Alphabet raised its 2026 guide $15B and the Mag7 wrote a $797B cheque; Intel raised $2B and got paid 12%. Same word, opposite reception, thirty minutes apart.
⚓ Weathervane. The Fed has turned hawkish for the cycle, and the customer-side leg of the AI capex bill has confirmed. Alphabet, Tesla and IBM signed the same invoice inside one hour Wednesday; the vendor rally was Tuesday, the confession Wednesday. The equity thesis now has to price both the tightening regime and the AI margin bill; ad revenue can fund the second, but not everyone has ads. Riding unchanged today, under material stress on the ad-revenue clause after Thursday. Meta Wednesday is the next definitive test.
Ahoy there, Trader! ⚓️
It’s Phil…
It is rare that the same market delivers opposite verdicts on the same word inside thirty minutes. Thursday’s post-close brought exactly that.
At 16:00 ET Alphabet raised 2026 capex from $180-190B to $195-205B, warned 2027 “increases significantly,” and posted the first quarterly free-cash-flow burn in its public history at -$5.9B. The tape took $180 billion off Alphabet, and $797 billion off the Mag7 across the session, biggest one-day drop since the April 2025 tariff tantrum.
At 16:15 ET, Intel raised its own 2026 capex from $18B to $20B, flagged 2027 “meaningfully” higher, and printed revenue up 25% year on year. The stock went from around $99 to $112 in the after-market.
Two invoices, thirty minutes apart, opposite reception. Which raises the day’s real question, and half of yesterday’s Weathervane hypothesis was tested by one of them.
When does capex reward and when does capex punish?
The textbook is clean. Capex is a call option on future free cash flow. If the market believes return on invested capital will exceed cost of capital, it buys the option forward and rewards the spend. If it disbelieves, the spend is a straight drag. Vendors like Intel sell into a bill somebody else is paying, so their capex reads as supply widening. Customers like Alphabet pay the bill, so their capex IS the question.
On the tape Thursday it was messier. Alphabet grew Cloud 82%, Search 17%, Ads 10%. If growth were the funding source, as the Weathervane hypothesised, Alphabet should have been the safest name in the group. It was the most punished by a wide margin.
So the tape may have moved past the “does the top line fund it” question and gone somewhere sharper: is the capex growth rate outrunning even accelerating top-line growth? Alphabet’s 2026 guide is up around 8-9% on prior; 2027 higher. Ad revenue grew 10%. That is a matched race, and the tape does not want to be told capex growth is nipping at revenue growth.
Intel’s raise is different because it is small ($2B on $16B revenue) and funded by a demand curve just demonstrated at 25% year on year. So the market’s implicit answer: capex rewards when it lags top-line growth, and punishes when it approaches or outruns even accelerating top-line growth.
That is a cleaner working rule than “ad revenue funds it.” Meta Wednesday tests it directly. If Meta’s ad revenue grows faster than its 2026 capex raise, the rule holds. If capex outruns 10% ad growth, we may need a bigger word than “punish.”
Phil’s Musing
The bit that keeps me awake here is the Kospi. Samsung and SK Hynix both traded +3% overnight while Nasdaq futures sat flat post -2.15%. The same capex line is a growth story in Korea and a margin bill in the United States, inside the same 12 hours. That is a rotation, not just a sell. If Meta gets the Alphabet treatment Wednesday, we should probably reckon with a global one.
Happy trading,
Phil
Less Brain, More Gain
…and may your trades be smoother than a cashmere codpiece
P.S. – Phils Footnote One thing I keep learning in public here: my instinct on Wednesday was that ad-revenue growth would be the pivot the market cared about. Thursday says the pivot may already have moved to the capex-versus-revenue growth-rate ratio, which is a sharper and much less forgiving question. If you have a better read on that ratio before Meta prints Wednesday, I would like to hear it.
🗒 Desk Notes | Friday, July 24, 2026
Raw briefing from the pipeline’s mechanism read. Observations, not trades.
