The Morning 10

What

The most binary print of the season resolved up. Intel reported $16.13 billion in revenue against a $14.4 billion consensus — up 25%, the fastest growth in more than fifteen years — with adjusted earnings of $0.42, double the $0.21 estimate. Data Center and AI grew 59% to $6.3 billion; foundry revenue rose 31%. The headline GAAP loss of $11 billion is an accounting event, not an operating one — a $12.5 billion mark-to-market on the CHIPS-escrow shares. The stock, which went into the print up 163% on the year and had run 10% higher the day before, still gapped up 12.4% after hours to around $112.70.

If

The regular session holds the gap, the record moves to 4-0-7 and the season's loudest question — whether the build-out's money reaches Western manufacturing — got its referendum answer with a raised hand. A faded gap after a run-in this hot would say the print was owned before it existed.

Why

A 59% data-center number at Intel is not an Intel story — it is the demand line of the entire compute thesis, confirmed at the one vendor the market trusted least. But the valuation math explains why the diary's compute exposure runs elsewhere: at the after-hours price, Intel trades near 69 times next year's consensus of $1.64 on roughly 13% revenue growth — while Nvidia, at 208.76, trades near 16 times its next-year consensus of $12.77 on growth above 40%. Four times the multiple for a third of the growth: the turnaround is priced as if already delivered, while the incumbent is priced cheaper than the market on next year's number.

Then

Score the close against the after-hours settle tonight. The book holds no tactical Intel — that was the contract, and a 12% gap it does not own is the cost of discipline, paid without complaint. The compute position stays where the growth-per-multiple lives.

What

SAP's print was two-handed: revenue rose — cloud up 22% to €6.3 billion, current cloud backlog €22.9 billion, up 27%, earnings of €1.89 beating by €0.22 — but the company lowered its operating-profit guidance for the current fiscal year, citing a sequential deceleration in growth, with the Dremio and Prior Labs acquisitions adding over €100 million of dilution on top. Frankfurt paid the print 6% off the 52-week low before the New York flush took the ADR back down 1.6% with everything else. The stat of the quarter: AI and SAP Business Data Cloud featured in more than 90% of the fifty largest deals.

If

The market keeps paying the print through a lowered guide, it is telling you which line it now prices SAP on — the orchestration claim and the backlog, not this year's margin. A fade back toward the 52-week low would say the deceleration language weighed more than the AI pitch.

Why

What SAP is actually selling changed this quarter: past basic copilots, toward an 'Autonomous Enterprise' where coordinated multi-agent systems execute business workflows end to end. The stack is concrete — Joule Studio as the build surface, the SAP Autonomous Suite, and n8n's low-code automation embedded into Joule Studio so developers can orchestrate agent workflows across SAP and external systems. The moat claim is the Knowledge Graph: a proprietary semantic model over decades of enterprise data that lets agents act with compliance-ready business context — the thing a peripheral point solution cannot replicate. Management tied agentic orchestration directly to cloud growth: outcome-as-a-service, software shifting from passive assistance to autonomous execution. ServiceNow made the same claim Wednesday from the workflow side; the orchestration race is now a two-horse field, and the layer below it — peripheral, single-point-of-excellence software — is where the extinction risk concentrates. The market has spent three weeks repricing exactly that layer: IGV closed a fifth straight day down at 87.10, and ServiceNow's bought beat round-tripped to 91.94.

Then

The winning position needs two things — the data, and the seat at the center of the enterprise where you cannot be switched off. Score both orchestrators on follow-through once the macro clears; the claim got made twice this week, and paid twice, briefly.

What

Thursday was the worst Mag 7 day of the season, and it was not close: Tesla −14.5% to 319.69, Alphabet −7.1% to 317.69 — closing below its 200-day near 323 — Amazon −4.6%, Meta −3.4%, Microsoft −2.2%, Apple −1.3%, Nvidia −1.6%. The Nasdaq 100 lost 1.9%, the S&P 1.2%. Two of the seven now sit below major levels, and the group gave back its year in relative terms in a single session.

