Investment analysis by Ruslan Averin — originally published at averin.com.
The busiest week of the quarter arrives with a single question hanging over it: is the AI spend worth it? Four of the largest companies on earth answer in the next few days — Microsoft and Meta after the close on July 29, Apple and Amazon on July 30 — and our analysts think the market's reaction will hinge less on earnings than on one line item.
The four tests
| Company | Reports | The market's real question |
|---|---|---|
| Microsoft | July 29 (after close) | Is Azure growth keeping pace with capex? |
| Meta | July 29 (after close) | Does ad revenue justify a $100B+ AI budget? |
| Apple | July 30 (after close) | Where is Apple's AI story — and China demand? |
| Amazon | July 30 (after close) | AWS margins versus the buildout |
The four hyperscalers have guided combined 2026 capital expenditure to roughly $725–785 billion, up around 77% year over year. That number is the story. Two weeks ago the megacap complex had its worst day since the April 2025 tariff selloff — the Magnificent Seven shed close to $800 billion in a single session — not because anything broke, but because investors put the soaring capex line and the flat revenue line on the same page and didn't like the gap.
Why the reaction won't match the results
The team's base case is that all four beat on earnings — these are extraordinarily profitable businesses and the bar for the past quarter is beatable. The risk isn't the print; it's the guidance. If any of them raises capex again — as Alphabet did on July 23, lifting its 2026 range to $195–205 billion and falling anyway — the market is now inclined to punish rather than reward it. The reflex has flipped: for three years higher AI spending was a bullish signal of demand; today it reads as a widening funding gap until proven otherwise.
The cleanest tell is the ratio, not the raw number. Our analysts will be watching capex growth against cloud revenue growth, company by company. As long as capex compounds near 60–80% and cloud revenue doesn't, multiple compression is arithmetic, not sentiment — and no earnings beat fixes arithmetic.
The other side
The bull case deserves a fair hearing. Under-building would be far more expensive than over-building if AI demand is real, and every one of these management teams has said the risk of spending too little exceeds the risk of spending too much. They may be right — AWS looked reckless in 2013 too. The difference is that AWS was already selling everything it built. That's the test this week's numbers have to start passing.
What we're watching
One, cloud and AI revenue growth versus capex growth — the ratio, not the headline. Two, any capex guidance revision, up or down, and how the stock trades on it. Three, free cash flow, where the damage from the buildout shows up first. Four, margin commentary, because depreciation from this capex lands on the P&L for years whether or not the revenue arrives on time.
Bottom line: four beats are likely; four rallies are not. In this tape, guidance on the buildout matters more than the quarter behind it — and the market has stopped taking the buildout on faith.
This is analysis, not investment advice.
More market analysis by Ruslan Averin at averin.com.
Top comments (0)