Amazon Passed the Test, Apple Failed It: The Invoice Thesis Closes the Month With a Clean Sweep

1. Macro Pulse

The setup: a broad, tech-led rally to close July, with Amazon’s cloud beat reinforcing confidence that AI investment is translating into results. The move is powerful and earned, but it sits on top of a bond market that has not relented: the 30-year at a two-decade high is the same warning flagged yesterday, still unaddressed. [CNBC/TipRanks, 07/31]

Thursday’s close was a strong rebound: the Dow rose 613.92 points (+1.2%) to 52,208.06, the S&P 500 climbed 1.66% to 7,437.63, and the Nasdaq Composite finished 2.8% higher at 25,122.18, recovering most of Wednesday’s 1,153-point Dow rout. Microsoft surged about 15% on the session after its Azure beat. JPMorgan noted hedge funds have likely finished selling their tech, chip and memory positions to reduce leverage, removing an overhang. [CNBC, 07/31]

Overnight, Korea staged the mirror image of Tuesday. The Kospi surged 17.9% to a record close of 6,595.45, its best single day ever, after two consecutive circuit-breaker crashes earlier in the week. The reversal, driven by Microsoft and Amazon validating the AI-demand thesis that SK Hynix’s selloff had questioned, shows how violently positioning-driven the memory complex has become in both directions. [CNBC, 07/31]

The bond market, still the story beneath the story: the 30-year Treasury yield sits near 5.24%, a two-decade high, after Wednesday’s hold drew three dissents favoring a hike and Warsh refused to signal any inflation tolerance. The 10-year is near 4.70% and the 2-year 4.28%. The curve has steepened sharply, reflecting reduced near-term tightening expectations against rising long-term inflation and fiscal concern. The dollar index trades near 100.8 to 100.9. [Bloomberg/UOB, 07/31]

This morning’s data: Thursday’s figures set the tone. June headline PCE fell 0.1% and core PCE rose just 0.1% month over month, both a touch softer than the 0.2% core expected, a mild relief on the inflation front even as the Q2 core PCE deflator held at 3.4% annually. Today brings Chicago PMI and the University of Michigan final sentiment and inflation expectations. [Schwab, 07/31]

Commodities: oil eased as tankers resumed transiting the Strait of Hormuz. Brent slipped 0.6% to about $88.70 and WTI 1% to about $83.40, though Saudi Arabia proposed a naval coalition to protect trade routes as Red Sea attacks continued. Gold holds near $4,020 to $4,050, capped by the elevated real-yield backdrop. VIX fell to about 18 Thursday as the rebound took hold. [TipRanks/StreetStats, 07/31]

2. Economic Calendar

Season scorecard: with the four megacaps now reported, aggregate Q2 S&P 500 earnings growth has tracked well above expectations, but the market’s reaction has been ruthlessly selective, rewarding proven monetization and punishing everything else. The dispersion, not the beat rate, has been the story of the season. [Schwab, 07/31]

[ANALYSIS] The softer June core PCE is the one piece of genuinely good news for the bond market this week, and it is why the 30-year has stabilized rather than kept climbing. But a single month of 0.1% core does not offset a 3.4% annual run rate or Warsh’s stated refusal to tolerate above-target inflation. Next week’s jobs report is the more important test: a hot number revives the hike, a weak one deepens the stagflation worry. Today is a month-end lull between this week’s earnings storm and next week’s labor data.

3. Earnings Spotlight

Amazon and Apple were the final two megacaps, and they split exactly along the line the whole week drew. The invoice thesis went four for four: monetization proven is bought, monetization absent or guidance soft is sold.

4. Pre-Market Movers: Upside

5. Pre-Market Movers: Downside

6. Sector Rotation Radar

[ANALYSIS] The month resolved into a barbell. At one end, proven-monetization tech (Microsoft, Amazon) with visible cloud returns; at the other, cash-harvesting energy (Exxon, Chevron) returning capital rather than consuming it. Both ends outperformed. What lagged was the muddled middle: companies spending on the future without yet showing the return (Meta), or priced so richly that any stumble broke them (Apple). The market’s message across the whole month is consistent: it will pay for a visible return on capital, whether that return comes from an AWS backlog or an oil dividend, and it will not pay for the promise of one.