Session read (§8.4)
SESSION BRIDGE: prior session (Thu Jul 23) full reaction Alphabet raised 2026 capex $195-205B (+$15B) with FCF -$5.9B first-ever cash burn plus Tesla FCF -$1.1B triggered Mag7 -4.8% (-$797B market cap, biggest one-day drop since April 2025 tariff tantrum), SPX -1.21% at 7,408.30, Nasdaq Composite -2.15% at 25,137.69, NDX -1.9%; oil surged 7% to Brent $100.69 (WTI $92.19) on Houthi tanker attacks + Saudi maritime blockade + Kazakhstan CPC drone-hit suspension + Trump “massive attack” threat; 10yr 4.71% highest since Jan 2025 +6bp, Sep hike odds jumped 61% to 78% inside blackout on oil feed, July hike odds 33%, VIX cash close 18.69 (+12.32%); Intel post-close BLOWOUT rev $16.13B (+25% YoY) EPS $0.42 doubled $0.22, Q3 guide $16.3B mid tops $15.06B, capex $18B to $20B, stock +12% AH to $112; live premarket Fri 01:27 ET ES +0.04% / NQ -0.09% / VIX 18.69 cash carry, WTI $91.78 (-0.44%), gold $4,029.4 (-0.51%), BTC $65,289 (+0.37%), DXY 101.44 (+0.29%); threshold: HARD (Nasdaq -2.15% breaches 2% carry-over trigger, Mag7 -$797B is the biggest one-day loss since Apr 2025 tariff tantrum, multiple named catalysts, ES flat premarket is itself abnormal after that reaction).
Mechanism read (full)
What moved.
- Front end: 2yr firmed materially again (level pending CME close, direction confirmed by the Sep hike odds move from ~55% Tue → 78% Thu). Wed close was 4.30%; Thu likely 4.35-4.38% area. Front end is now pricing what Warsh cannot say.
- Long end: 10yr 4.71%, +6bp, highest since Jan 2025. Fourth consecutive up session. Oil is doing the tightening while the FOMC is in the soundproof booth.
- Dollar: DXY 101.44 (+0.29%), bid on the hawkish reprice.
- VIX: cash close 18.69 (+12.32%), fresh above 18 designation returns after the one-day sub-18 (Wed 17.05) print looks like a positioning trade in hindsight – Part B “17.59 was positioning” MARKED-HIT on direction AND level (17.05 was the low, 18.69 clears the 17.50 arbiter).
What it implies.
- The macro-monetary transmission is running through oil this week. Kazakhstan CPC out, Red Sea widened by the Houthi blockade on Saudi ports, twelfth night of US strikes on Iran. That is not an earnings-week backdrop – it is an oil-shock backdrop that the earnings tape is metabolising as a further hawkish reprice.
- The equity story is being priced through capex maths: Alphabet -7.13% on a $15B raise plus the 2027 “increases significantly” tell; Tesla -12.6% on -$1.1B FCF; then Intel +12% AH on a $2B capex bump. The market is not selling AI capex. It is selling customer-side capex.
- The AI capex reckoning has now propagated: Amazon caught -4.6% (~$120B cap gone) as a peer read, Meta -3.36% and MSFT -2.24% pricing their own coming Q2 confessions Jul 29-30. Meta prints Wednesday.
The one artery. The customer-side leg of the AI capex bill got paid at the tape, in cash, for $797 billion, in six hours. That is the story. Everything else (oil, front end, hike odds) accelerated in step – oil provided the inflation feed the front end used to firm hike odds, which provided the discount-rate leg on the same Mag7 sell.
Editorial carry-over (§10.6 – fires today)
The prior session earns the backward glance. Two beats to carry:
- AVE: the $53-per-dollar exchange rate on Alphabet’s capex guide, and the half-hour gap between Google getting sold and Intel getting bought for the same word.
- Snippet: “A footnote on a capex slide has never been more expensive.”
- Macro Edge: the actual dissection – the market has been telling us since IBM’s July 14 letter that the customer-side leg was coming; Wednesday delivered the three-signature print; Thursday delivered the price. The ad-revenue-funds-it rule the Weathervane hypothesised is now under material stress because Alphabet HAS ads, ads grew 10%, and Alphabet got sold anyway. Meta Jul 29 is the definitive test.
Forward catalyst slate
- Today (Fri Jul 24): US Flash PMIs at 09:45 ET. UMich final at 10:00 ET. Nasdaq stress test as futures open at 09:30 ET after Intel AH gain vs Mag7 sympathy.
