Amazon's cloud boom no longer buys AI capex a free pass

What does it take for a cloud business to grow 39% and still get punished by the market? Apparently, spending more than the whole company can pay for. That is the lesson of Amazon's third quarter, and we think it marks a change of mood in the AI trade.

Amazon nailed the growth and still got sold

AWS grew revenue 39% to $46 billion, and as dpa reported, its operating income reached $17.8 billion, up 55%. Group revenue rose 12% to $201.9 billion, yet the quarter's numbers and the holiday-quarter outlook came in below expectations. Amazon has budgeted $220 billion in capital expenditure for this year alone.

The cost showed up in cash. ORF reports that the cash outflow over the past three months nearly sixfold to $42.4 billion, and the stock dropped as much as roughly 8% in after-hours trading.

Growth is no longer the argument; cash coverage is

For two years the implicit deal was simple: build the data centers, and the cloud numbers will vindicate you. Amazon just delivered its fastest cloud growth in about five years, and the deal didn't hold. That tells us the market has stopped grading the growth line in isolation and started grading it against what it costs to produce.

Look at Alphabet, which got hit the very same day. According to dpa-AFX, Alphabet's free cash flow slipped to about minus $9 billion, and the stock closed down 12.6%, wiping out almost $500 billion in market value in a day. One analyst quoted there, Jochen Stanzl of Consorsbank, summed it up: the market rewards spending discipline and punishes investment without cash backing. It's a blunt line, and we think it's basically right.

This doesn't mean AI spending is being rejected. It means the burden of proof moved. When a company funds capex out of operations, it gets the benefit of the doubt. When it funds it by draining the treasury, every dollar needs a story.

Microsoft shows what the passing grade looks like

The contrast with Microsoft is the most useful part of the week. Microsoft's capex rose 30% to $45.2 billion, and yet, per the Raiffeisen analysis, free cash flow was $18.6 billion, positive and well above consensus, while Azure grew 48%. Even so, the stock ended slightly down, partly because the company is raising billions in the bond market to keep investing.

Read that carefully. The hyperscaler in the best shape still isn't getting a rally, and it is tapping debt markets. We read that as a sign that this race is increasingly decided by balance-sheet strength rather than enthusiasm. In our view, that favors whoever has the fattest operating cash flow, and it puts pressure on anyone whose core business, like Alphabet's search, might itself be eroded by the technology it is financing.

Amazon's defense is real, but it's a promise

To be fair to Amazon, the bull case isn't flimsy. Management says, per its earnings release, that AWS investment is working, pointing to a $184 billion revenue run rate and projected growth above 40% in both 2027 and 2028. A Seeking Alpha headline also cites a $496 billion backlog. Those are the kind of numbers that make a capex plan look like pre-selling rather than speculation.

But note the tense: projected. Reuters notes that Amazon also signaled higher capital expenditures next year on top of an already huge 2026. Investors are being asked to finance a forecast with real cash today. Meanwhile, the retail side, which guided the holiday quarter below expectations, doesn't look like the comfortable cash engine that could absorb any overrun without a murmur.

Our verdict

The AI race has entered its balance-sheet phase. Model quality and cloud growth still matter, but the question that now moves stocks is who can keep building without running out of financial oxygen. Our read: Microsoft is the model of a defensible position, Alphabet is the cautionary tale, and Amazon sits in between, with the best growth and the ugliest cash line.

If you follow these stocks, stop reading the cloud growth percentage as a verdict. Watch free cash flow, bond issuance and how capex guidance evolves for 2027. If AWS really does stay above 40%, Amazon will look prescient. If it doesn't, $220 billion will look like the price of a lesson the whole sector is now learning in public. This isn't investment advice, just an opinion on where the scrutiny is headed.