7. Today’s Session Playbook

1. Month-end, so respect the mechanics. The final session of July can see rebalancing flows that move prices independent of news. With a strong month for equities behind it, some of today’s action is portfolio mechanics, not fresh information. The real catalysts are next week.
2. Top story: the invoice thesis closed four for four. Microsoft and Amazon proved monetization and rallied; Meta and Apple stumbled and fell. The rule that has defined this tape since early July held through every megacap report. Show the return or get sold.
3. S&P 500 levels: Thursday closed 7,437.63, reclaiming the 7,400 shelf and the 7,428 pivot. Futures point to an open near 7,470. Resistance is 7,500, then the 7,620 all-time high; support is 7,400, then 7,316 (Wednesday’s low). The index has clawed back the entire Fed-day rout in a single session, a sign the AI-monetization proof outweighed the hawkish Fed for now.
4. Event risk: the 30-year and next week’s jobs data. The equity rally is running over a two-decade-high long-bond yield. The July jobs report next Friday is the trigger that decides whether the 30-year challenges new highs and caps the rally, or eases and lets it run. Today is the calm before that.
5. Under-the-radar: the memory-price squeeze in Apple’s guidance. Apple guided down partly on memory-price surges, the clearest corporate confirmation that the AI buildout is causing real component scarcity. That is bullish for memory makers and a cost headwind for every hardware assembler. Watch it ripple through the supply chain next quarter.

Four weeks ago this brief proposed that the market had stopped paying for AI spending and started paying for a visible return on it. This week was the definitive test, four hyperscalers reporting their own capex against their own monetization, and the thesis did not break once. Microsoft grew Azure 43% with steady capex and rose. Amazon grew AWS 37% with a $496B backlog and rose 13%. Meta let free cash flow collapse 91% with no cloud to show and fell 8.5%. Apple, on the paying end of the invoice as a component buyer, guided down on supply costs and fell 7%. Two proved the return and were rewarded; two could not and were sold. As a controlled experiment in how this market prices artificial intelligence, the result is as clean as markets ever produce.

The refinement worth carrying forward is that the invoice has two sides, and this week illuminated both. The earning side, Microsoft and Amazon, books the AI spend as cloud revenue with proven margins, and gets paid. The paying side runs deeper than the hyperscalers: Apple guided down because the same memory-price surge and chip scarcity that reward SK Hynix and Micron are now a cost line for every device maker. AI capex is simultaneously revenue for the suppliers, a justified investment for the monetizers, an unjustified cost for the spenders who cannot yet show returns, and an input-cost shock for the buyers downstream. The market spent July learning to price each position on that chain separately, and by month-end it did so with real precision. That is a more discriminating market than the one that bought everything AI-labeled in the spring, and a healthier one.

But the closing of the AI question only sharpens the macro one, and intellectual honesty requires ending the month on it rather than on the earnings win. Every one of these equity moves happened above a 30-year Treasury yield at 5.24%, a two-decade high, put there by a Fed chair who will not signal any tolerance for inflation and a Q2 that paired 1.5% growth with a 3.4% core PCE run rate. The softer June monthly core print stabilized the long bond this week, which is the single reason the rally could breathe. But the term premium has reset, the curve has steepened, and the bond market is pricing a future in which inflation and fiscal risk keep long rates high regardless of what the AI winners earn. A 5.24% long bond discounts Amazon’s backlog and Microsoft’s Azure just as surely as it discounts everything else.

So the edge going into August is a two-front one. On the microeconomic front, the rule is settled and tradeable: own the proven monetizers and the cash-returning energy majors, treat spend-without-return as a value trap until margins turn, and watch the AI supply chain for where component scarcity creates winners and losers. On the macroeconomic front, the question is wide open and more important: whether the bond market’s revolt is a repricing that stabilizes here or the start of a sustained move that eventually caps even the best equity stories. Next Friday’s jobs report is the first real test. July proved that AI, for the companies that can monetize it, is real and is being paid for. August will test whether that matters when the risk-free rate on thirty-year money sits at a level the market has not seen since before the financial crisis. Trade the winners, but keep one eye on the long bond, because that is where the next regime is being written.