6-3, with dissents rising from two to three as Catherine Mann, Megan Greene and Huw Pill all pushed for an immediate 25bp move to 4.00%. Note the stated reasoning: one member’s formulation was that “the key change in the environment for my decision is the collapse of the U.S.-Iran Memorandum of Understanding, the widening of the Middle East conflict and the associated volatility in energy prices.” Meanwhile UK June CPI came in at 2.6%, half a percentage point below the Bank’s own recent forecast. Those three are voting against a scenario, not against a data point — the Bank’s own scenarios put $100 oil at 3.2% UK inflation for 2026 and oil falling back toward $71 at roughly 3.0%, with a warning that further escalation could push next year’s inflation above 4%. Governor Bailey immediately leaned the other way, telling the public not to conclude that the Bank is edging toward a hike.

That structure is now symmetric across the Atlantic: a supply shock originating in a shipping chokepoint is generating hawkish minorities inside multiple central banks at once, while the majorities in both places decline to act on the grounds that this is not a monetary phenomenon. Markets are answering symmetrically too — both currencies fell, and metals were bid. That is not a coincidence. It is one reaction function.

The yen rose 3.5% today, but this is not a carry unwind, and the test is clean. Between roughly 09:50 and 11:15 ET, USD/JPY dropped about 480 pips to a low of 158.34, the largest single-day yen gain since December 2023. Analysts said the move “bore the hallmarks of official intervention”; Tokyo has confirmed nothing. The dollar index slid below 100 to around 99.98, its lowest since June 17.

A genuine carry unwind has three simultaneous signatures: the yen rises, risk assets fall, volatility rises. Today all three ran the other way — the VIX fell 11.7%, the Nasdaq 100 ETF rose 3.1%, the semiconductor ETF rose 6.8%. More decisive still is the direction of the spread: Japanese 2s, 10s and 30s fell 1.6, 2.6 and 3.9bp today while the US long end rose, taking the US-Japan 10-year spread out to roughly 190bp and the 30-year to about 127bp. If the market were pre-positioning for a hike at tomorrow’s Bank of Japan meeting, JGB yields would be rising alongside the yen. They moved the opposite way. A yen appreciating while its rate disadvantage widens has only one self-consistent explanation: this is the fiscal authority operating, not the market repricing. Which means it is reversible — Japan deployed roughly $70 billion in late April and early May and has about $1.09 trillion in reserves left, and from tomorrow the market will be counting that ammunition.

A correction to a headline that traveled widely today: “gold clears $4,100” refers to futures, not spot. What cleared that level was COMEX August gold, which traded at $4,130.90 at 8:33 a.m. ET. Spot gold stayed below $4,100 throughout — CNBC had it at $4,071.47 at 9:00 a.m. ET, Forbes at $4,079.90 at 8:45 a.m. ET with a 24-hour range of $3,996.32 to $4,114.04, and Kitco described spot as roughly flat and approaching rather than clearing $4,100, still inside the $4,000-4,200 band it has occupied for a month. This correction is not pedantry; it is evidence against my own weak-dollar, debasement-pricing lean above. On a day when the dollar index broke below 100, spot gold stayed mid-range and did not break out. What actually led today was industrial metals and silver (copper about +2.5%, spot silver around $58.61, +1.7%) — which looks more like a reflation-plus-weak-dollar trade than a pure bid against fiat.

Hormuz: today, “traffic is recovering” and “Iran controls the strait” are the same sentence. This is day 153 of the war. Kpler counted 12 two-way commodity vessels on 7/29 (Bloomberg, citing Kpler, put it at 14 — the two readings disagree, so both are listed by source). The New York Times says transits remain below a tenth of pre-war levels, and the IMF PortWatch seven-day moving average of tankers through the strait stands at 11 — roughly 8% of the ~138 vessels per day cited as the pre-war baseline. The genuinely new fact is a single LNG carrier: QatarEnergy-controlled Al Areesh exited the strait overnight on 7/29-30 with its transponder on, bound for Port Qasim in Pakistan, the first QatarEnergy LNG carrier out since Al Rekayyat was struck in early July. Iran’s Fars news agency says the vessel transited through Iran’s designated corridor, with Iranian permission; NBC, citing MarineTraffic, reported four additional vessels in the Iranian corridor the same day.

