AWS Grew 39% and Amazon Still Fell: The Market Has Stopped Asking How Fast the Cloud Grows

中文版 Chinese version

Figures below are from Amazon’s Q3 2026 earnings release on the evening of October 29, 2026, with analyst expectations as reported by Zacks (consensus figures vary slightly by provider). Share-price moves are snapshots from Friday, October 30 premarket trading, not closing prices.

The quarter that should have worked

On the numbers most investors claim to care about, this was a strong quarter. Amazon reported Q3 2026 revenue of $201.94 billion, up 12% year over year, and operating income of $28.1 billion, up 61%, for a record operating margin of 13.9%. AWS revenue grew 39% year over year to about $46 billion — its fastest pace in nearly five years — with AWS operating income up 55% to $17.8 billion. CEO Andy Jassy projected AWS growth above 40% in both 2027 and 2028, and pointed to new agreements with OpenAI, Meta and Nvidia, plus Amazon’s custom Trainium chips and the Bedrock platform, as evidence that the AI demand is real and contracted, not speculative.

And the stock fell roughly 7% in Friday premarket trading, according to Zacks. The misses that triggered it were, by themselves, tiny: revenue of $201.94 billion against a $202.11 billion Zacks consensus, and EPS of $1.89 against a $2.01 consensus (down from $1.95 a year earlier, a comparison distorted by a $9.5 billion Anthropic-related gain in the year-ago quarter). A $170 million revenue miss on a $202 billion quarter does not explain a 7% selloff. Something else did.

The spending line overwhelmed everything

That something is the capital-expenditure line. Amazon reaffirmed about $220 billion of capital spending for 2026, mostly for AI infrastructure, and warned that 2027 will cost “meaningfully more.” Third-quarter spending alone came in at $77.3 billion, well above the roughly $58.7 billion analysts expected, according to contemporaneous reporting. The cash consequence is now stark: quarterly free cash flow was negative $33.5 billion, and trailing-twelve-month free cash flow swung to a negative $42.4 billion, against positive $14.8 billion in the prior twelve-month period.

Put those two halves together and the market’s logic becomes legible. AWS accelerating to 39% is the demand story. Negative $42.4 billion of trailing free cash flow is the cost story. For years, hyperscaler earnings were judged mainly on the first half: is growth accelerating? This week — Alphabet disappointed on the same question days earlier — the judgment flipped to the second half: at what cash cost, and with what payback period? Jassy’s most quoted line from the call, that Amazon is installing an AI accelerator roughly every 14 seconds with customers absorbing capacity almost as fast as it comes online, was meant as reassurance. Read against a negative-$42 billion cash-flow print, it sounds like a treadmill: the faster you build, the faster you must keep building.

Why this is not simply a “bubble” verdict

It would be lazy to file this under “AI bubble pops.” Two facts resist that reading. First, the business is not unprofitable: a record 13.9% operating margin and 55% growth in AWS operating income describe a company converting growth into operating profit, then choosing to plow it — and more — back into capacity. That is a reinvestment decision, not a loss-making operation. Second, the holiday quarter itself is not the problem: Q4 guidance of $240–247 billion in revenue, 12–16% growth, landed roughly in line with the $245.65 billion consensus, with operating income guided to $28.5–33.5 billion. Investors did not sell a bad Christmas. They sold a 2027 spending plan they have not yet seen pay back.

The honest bull case and the honest bear case now share the same hinge, and it is Jassy’s above-40% AWS growth projection for 2027 and 2028. If AWS compounds at that pace, today’s $220 billion — and whatever “meaningfully more” turns out to mean — looks like buying the capacity that prints the next decade’s cash flow, and the negative free-cash-flow print is a trough, not a trend. If AWS growth decelerates toward the low 30s or below while spending stays at this level, the same print is the first year of a structurally cash-hungry business, and the multiple compresses accordingly. Nothing in this quarter settles which world we are in. The backlog number — contracted commitments reported at $578 billion, more than double a year earlier — is the bulls’ best evidence that demand is not imagined. The cash-flow number is the bears’ best evidence that demand at any price is not the same as demand at a profit.

The judgment

Amazon’s quarter marks a regime change in how the market grades hyperscalers, and it is worth naming precisely: growth is now necessary but not sufficient. A 39% AWS print that would have been celebrated two years ago was not enough, because the question has moved one level down — from revenue growth to cash conversion. Until Amazon can show a quarter in which AWS growth stays high while the capex line stops accelerating, every strong cloud print will be read as evidence for both sides at once: proof of demand, and proof of cost. The next datapoint that matters is not Q4 revenue. It is the 2027 capex number, and whether Jassy’s above-40% AWS guide survives contact with it.

Sources

• Zacks, “Amazon (AMZN) Stock After Q3 Earnings: Buy the Dip, Sell, or Hold?” October 30, 2026: https://www.zacks.com/commentary/3019984/amazon-amzn-stock-after-q3-earnings-buy-the-dip-sell-or-hold
• BeInSure, “Amazon (AMZN) shares drops as $220 bn AI spending pressures cash flow,” October 30, 2026: https://beinsure.com/news/amazon-shares-drops-as-220bn-ai-spending-pressures-cash-flow/