If

Alphabet reclaims the 200-day quickly, Thursday reads as a macro flush that used the capex print as an excuse. If it stays below while the raised guide stands, the market is repricing the spend itself — the cash-burn read, not the multiple-compression read.

Why

The spender-pays-vendor-collects frame just got a discount-rate multiplier attached. The Mag 7 are the largest duration assets on earth — their terminal values live furthest out on the curve, and an 82% hike probability moves the denominator under all of them at once. Note what did not happen: the semiconductor complex fell less than the index, and the season's biggest beat came from a vendor after the close. The tape is not questioning the build-out; it is questioning what a dollar of 2030 cash flow is worth.

Then

GOOGL against 323 is the level of the day on the spender side. The scoring line from Wednesday's print — sold 3.3% after hours — has now compounded to −10% in two sessions.

What

The repricing under everything: fed funds futures now put a September hike at 82%, up from 52% a week ago and roughly one-in-five before the oil shock began; December is 92%, and even next week's meeting carries a 38% hike probability, up from under 12%. The two-year yield hit 4.24%, a 16-month high; the 10-year real yield is closing in on 3%. IEF closed at 92.85 — through the 93 floor that held all month — TLT sits at new lows, and gold fell 2% to 371.52 on a day equities were sold: the haven failed because gold is a duration asset too when the real yield is the thing moving.

If

Next week's Fed meeting confirms the hold but validates the hike path, the tape has its answer: the cutting cycle is over, and every asset gets re-marked against a rising discount rate until the oil curve or the data says otherwise.

Why

This is why Thursday had no hiding place — equities, bonds and gold all fell together because they are all the same trade when the denominator moves. An oil-shock hike bet is the nastiest kind: it tightens into a supply shock, not demand strength. The one mitigant is in point 5 — the curve does not believe the shock lasts.

Then

The 93 line on IEF is now resistance to reclaim rather than a floor to hold. Real yields at 3% is the number that turns a flush into a regime — watch TIP, not headlines.

What

Brent holds above $100 at 100.33 this morning, off its 101.16 high — while the December US crude future still sits around $80, exactly where it sat before the spike. A twenty-dollar backwardation, and so far only spot and the short end are moving; the long-dated strip has not budged. The oil ETF tells the same story on one chart: USO at 139.49 trades below the 142 shelf and below the ~151 crisis highs printed in May — spot crude above 100, and the vehicle that holds it still below its own panic levels from four months ago.

If

The deferred contracts start rising toward spot — December bidding from 80 toward the front — that is the curve declaring the disruption structural, and every risk asset needs re-marking. Until then, the backwardation is doing the opposite: paying the market to bet on normalization.

Why

The watch-line established Wednesday holds: crisis is when the back of the curve believes the front, and it does not yet. A $100 handle on spot with an $80 December is the futures market pricing blockade theater, not lost supply. The morning's tape agrees — oil slipped and futures rose after Washington's overnight line shifted from military threat to making frozen Iranian funds pay for tanker damage.

Then

Two numbers carry the whole macro: December Brent against spot, and USO against 142. Everything in point 4 re-prices off what these two do.

What

Into the rate scare, Washington added a tariff round: new duties on imports from 60 countries targeting weak forced-labor laws, 50% tariffs on a range of Canadian goods effective in 30 days, and the Section 122 surcharges expiring today unless extended — with more targeted successor pathways reportedly prepared. Meanwhile the calendar itself turns: late July marks the start of the market's seasonally weakest stretch, now loaded with rate fears and an open geopolitical tail.

If

The Section 122 decision lands hawkish today, the tariff impulse compounds the oil impulse in the same inflation prints the Fed will read in September — the two shocks are additive exactly where the hike bet lives.

Why

Seasonality alone is noise; seasonality plus a hiking repricing plus two live policy shocks is a regime where rallies get sold and dips get bought slowly. This is the backdrop the book's discipline was designed for — the reason the buying phase closed at Thursday's open rather than staying open into August.

Then

No prediction, one posture: expect chop, respect the watch-lines, and let the curve — not the headlines — say when the macro has actually changed.