So Bloomberg’s “shipping volumes have recovered in recent days” and the institutionalization of a toll regime are two readings of one dataset. Deputy Foreign Minister Gharibabadi also spelled out how narrow the negotiating space is: Oman’s “50% Iranian control, 50% Omani control” routing does not address Iran’s concerns, and Iran demands full control of the inbound lane plus part of the outbound lane. Iran also rejected third-party mine clearance, insisting it will demine the strait itself, and stated that Hormuz “will never return to its pre-war state” and that Iran has made no request for talks with Washington in the past 15 days. The same day, Pakistan’s foreign ministry said negotiations “are under way.” The mediator and the principal are directly contradicting each other, and CNN notes the only publicly announced channel recently is the Iran-Oman deputy-minister-level talks on transit management. Militarily the escalation resumed: CENTCOM says it struck “dozens” of IRGC targets inside Iran on 7/30, and Iran answered with drones and missiles against Kuwait and Jordan, with Jordan’s military reporting five interceptions.

The toll model is spreading, and that matters more than any single day’s transit count. Reuters reports the Houthis are weighing a transit fee on commercial vessels through the southern Red Sea, a mechanism similar to Iran’s Hormuz fee, coming a week after they declared a maritime blockade of Saudi Arabia. Vespucci Maritime has named the Panama Canal as the next chokepoint risk, and Fertistream describes Hormuz, the Black Sea and Bab el-Mandeb as one converging theatre — while the Bank of England today formally listed shipping-route disruption, the Panama Canal included, among its inflation factors. A year ago, charging for passage through an international strait was unthinkable inside the shipping industry. It is now a replicable business model. That externality outlives any oil price.

And today’s inflation is not in crude. It is in freight and refined products. This is the most under-covered set of numbers of the day: the ICE gasoil crack spread broke above $70/bbl, a record, ING put front-month gasoil calendar spread backwardation above $80/bbl, and the Dated Brent EFP peaked near +$5 and stayed positive. Meanwhile September Brent slid from a European morning high of $93.31 to about $89.92 by 12:20 ET, down 0.9% from the 7/29 settlement, on an intraday range of roughly $4.4; and US front-month natural gas sits near $2.769, down 16.8% month-over-month and 10.5% year-over-year — the one energy commodity this war has not pushed up.

Line those three up: flat crude, gas down 17% on the month, a record diesel crack. This is not an energy shock; it is a logistics and refining shock. It also explains the strangest leg of the June PCE report — goods prices up 0.7% MoM against services at 0.3%, with goods running hotter than services. Diesel is an input to freight, agriculture and industry, and its transmission channel is shipping rates, not wages. The policy implication is larger than “core PCE 3.3%”: rate hikes do not compress crack spreads. One footnote — the Cleveland Fed’s 7/30 nowcast update puts Q3 annualized CPI at 1.15%, core CPI at 1.90% and core PCE at 3.01%. Against Brent up more than 20% in a single month, that is a model that has not caught up with reality; the 8/26 release of July data is where energy pass-through actually lands.

On tariffs, a new fiscal fact arrived today: US net customs revenue turned negative in June. IEEPA tariff refunds exceeded the month’s collections, against monthly IEEPA revenue that had run $27B-$31B from June 2025 through February 2026. Roughly 330,000 importers paid about $166B in IEEPA duties across more than 53 million entries, and that money is now flowing back. Treasury Secretary Bessent says total 2026 tariff revenue is “virtually unchanged” once IEEPA, Section 122 and Section 301 authorities are combined — but New York Fed research estimates that about 90% of the roughly $175B collected under IEEPA was borne by US consumers and businesses. Which is to say: a system that was mostly not taxing foreigners is now returning part of that to domestic payers. At the same time, the Section 301 tariffs of 10%-12.5% on some 60-80 economies were sued over almost immediately after taking effect on 7/23, and trade lawyers think they may fail judicial review as well. The upshot is rising uncertainty on the revenue side and rising certainty on the inflation-cost side (100% on pharmaceuticals 7/31, 50% on Canada 8/19). For a long-end curve already sitting at 52-week highs, those two things together are the worst possible pairing.

