Amazon Free Cash Flow: -$42.4B After a 39% AWS Quarter

Amazon reported third-quarter results on Thursday, October 29, and the wire headlines said it missed. By the numbers, the shortfall looks small to me. Earnings were $1.89 a share against $1.97 expected by FactSet, per the AP, and sales were $201.9 billion against $202.2 billion. Amazon Web Services did better than expected.

The stock fell about 7% in after-hours trading anyway. The AP said investors worried about Amazon burning cash on AI. That number sits lower down the release, in the supplemental figures and the cash flow statement it comes from.

The one number

Amazon says free cash flow for the trailing twelve months was an outflow of $42.4 billion. A year earlier it was an inflow of $14.8 billion. Amazon’s own explanation is short: purchases of property and equipment, net of sales and incentives, rose by $94.8 billion, mostly for AI.

Operating cash flow didn’t fall. It rose 29% over those twelve months, to $168.3 billion. The spending just rose faster.

The release also lets you work out the quarter on its own. This is my arithmetic from the cash flow table, not a figure Amazon reports:

Quarter	Operating cash flow	Net spending on property and equipment	Free cash flow
Q3 2025	$35.5B	$34.2B	+$1.3B
Q3 2026	$42.4B	$75.9B	-$33.5B

Cash coming in grew about 19%. Cash going out for data centers, chips and the rest more than doubled. Amazon still says it expects about $220 billion of cash capital spending for all of 2026, and CEO Andy Jassy said on the call it will rise meaningfully in 2027.

So was it a miss?

Against analyst estimates, mildly, in my reading.

Line	Reported	Estimate
Sales	$201.9B	$202.2B (FactSet, via AP)
Earnings per share	$1.89	$1.97 (FactSet, via AP)
AWS sales	$46.0B	$45.3B (StreetAccount, via CNBC)
Q4 sales guidance	$240B to $247B	$244.2B (analysts, via AP)

Two things soften the sales number. Prime Day moved into the second quarter this year for most large countries, and Amazon says that alone shaved about 4 percentage points off third-quarter growth. And last year’s earnings included $9.5 billion in pre-tax gains from Amazon’s Anthropic investments, so this year’s net income of $20.7 billion looks lower than last year’s $21.2 billion and in my reading there’s no operating weakness behind that, since operating income rose 61%. Operating income was $28.1 billion.

The fourth-quarter range does sit slightly under the $244.2 billion analysts wanted, at a midpoint of $243.5 billion. I’d call that a small gap.

What AWS looks like

AWS sales rose 39.3% to $46.0 billion, which Jassy called the sixth straight quarter of acceleration, with an annualized run rate of $184 billion. AWS operating income was $17.8 billion. Jassy said backlog is $578 billion and projected growth above 40% in both 2027 and 2028. He also said Amazon is installing an AI chip about every 14 seconds.

Those are Amazon’s numbers and its forecasts. In my view a forecast two years out is a target, not a result, and the backlog is the harder figure to wave away, though I haven’t seen the terms behind it.

My read

Alphabet showed the same shape a day earlier. As I wrote in the Alphabet cash flow post, its cloud revenue doubled from a year earlier and it still burned $9.1 billion. Amazon’s burn for the same three months, by my math, was about 3.7 times that.

I don’t think the market is saying AWS demand is fake. The AFP report quoted a Floodlines Research analyst making the same distinction: the worry isn’t that AI demand has slowed, it’s that higher funding costs and credit risk could eventually force the big cloud companies to slow the build-out. Intel, which fell about 23% the same week for different reasons, is covered in our Intel post.

What I’d watch is the 2027 capex number. Amazon projects AWS growth above 40% for 2027 and 2028 while spending more. If that growth comes in a few points short, the free cash flow line is where everyone will see it first.

Note on sources

Amazon’s press release was read directly. Analyst estimates differ by data provider (LSEG, FactSet, StreetAccount), so the table names each source. The AP and CNBC figures were checked from the articles’ text and cached pages, and the after-hours move was reported as ‘about 7%’ by the AP and ‘more than 7%’ by Euronews, while CNBC’s page said more than 6%. Call details come from MarketBeat’s summary of the call, not the full transcript. The after-hours stock move is as the AP and AFP reported it, and prices may have changed since.

Amazon's cloud grew 39% and the stock slid 7% after hours. The cash flow statement has the answer.