Amazon Q3 2026: $77.3B Capex in One Quarter, AWS Up 39%, FCF at -$42.4B

Amazon spent $77.3 billion on property and equipment in the third quarter of 2026, more than twice the $35.1 billion of a year earlier and, according to SiliconANGLE, far above the $58.7 billion analysts had pencilled in. The earnings release published on 29 October 2026 also showed AWS revenue up 39% to $46.0 billion, its fastest growth in nearly five years, and trailing twelve-month free cash flow of negative $42.4 billion. Amazon kept its 2026 plan of roughly $220 billion in cash capital expenditures, and CEO Andy Jassy said spending will rise again in 2027.

What happened

Net sales rose 12% to $201.9 billion. Operating income reached $28.1 billion, a 13.9% margin that Amazon says is its highest ever. AWS contributed $17.8 billion of that, up from $11.4 billion a year earlier, and the business now runs at a $184 billion annualised revenue rate. Jassy's statement in the release says AWS is projected to grow "over 40% both in 2027 and in 2028".

Earnings per share of $1.89 missed the $1.96 consensus, and revenue was a touch under estimates. The shares fell more than 7% in late trading, then recovered most of that: SiliconANGLE reported a drop of only about 2% on the Friday session. On the call, Jassy said Amazon plans to "spend more meaningfully in 2027" and cited a "clear line of sight to attractive financial returns".

The numbers

The free cash flow swing is the headline number for anyone thinking about the balance sheet. Over twelve months, purchases of property and equipment net of proceeds and incentives rose by $94.8 billion, and the release states plainly that the increase "primarily reflects investments in artificial intelligence".

One thing in the guidance does not add up neatly, and I want to flag it rather than paper over it. Amazon has already spent $175.7 billion gross on property and equipment in the first nine months of 2026. If the $220 billion "cash capital expenditures" figure maps directly to that line (net of about $3.5 billion of proceeds so far), Q4 would come in near $48 billion, far below Q3. The release does not reconcile the two, so either Q3 pulled spending forward or the definitions differ. I would wait for the 10-K before drawing conclusions.

Why it matters

The 10-Q puts long-term performance obligations, primarily AWS, at about $578 billion, with a weighted average remaining contract life of 6.1 years. That is the demand Amazon is building against. The same filing shows energy contracts covering roughly 690 million megawatt-hours over an average of 18 years, most of it to be delivered more than a decade out. Amazon is not just buying servers; it is locking in power for the 2030s.

There are also hints of where the capacity goes. The release lists new AWS agreements with Meta, NVIDIA, OpenAI and Oracle among others. The 10-Q describes a financing facility of up to $20.0 billion for Anthropic, drawable as Amazon delivers compute capacity milestones. In Q2 Amazon also put $5.0 billion into Anthropic Series G preferred stock, and a further $5.0 billion into Series H under the facility, which cut what remains available to $15.0 billion. Some of AWS's growth is effectively vendor-financed.

Local politics now shows up in the earnings release too. Amazon announced "Built Together", a $1 billion programme for communities near its US data centers, funding trades training, efficiency upgrades and local priorities like roads and fire equipment. You do not spend that money unless permitting and community opposition have become a real constraint on where you can build.

What sceptics say

The bear case: a company that generated $14.8 billion of free cash flow a year ago is now $42.4 billion in the hole on a trailing basis, and management has promised to spend more next year. The guidance also lists "resource and supply volatility, including for memory chips" among its risks, which means each dollar of capex may buy less hardware than planned.

The bull case is that AWS margins rose while it grew, and that analysts have read the spending as a supply problem rather than a demand problem. Forrester's Tracy Woo told SiliconANGLE that the profile suggests "Amazon sees future capacity constraints rather than demand constraints." Compare Microsoft's quarter, where Azure grew 48% while some third-party capacity arrived late, and Alphabet's, where Cloud doubled. All three say demand exceeds supply. If that is wrong, the correction will be painful for everyone at once.

What it means for engineering and DevOps teams

My view: the line in these filings that matters most for practitioners is not the capex figure, it is the 6.1-year average contract life on the backlog. AWS is filling its new capacity with long commitments from a small number of very large buyers. Teams that buy on demand are at the back of that queue for GPU instances, especially in the newest regions.

Practically, that means three things. First, reserve what you know you need, and keep what you do not know on a provider-neutral path (containers, Terraform modules that can target more than one cloud, and inference behind a gateway). Second, treat Trainium as a real option rather than a curiosity: Jassy said the chips business (Graviton and Trainium) is growing at triple-digit rates, and SiliconANGLE reports Amazon plans to make Trainium chips available to external customers toward the end of next year, and PyTorch support via TorchNeuron lowers the porting cost. Third, revisit your Kubernetes cost work: when capacity is scarce, the cheapest instance is the one you are already paying for and not using.

None of this needs new tooling. It needs the boring discipline of knowing your utilisation and your commitments, which is what platform teams have always been for.

FAQ

How much did Amazon spend on capex in Q3 2026?

Purchases of property and equipment were $77.3 billion in the quarter ended 30 September 2026, against $35.1 billion a year earlier. Amazon still expects about $220 billion of cash capex for 2026.

How fast is AWS growing?

AWS revenue grew 39% year over year to $46.0 billion in Q3 2026, an annualised run rate of $184 billion. Amazon projects growth above 40% in both 2027 and 2028.

Why is Amazon's free cash flow negative?

Trailing twelve-month operating cash flow was $168.3 billion, but net purchases of property and equipment were $210.7 billion. The result is free cash flow of negative $42.4 billion, driven mainly by AI infrastructure.

Related: AI Power in 2026: 485 TWh, a 0.3 GW Stargate, a 116 GW Turbine Backlog, and 1,800 MW That Dropped Off the Grid in Seconds, Kubernetes Cost Optimization in 2026: 8% CPU Use and What Fixes It and AWS vs Azure vs GCP 2026: Which Cloud to Learn First, by the Numbers.