The Bond Market

Today’s curve delivered a textbook bear steepening — and it did so on cool inflation data. On the CBOE yield-index basis, the 5-year rose from Wednesday’s close of 4.352% to 4.370% intraday (+1.8bp), the 10-year from 4.622% to 4.659% (+3.7bp), and the 30-year from 5.143% to 5.206% (+6.3bp). The move increases monotonically with maturity. 5s30s = 5.206% − 4.370% = 83.6bp, 4.5bp wider than Wednesday’s 79.1bp, while 10s30s widened from 52.1bp to 54.7bp. A front end that barely moves while the ultra-long end rises on its own is term premium moving, not policy expectations. That is the cleanest single piece of evidence behind today’s core judgment.

A threshold broke as well. After 17 consecutive sessions closing above 5.00%, the 30-year traded at 5.206% today, roughly 2.6bp above the 5.18% close set on May 19 — the 52-week closing high. Note that this is an intraday reading and not yet a close. But it means the long end went to test cycle highs on the day inflation data cooled, rather than on a day it reaccelerated.

The supply calendar is closing in. Bloomberg reports that JPMorgan expects the Treasury to avoid changing auction-size guidance in next week’s quarterly refunding statement, given the approaching midterms, while the same team estimates a $3.7 trillion funding gap over the next four fiscal years and argues for deleting the words “at least” from “expected to remain steady for at least the next several quarters.” Removing an adverb is not a technical suggestion — it is a demand that the Treasury stop implying that “steady” has a floor. A long end making room for that statement is rational, and this is the strongest of the counterarguments I listed above.

Japan ran the other way. JGB 2s printed 1.488% (−1.6bp), 10s 2.757% (−2.6bp) and 30s 3.941% (−3.9bp) — the whole curve lower, with the 30-year still pinned below the 4% threshold. That happened on a day the yen appreciated 3.5%, with a Bank of Japan decision tomorrow. If the market believed a hawkish move were coming, JGBs would not trade like this. So today’s yen strength was imposed from outside, not generated from within. Take both ends together and the global ultra-long end tells one story: the US long end printed a cycle high on a cooling-inflation day, the Japanese long end fell on a currency-surge day, both governments’ long bonds need buyers, neither central bank wants to be that buyer at current inflation levels — and one of the two countries’ fiscal authorities chose today to act in the currency market instead of the bond market.

Sector Spotlight

AI platforms: two reports on the same night split “AI capex” into two categories. Microsoft traded up 16.6% (from Wednesday’s $390.54 close to $455.29 intraday) while Meta fell 9.4% ($585.61 to $530.73, RSI 13.3, 32.3% below its 52-week high). Both are spending unprecedented amounts: Microsoft’s Q4 capex including leases was $41B, up 69% year-over-year, with guidance above $50B for Q1 FY27 and roughly $175B for FY2027; Meta’s Q2 capex was $31.08B, with the FY26 guidance floor raised from $125B to $130B. The difference is whether there is revenue on the other side of the spend. Microsoft: Azure grew 43% in constant currency against 39-40% guidance, its fastest in four years, Azure passed $100B in annual revenue for the first time, and remaining performance obligations rose from $627B to $678B. Meta: revenue beat at +28%, but net income fell 14%, diluted EPS of $6.18 missed expectations of roughly $7.13-7.22, operating margin compressed from 43% to 31%, total costs and expenses rose 55% — and $31.08B of capex converted into just $784M of FCF, about 2.5% of the capex line. Last week the market punished the act of raising capex itself (SK hynix was hit despite a 76.3% operating margin; Vertiv fell despite raising full-year guidance). Today it changed the standard: it punished only the portion that has not been monetized.

Devil’s Advocate: if you are going to execute Meta over its operating margin, ask what depreciation schedule Microsoft’s margin rests on. Of Microsoft’s $4.74 adjusted EPS, $0.27 came from discrete items outside the April 29 guidance. Net gains on the OpenAI investment contributed $480M ($0.07) in Q4 and $4,963M for the full fiscal year — $0.67 of EPS derived from marking a private stake. More importantly, there is that accounting change: data center and office building depreciable lives extended from 15 to 25 years starting FY27, with more future leases shifted to operating treatment and therefore out of capex. The company says calendar-2026 capex plans are “unchanged after adjusting for the accounting change” — a sentence that only needs to exist if the accounting change moved the number. In fairness, a 25-year life is defensible for building shells; servers and accelerators sit on a separate, much shorter schedule. But it landed on the fastest-growing and largest line item, at precisely the moment the market is asking whether AI capex can cover its cost of capital. Azure at +43% and RPO at $678B are real, and they are what deserved to be paid for today. The EPS line contains a private-mark contribution and a useful-life extension.

