Red Sun at... Night?
morning musings 7.17.26-- views from the smoke-covered island
“I had a dream, which was not all a dream. The bright sun was extinguish’d...” — Lord Byron, ‘Darkness’
The sun went red last night Wildfire smoke, they say, and I believe them, but it made for a fitting sky. Because there are at least three separate fires burning under it, and the market has decided to price exactly one of them.
FIRE ONE.
The President went on primetime and told the country that China stole 220 million voter files and rigged 2020, then declassified the ‘intelligence’ to prove it (BBG).
Set aside whether a single word of it is true. Ask what it is FOR. You do not manufacture a stolen-election narrative eight weeks into openly talking about the midterms because you are worried about archival integrity. You build it as the predicate for federal ‘election security’ intervention, and the map of where that intervention lands is not a mystery. It is seven states, the same seven we always end up talking about.
Did I miss the part where the Constitution made room for this, or are we filing that document under ‘suggestions’ now?
FIRE TWO.
The Middle East stopped pretending. Sixth straight night of US strikes, and Iran answered by hitting Kuwaiti water and power plants, civilian infrastructure, the first time since April (BBG). The IMO has declared Hormuz too dangerous to transit.
And the new wrinkle, the one that belongs at the top of your risk sheet: Tehran has reportedly asked the Houthis to close the Red Sea oil route if the US touches its power grid (PDB). That is not one chokepoint. That is two. Brent is up twelve percent on the week (FMP) and the tape is treating it like a rounding error.
FIRE THREE, and this one is almost funny.
SpaceX scrubbed Starship Mission 13 on an engine failure, the stock has cracked below its IPO price, and the most-hyped listing of the year has dragged the 2026 IPO cohort’s average return down to six percent (BBG).
The company that was going to colonize Mars cannot currently clear the pad, and its equity is heading the direction its rocket did not.
Here is what the market actually priced. Not the election. Not the second chokepoint. It priced the fourth fire, the one nobody set on purpose. The AI trade finally caught.
TL;DR
Consensus is calling this a buyable Goldilocks dip. Soft CPI, Fed on hold, quality AI names on sale.
I think consensus is reading the wrong instrument. This is not a dip in a monolith and it is not a soft landing. It is a de-gross in the most crowded trade on earth, chips and AI, colliding with an inflationary energy supply shock, under a Fed that has openly stopped talking about cuts.
The tell is not the Nasdaq. The tell is gold, which broke below 4,000 on a risk-off morning (FMP) because the market is now choosing higher-for-longer over safety. The disinflation leg the whole dip-buy rests on already cracked at 8:30 this morning, when import prices printed plus 0.3 against a minus 0.7 estimate.
The book: net long the rotation into financials and defensives, short the crowded chip complex, short duration on higher-for-longer, hedged with short HY and long energy. What kills it is one print. Gold reclaiming 4,000.
MY READ
The market wants to tell you a comforting story this morning, and it goes like this: June CPI was soft, the Fed is locked on hold, and a Chinese lab shipping a cheap model plus a Google product delay is a positioning wobble in a secular bull. Buy the quality names on the flush. That is the consensus, it is coherent, and it is what the sell-side will publish by noon.
I do not believe it, and the reason is mechanical, not emotional. The comforting story has one load-bearing beam, confirmed disinflation, and that beam broke this morning while everyone was staring at semiconductors. June import prices came in plus 0.3 percent against a consensus of minus 0.7, up 7.1 percent year on year (BBG). That is not noise. That is the goods-inflation channel reopening at exactly the moment the Fed has gone hawkish in chorus. Warsh told Congress the committee has ‘no tolerance’ for elevated inflation (BBG). Cook said she is ‘prepared to act.’ Logan wants ‘modestly higher rates.’ Williams named AI demand as his primary inflation worry. That is not a Fed about to cut into a dip. That is a Fed looking for a reason to go the other way.
And the cross-asset tape agrees with the Fed, not with the dip-buyers. Run the haven test. On a morning where the Nasdaq is down nearly two percent, gold is DOWN, through 4,000 to the downside, with ETFs bleeding ounces (FMP). The yen is offered and sitting near its year low against a firm dollar (FMP). Only bonds caught a bid, and even that bid is capped by rate-hike repricing. When the risk-off asset that ALWAYS works, gold, cannot work, that is the market telling you the dominant force is not fear of a slowdown. It is fear of higher-for-longer. In a genuine growth scare, gold rips. It did the opposite. That single fact reprices everything.