What

On the worst tape day of the month, the semiconductor complex outperformed again: SOXX fell 0.77% against the S&P's 1.23%, closing at 551.24 — nineteen points above the 530–532 band, the breakout intact through a discount-rate flush. The five-day tape says it louder: SOXX +8.4%, SMH +5.9% while the equal index went nowhere. Then Intel's print landed after the close, and Nasdaq futures point up 1.3% this morning.

If

The complex opens on the Intel gap and holds it through the first hour, the leadership signal survives its hardest test yet — outperformance on the down days, confirmation on the prints, and a bid that macro cannot shake loose.

Why

Relative strength during a flush is the single most reliable leadership tell this tape offers. The layer being sold is the one that owns the spend; the layer being held is the one that collects it. Thursday put a rate shock, an oil shock and a Mag 7 rout on the tape at once — and the money still would not leave the collectors.

Then

SOXX against the band remains the structural line. A close above 552.69 — Tuesday's breakout print — on Intel follow-through would put the whole sequence at new highs on the week of its hardest macro.

What

Step back from the flush and count the week: Alphabet beat on revenue and raised its capex guide, ServiceNow beat every metric, SAP beat and broke its unpaid streak, Nokia beat with AI-and-cloud sales up 105%, and Intel printed the fastest growth since 2011. The misses — Tesla's earnings, IBM's revenue line — were company stories, not cycle stories. Five beats, two idiosyncratic misses, and a capex path that was raised for 2026 and flagged higher again for 2027.

If

Next week's five prints — CDNS and Meta Monday, Amazon Tuesday, Apple Wednesday, Microsoft Friday having pulled to the 29th — hold this hit rate, the season closes with the fundamental case stronger than it opened, whatever the tape did to multiples in between.

Why

This is the asymmetry the flush obscures: the numerator is improving while the denominator is repricing. Discount-rate corrections against rising earnings are historically the buyable kind — they compress the multiple, not the cash flow. The 2027 capex escalation flagged by the largest spender means the demand line under the compute winners extends, not peaks, into any weakness this macro produces.

Then

The diary's read, stated plainly: prolonged discount-rate weakness in the AI compute winners is weather to be used incrementally, not fled — on confirmation, per the rules, never all at once. The season is validating the thesis while the macro discounts it.

What

Read what fell hardest Thursday and the flush identifies itself: the casualties were duration — Tesla, Alphabet, software, gold, the long bond — while the cyclical and hedged corners held: small caps fell just 0.58%, the Dow half the Nasdaq's loss, semis outperformed, the dollar rose 0.39%. VIXY woke up 5.1% to 21.78, the first real vol bid of the week, but from a level that never priced fear to begin with.

If

The pattern holds — duration sold, cyclicals sponsored — the market is trading a discount-rate reset inside an intact growth story, which is a rotation, not a top. A growth scare would look inverted: small caps and cyclicals leading down, bonds bid.

Why

Gold falling with stocks confused the haven crowd and should not have: when real yields are the thing moving, gold is on the same side of the trade as a hundred-dollar tech multiple. The board is internally consistent — every line on it is the same bet on the price of time, and the price of time went up.

Then

Watch whether today's vol bid extends or fades on the Intel gap. A tape that sells its havens and holds its leaders is uncomfortable — but it is not broken.

What

Ed Yardeni titled Thursday's QuickTakes 'Apocalypse Now! Or, TACO Now?' — the massive-attack rhetoric, B-1 deployments and a-head-for-an-eye doctrine on one side, and on the other his observation that the market is betting 'either Iran caves or Trump does.' He notes the S&P fell just 1.2% through the escalation and that geopolitical crises have historically been buying opportunities more often than turning points.

If

N/A

Why

The oil curve in point 5 is voting with Yardeni's second scenario: a $20 backwardation is not what sustained-war pricing looks like. And the political calendar leans the same way — a full-scale, deeply unpopular war weeks before the mid-term elections is the one trade the White House can least afford to put on. The morning's shift from attack rhetoric to making frozen Iranian funds pay for tanker damage reads like the first TACO tell.

Then

Read the piece against the USO chart: spot above 100, the ETF below its May highs, December at 80. For now, the market — like Yardeni's title — is asking the question and leaning toward the second answer.

Ed Yardeni · Yardeni Research ↗