Semiconductors: the supply side got bought back, and the largest forced seller happened to finish selling. The semiconductor ETF rose 6.8% intraday, with Micron +16.5%, Intel +13.4%, AMD +13.2%, TSMC +7.8%, Arm +6.4%, Dell +10.1%, Broadcom +4.0% and Nvidia +2.0%. Two hard facts underpin the fundamental side: Samsung Electronics posted a record Q2 operating profit of KRW 89.5 trillion against expectations of 88.13 trillion, with record DRAM and NAND revenue, HBM4 in volume production and first HBM4E samples shipped, guidance for memory supply to stay tight through 2028, and HBM4 shipments set to more than triple sequentially in Q3. Alongside that, UBS initiated SK hynix’s US depositary receipts at Buy with a $204 target, SanDisk traded up 18%, and Apacer was reported to be raising Q3 DRAM contract prices by roughly 30%.

Structurally, something more mechanical also happened. Leopold Aschenbrenner’s hedge fund Situational Awareness, having lost money simultaneously on long AI exposure and on short software exposure, was forced to meet prime-broker margin calls and liquidated its entire public equity book in a single block trade, with Citadel taking the other side. The fund has been reported at $20B-$24B in size, having peaked near $45B, and its 7/24 investor letter cited a net return of +439% for the first half of 2026. Its 13F disclosures centered on the semiconductor ETF, Nvidia, Oracle, Broadcom and AMD, plus CoreWeave, Core Scientific, Bloom Energy, IREN and Applied Digital. Devil’s Advocate: a roughly $13.7B disclosed book cannot explain the more than $1 trillion of equity value erased since 7/24 — and Citadel can now sell it too. So the significance of that trade is not that it validates fundamentals; it is that it removes a known forced seller and clears an overhang. The fundamentals were validated separately, by Samsung and by Microsoft, and those are revenue-side facts rather than price-side facts. Scale also matters: even after today’s rally, the semiconductor ETF is down 7.0% over five days, 17.7% over a month and 19.3% from its 52-week high; Micron is still 12.8% lower over five days and 25.2% over a month; and Arm, having delivered a record quarter — $1.29B of revenue up 22%, data center royalties more than doubling year-over-year for a second consecutive quarter, FCF up 343%, guidance above consensus — still carries an RSI of 28.1. Today was a correction, not a recovery.

Power: the compute leg came back, and the power leg came with it. NRG rose 7.4%, Constellation 3.7% and Vistra 4.2%. But their one-month readings are −8.1%, +8.2% and −6.3% respectively — this leg has spent five weeks trading as second-order beta to AI capex, rising and falling with compute, while verification from its own financials has yet to arrive: Constellation’s call is 8/6 and Vistra reports 8/7. Until then, treating power as a narrative independent of semiconductors has no data behind it.

Energy: for the first time, the commodity and the companies pointed opposite ways. ExxonMobil traded down 0.3% with an RSI of 80.3, Chevron down 0.5% at RSI 72.4, and Occidental down 1.2% — while September Brent sat 0.9% below the 7/29 settlement, front-month WTI near $84.2, and crude down 8.6% over five days. Energy equities have pushed their technical readings into extreme overbought territory on the back of a 13.9% one-month advance, while the commodity itself gave back 8.6% in a week. Last week the pattern was “commodity leads, equities lag”; this week it is “equities at RSI 80, commodity retreating.” There is one genuine justification for that divergence — a record diesel crack spread is real profit for refining and integrated operations, and it does not appear in the crude flat price. But RSI 80 also means this leg has priced the good news fairly fully. Fertilizer moved with the commodity today: CF Industries −2.0% and Nutrien −1.1%, with the 8/5 quarterly results the only hard data check this leg gets.