So my variant, stated plainly: this is a rotation OUT of the crowded AI/chip complex and duration, and INTO financials, defensives, and the energy supply shock, all happening inside a hardening rates regime. Not a dip. Not a de-risk to cash. A regime that punishes the exact positioning that worked for six months. And the crowd is offsides because it is still trading the last regime, pricing a Fed cut that the Fed just spent the week telling you is not coming.
WHAT MATTERS TODAY
The AI de-gross has a name now, two of them. Moonshot’s Kimi K3 landed overnight claiming parity with the frontier US labs at a fraction of the cost, and Alphabet confirmed Gemini 3.5 Pro is months behind on coding (BBG). Together they are a second ‘DeepSeek moment,’ and they attack the one premise the chip multiples rest on: that compute is scarce and expensive forever. It is not the demand that broke. TSMC beat and RAISED capex (PDB). It is the willingness to pay 30-plus times earnings for the arms dealers when the arms are getting cheaper and a Chinese lab just proved it.
The macro data cut against the dip-buy, twice. Housing starts ripped to 1,427k, well above every estimate, up 19 percent on the month (BBG). Strong. Import prices ran hot. Also, for the Fed’s purposes, the wrong kind of strong. Michigan sentiment came in soft at 49.5 (BBG). The composite says: growth is not rolling over, but inflation is not done, which is precisely the mix that keeps the Fed hawkish and duration under pressure.
The energy tail is live and being paid, not hedged. Brent 86.13, up 1.9 on the day and up roughly twelve percent on the week, the biggest weekly gain since April (FMP). Gasoline running 32 percent above pre-war levels, a diesel crunch building, European gas up 14 percent on the week (PDB). This is a supply premium the market is actually paying, not a headline it is fading. The Houthi Red Sea threat is the discrete event a mid-80s Brent is not priced for.
MOVERS THAT MATTER
GOOGL, -4.44% to 354.46 (FMP). The Gemini 3.5 Pro delay dropped the one mega-cap that was supposed to be a rotation WINNER into the loser column. So what: the ‘own the hyperscaler platforms’ trade is name-specific, not a basket. AAPL and MSFT caught the bid, GOOGL got sold with the chips. It is exactly why we cut the long-GOOGL idea rather than fold it in. You cannot be long the buyers by owning the one buyer the tape is selling.
Regional banks and Travelers, all beating (BBG). TFC 1.23 vs 1.08, FITB 1.02 vs 0.95, RF 68c vs 63c, and TRV a blowout at 10.04 vs 5.38, with provisions coming in BELOW estimates. So what: the financials-long leg got live earnings confirmation while the tape fell, and the light provisions are the tell that the consumer-credit crack the bears keep forecasting is not in the bank numbers yet. The rotation into financials is not a chart, it is a fundamentals print.
Gold, 3,983.80, -8.3, sub-4,000 (FMP). So what: this is the whole regime read in one line. On a risk-off morning with a war escalating, the classic haven is being sold because higher-for-longer beats safe-haven demand. That is why the expressions are a duration short and a firm dollar, not a gold long, and it is the single print that, if it reverses and holds back above 4,000, tells you I am wrong and this is a growth scare after all.
FINANCIALS
Financials are the one place the tape is rewarding fundamentals, and the earnings this morning made the case in ink. Regionals beat across the board, TFC, FITB, RF, and every one of them printed provisions for credit losses BELOW consensus (BBG). That is the number that matters more than the EPS beat, because it says the credit deterioration the bears are underwriting has not shown up in the regional book. The driver of the day is the steepener plus deposit-cost relief: 2s10s around +41bp (Fin Daily), deposit betas falling (Fifth Third’s cost of deposits 1.71 against 2.04 a year ago), and a NIM inflection that the Q2 prints are now confirming rather than promising.
The dispersion is the edge, not the beta. Capital-markets names SOLD the news, GS off 4.9 and MS off 4.5 on their prints despite GS posting a record 20.3 billion quarter, because the positioning was crowded and the multiple was full (Fin Daily). Regionals ripped. That is the whole trade inside financials: regionals over brokers. Insurers are the other clean corner, KIE up 2.2 on the session and TRV blowing the doors off, the defensive-financials expression that works whether this is a rotation or a de-risk.