- Sunday Jul 27: Moonshot AI Kimi K3 full open weights release. This is the model that took SOX into bear-market territory Jul 17 in intra-day trading; the full weights print could re-test the vendor-side floor.
- Mon-Tue Jul 27-28: ECB/BoJ policy statements around FOMC. Front end sensitivity elevated.
- Wed Jul 29: FOMC + Warsh presser 14:00 ET. Meta Q2 post-close is the first ad-revenue-funds-it test with a peer that has ads.
- Thu Jul 30: Apple + Amazon post-close. Two customer-side names with different revenue engines; if both get sold, the customer-side leg fully generalises.
- Fri Jul 25 (today): US Q2 GDP first estimate. Note weakness would compound the front-end paradox (oil-driven inflation into slowing growth).
Divergence flags
- The vendor-vs-customer split priced live inside 30 minutes. Intel +12% AH on capex raise; Alphabet -7.13% on capex raise. Same word, opposite verdict. The market’s implicit rule Thursday was “AI capex is a virtue when the vendor says it, a vice when the customer says it, regardless of the top-line growth funding it.”
- Bitcoin refused to sell into the equity risk-off. BTC touched $64,700 intraday during the Mag7 -$797B session, closed $65,556 area, back to $65,289 by Friday premarket. Seven-session ETF inflow streak neared $1B while the equity Mag7 wrote a $797B cheque. Correlation break, worth tracking whether it holds a second session.
- The front end firmed inside a blackout by 17 hike-odds points. Sep odds 61% Wed close → 78% Thu close, with no Fed voice available. The oil surge did the tightening. When the referee is out and the tape moves this much, the presser Jul 29 has to either confirm or repudiate what the tape has already priced.
- Kospi +2.8% overnight while Nasdaq futures sat flat post -2.15%. Memory-supplier read of the same Alphabet capex raise US customers got sold on. Same event, opposite verdict across hemispheres. This is the vendor-customer split rendered at the country-index level.
- Ether ETFs beat Bitcoin ETFs Wed for the first time in weeks. $72.64M vs $68.99M. ETH staked supply at record 33.9% of circulating. Rotation inside crypto ETFs worth noting; not a full pattern break yet.
Part C – regime tracking read
⚓ Weathervane (persistent banner). “The Fed has turned hawkish for the cycle, and the customer-side leg of the AI capex bill has confirmed. Alphabet, Tesla and IBM signed the same invoice inside one hour; the vendor rally was Tuesday, the confession Wednesday. The equity thesis now has to price both the tightening regime and the AI margin bill; ad revenue can fund the second, but not everyone has ads.”
Today’s Weathervane read. Rides UNCHANGED but under material stress on the ad-revenue-funds-it clause. Alphabet HAS ads, ads grew 10%, and it got sold $180B anyway. That is a live rebuttal of the working rule, but from Alphabet’s own reaction rather than a peer with ads. Meta Jul 29 remains the definitive rewrite trigger.
Regime Flag (internal tripwire). Chipflation customer-side thread: TRIPPED per Wed Jul 22 (three-of-three signatures). Vol regime compression: BROKEN designation revived – Thu cash close 18.69 clears the 17.50 threshold definitively. Hormuz price-side: BROKEN hardening cycle 16 (Brent above $100 first time since May, Kazakhstan CPC out, Red Sea widened). Fed hawkish-locked: CONFIRMED cycle 13 (78% Sep hike odds inside blackout). Bitcoin ETFs: eight-day streak building; first equity-risk-off stress test passed intraday.
Sensitivity read (§17.3). HIGH-TIGHT and tightening further. Six live candidate triggers now open (per yesterday’s ledger); Thu’s cash close resolved three of them in the confirming direction (VIX above 18, Mag7 pricing the capex bill, 10yr above 4.70%). One remains open on the SPX-7,400 line (Thu close 7,408.30, within touching distance). The Meta / Apple / Amazon wave Jul 29-30 stays the definitive multi-trigger event.
Public tell status. NOT triggered. “Hoist the mainsail” language remains holstered. Recommendation to Phil: hold at yellow through FOMC; if Meta gets sold on the same ad-revenue+capex combination that just failed to save Alphabet, move to green subject to Phil’s greenlight.