Digital assets: the miners are not trading Bitcoin today. Bitcoin sat near $64,790 (+1.4%) and Coinbase was roughly flat, while MARA rose about 16%, RIOT about 19%, CLSK about 19% and IREN about 25%. That distribution says these names are being priced as AI data center assets rather than crypto assets — they tracked the semiconductor complex’s 6.8%, not Bitcoin’s 1.4%. Crypto itself was the quietest place all day: the Fear & Greed index read 28 (fear); spot Bitcoin ETFs took in a net $32.1M on 7/29 on the Farside primary-source basis, ending four consecutive sessions of outflows totaling more than $500M; Ether ETFs saw a net $32.9M out the same day, though month-to-date Ether ETF inflows of $342.9M still exceed Bitcoin ETFs’ $204.7M. Perpetual and futures liquidations over 24 hours totaled $286M across 87,294 accounts, split $186M long against $100M short — the price barely moved and both sides got washed out. And the 7/31 monthly options expiry carries roughly $9.61B of notional open interest with Deribit max pain at $64,000, essentially where spot is sitting.

China: the only major equity market with a positive one-month trend. The China large-cap ETF is up 15.4% over a month and 5.8% over five days, with an RSI of 79.3, already overbought. Over the same month the Nikkei is down 10.9% and the Nasdaq 100 down 7.3%. This implies nothing about Chinese fundamentals, but it does say global equity allocation has shifted noticeably at the margin: in the same month the US long end sat at cycle highs and Japan’s currency needed official support, money went to the cheapest large market available. An RSI of 79 says that shift has not been slow.

What to Watch and How to Read It

7/30 after the close — Strategy (MSTR) and Coinbase Q2 results: As of 12:44 PM ET, MSTR’s most recent 8-K on SEC EDGAR was still dated 7/27 and the Q2 filing had not been submitted. Three mutually inconsistent figures for its Bitcoin reserve are circulating — 843,775, 846,000 and 847,363 — none confirmed by an 8-K, and none should be treated as fact before the filing lands. What is already pre-disclosed is the hard part: a Q2 digital asset loss of $8.32B (of which $8.31B unrealized), a quarter-end Bitcoin carrying value of $49.67B below what was paid, and a full valuation allowance taken against the related deferred tax assets. The read: the real information is not the coin count, it is whether that full valuation allowance comes with a change in financing cadence. Coinbase posted a $394M net loss in Q1, its second consecutive loss-making quarter, followed by roughly 700 job cuts; the thing to watch is whether transaction revenue’s share can keep falling in an environment where spot activity has shrunk to 2023 bear-market levels.

7/31 — Bank of Japan decision (policy rate 1.00%): This is the judge of today’s yen call. If the BoJ holds with no hawkish signal, then a suspected intervention stands alone — the yen likely hands back today’s gain, and the market starts counting the fiscal authority’s ammunition (roughly $70 billion deployed in late April and early May, about $1.09 trillion of reserves remaining). If the BoJ hikes or turns explicitly hawkish, the intervention acquires policy backing and the carry-unwind clock restarts; at that point the test remains the same three things arriving together — yen up, risk assets down, volatility up. Today only the first of the three was satisfied.

7/31 — Cameco Q2 before the open, and 100% Section 232 pharmaceutical tariffs take effect: For the former, watch realized selling prices and recurring FCF rather than production volume. The Department of Energy’s selection of five states for nuclear fuel campuses this week, and the framing of the US-Saudi nuclear agreement as a Westinghouse-related deal rather than a uranium supply deal, both suggest the policy tailwind currently lands on construction rather than on fuel. For the latter, the tariffs cover patented drugs and active pharmaceutical ingredients, applying first to the 17 companies listed in Annex III of the proclamation, with other covered importers following from 9/29. This is the most certain endogenous tariff-inflation channel in the second half of the year.

7/31 — Bitcoin monthly options expiry: Roughly $9.61B of notional open interest, 116,260 in call open interest, Deribit max pain at $64,000, with spot effectively glued to it. If the price leaves the $64,000 area quickly after expiry, this week’s flatness was options gravity rather than genuine equilibrium. If it stays, the fragile structure of weak spot demand layered with rebuilding leverage has further to run. The September expiry is the largest on the board at $7.53B notional with max pain near $74,000.