The one thing I will not let slide is the private-credit tail, because it is the thing that is not in the bank multiples. BDC non-accruals are creeping toward 6 percent, redemption pressure at the non-traded vehicles is running into the 5 percent quarterly gates, and ARI printed a 33 percent single-session drop on the 16th on a CRE-mortgage-REIT stress (PDB). Layer on hyperscaler bonds ranking among the worst performers in the global indexes year to date (BBG) and HY primary stalling to a single deal on Thursday (BBG), and you have the early tells of a credit tape that is tight on the screen and fraying at the edges. The so-what for positioning: own the insurers and the regionals, underweight the BDCs, and treat ARI as a flare, not an anomaly, until the Q2 prints prove otherwise. The level that arms the credit hedge in size is HY OAS through 300 with a rising 10Y. We are at roughly 266 (context). Not there. Watching it.
SECTORS
Levels are 16 July closes, the last cash prints; 17 July direction is live futures (FMP).
XLK 177.52, -2.24%. The crowded trade paying for its own concentration; tech is the epicenter and the read-through into next week’s mega-cap prints is the whole ballgame.
XLV 161.80, +2.22%. Healthcare broke out as the defensive-rotation leg; UnitedHealth’s guidance raise ($19.50-20.00 from >18.25) is the fuel and the single-name tell.
XLF 56.75, +0.34%. Another 52-week high on a down day; the rotation leader, now live-confirmed by regional and insurer beats, and the most extended long in the book.
XLE 57.02, +0.92%. The only sector with a real fundamental tailwind; oil is being paid, not faded, and energy is the natural hedge inside the rotation.
XLP 85.81, +2.80%. Staples ripping and the strongest SPDR on the session; classic risk-off defensive bid, and confirmation this is a rotation with a defensive core, not a melt-up.
XLRE 45.46, +2.02%. REITs bid on the bond bid; enjoy it, but this is the sleeve that hurts first if the CRE-credit tail bites.
XLU 45.47, +0.56%. Utilities firm on the duration bid but muted next to staples; the defensive bench, not the starter.
XLY 117.34, +0.29%. Discretionary flat; the consumer is neither confirming the soft-landing bull nor breaking, which is its own quiet warning with Michigan at 49.5.
XLI 180.15, +0.05%. Industrials flat as the defense bid and the housing-plus-trade drag net each other out; a mid-beta way to own nothing in particular today.
XLB 50.89, +0.77%. Materials up modestly while copper fell 2 percent; the divergence is the growth question nobody is answering.
XLC 112.65, -0.64%. Comm services down with the AI complex, dragged by GOOGL and META; the sector that most needs next week’s capex prints to go well.
FLOWS & POSITIONING
The setup that makes this dangerous: exposure was near the highs going into the flush. The NAAIM Exposure Index jumped to 95.64 this week from 82.95, sitting just under the 98.59 read on 6/24, itself the highest since December, against a 52-week high of 100.83 and an April trough of 35.16 (SmartMoneyPassport). Managers reloaded risk right into the teeth of the de-gross. That is the ignition source for a real move: crowded, extended, and long the exact complex that is unwinding.
The corroborating flow tells line up. US insiders sold 77.6 billion of stock in the first half, up 20 percent year on year, the second-fastest pace in over two decades, and the only comparable period is 2021 (BBG). The quality factor is the best-performing basket of the last two months, up 1.9 percent Thursday as the Nasdaq fell (BBG), which is the balance-sheet flight in miniature. The leveraged single-stock AI ETFs, the 2x Marvell, NBIS, IREN, OKLO cohort, got flushed 18 to 28 percent, and Korea has now banned new single-stock leveraged listings and tripled the cash-deposit requirement (PDB). That is regulators reacting to a speculative unwind, not a healthy pullback.
Two structural drains are landing at the same time. The supply wall: 140 billion of 2026 IPO issuance into a de-rating tape, with SpaceX now below its listing price and dragging the cohort return to six percent (BBG). And the private-credit gate risk from the FINANCIALS section, which is the tail that is not in any multiple. Perspective: this is what a de-gross looks like BEFORE it decides whether it is a rotation or a liquidation. Gold tells you which.
PAID BRIDGE
Behind the wall: the full book marked to this morning’s live tape, six cross-asset expressions with entries, stops, targets and the exact print that falsifies each. The duration short as the cleanest higher-for-longer expression and why I am holding it core rather than trimming it the way the mechanical ladder wants. The financials sleeve, two intra-sector pairs that express dispersion instead of beta, regionals over brokers and insurers over BDCs. The decision tree branched on the four prints that actually drive the book, gold 4,000, SOXX 555, Brent 95, and the 10Y at 4.40. The conviction ranking. And the watchlist, including the single trigger that re-arms a long GOOGL and the one that arms the credit hedge in size.
Keep reading with a 7-day free trial
Subscribe to Morning Musings- Thoughts from Dauber Island to keep reading this post and get 7 days of free access to the full post archives.