8/2 — OPEC+ core-seven video conference: Confirm whether September’s quota increase is the last of 2026 and whether new quotas slip to January 2027. Confirming a pause in the same week US commercial crude hit its lowest level since 2018 would remove the only supply buffer unrelated to Hormuz. Continuing to add barrels instead would say producers would rather absorb price pressure than concede share.

8/3 after the close — Palantir Q2: Company guidance of $1.797B-$1.801B sits below the roughly $1.81B consensus, and a guidance ceiling below consensus is itself a signal. Options are pricing roughly ±15% on the print. The thing to watch is not whether revenue clears the bar, but whether the software layer can keep trading growth rate for valuation multiple with rates at cycle highs.

8/4 — June goods and services trade (8:30 AM ET); 8/4 ~4:30 PM ET API and 8/5 10:30 AM ET EIA weekly reports: The trade release is the first test of the claim that +11.5% imports were tariff front-running; if June already shows a pullback, Q3’s mechanical improvement in net exports has a starting point. The inventory reports need one extra line read this week: with the diesel crack at a record, distillate stocks and refinery utilization carry more information than headline crude. Utilization already reached 97.2% with essentially no headroom, which means next week’s crude direction is set mainly by imports and reserve releases.

8/5 after the close — nitrogen fertilizer and precious metals royalty Q2 results: This is the first reported answer to whether energy equities at RSI 80 have the earnings power to back it. On the fertilizer side, the question is whether the spread between the cheapest feedstock available (gas down 16.8% month-over-month) and the scarcest product has actually shown up in the financials. If it has, this week’s commodity pullback is not a reason to re-rate the companies.

8/8 — the CLARITY Act before the Senate recess: Majority Leader Thune has already demoted it behind the Russia sanctions bill and personnel nominations. Market-implied odds are scattered and are listed here by source: the relevant Polymarket contract fell to an all-time low of 27% on 7/29, Galaxy Digital cut its own estimate to 30%, and other outlets have headlined 35%. The open disputes are a bipartisan counterproposal on ethics provisions and bank-lobby demands to restrict stablecoin yield and reward products. If nothing moves before 8/8, the legislative path effectively slides past the midterms — and SEC Chair Atkins has said that if Congress does not act, the SEC is “ready, willing and able” to write crypto rules under existing authority. Regulatory certainty would then arrive in a worse form.

8/18 — the original Day-60 marker, whose meaning needs redefining again. Today’s LNG carrier exited through Iran’s designated corridor with Iranian permission, which means part of the core Day-60 question has already been answered in physical data: the issue is not “will there be a fee,” it is that “who grants the right of passage” is already operating in fact. Gharibabadi’s terms — full Iranian control of the inbound lane, Iranian responsibility for demining, “never return to its pre-war state” — mean the negotiated solution space is far narrower than a dispute over rates. The reading rule is unchanged: only when a military pause, corridor rules, insurance underwriting and insured mainstream vessel flow all improve together does Hormuz downgrade from a structural constraint to an event premium. Right now only the fourth of those four showed a single-day improvement, and it improved because Iran authorized it.

8/19 — Section 338’s 50% tariff on 554 Canadian tariff lines covering roughly $20 billion of goods: Canada’s trade minister travels to Washington this week for the first face-to-face talks since the proclamation, while Trump has said publicly that he does not care whether a deal happens. Watch whether the exemptions for energy, potash, critical minerals and fish products hold. If they narrow, the Fed’s “supply shocks” phrase gains another source that is entirely endogenous to policy.

8/26 — Q2 GDP second estimate with corporate profits, and July PCE: This is the real adjudication date for this brief’s core judgment. If July core PCE stays mild after Brent rose more than 20% in a month, then the Cleveland Fed’s 3.01% Q3 nowcast is right and the long-end move belongs more purely to fiscal supply. If energy and freight begin passing through in July, then “monthly core cooled” was itself point-in-time noise and the quarterly deflator is the main line. The corporate profits data in the second estimate matters just as much — it is the first direct evidence on whether the capex behind that +15.2% equipment number is earning anything.

Disclaimer

This article is public market commentary and personal research notes. It does not constitute investment advice.