Amazon (AMZN) reported its June quarter after the close on July 30. The headline that traveled furthest was earnings per share of $5.75 against a consensus estimate near $1.81, a beat of more than three times. That number is close to meaningless as a read on the business. What matters was sitting underneath it.
What Amazon reported
Net sales were $200.6 billion, up 20%, about $4 billion above consensus and the company's first $200 billion quarter. Operating income rose 43% to $27.5 billion. That is the record worth keeping.
The segments:
- AWS: $42.2 billion, up 36.7% — the fastest growth in 18 quarters. Segment operating income of $16.6 billion on a 39.4% margin.
- North America: $116.2 billion, up 16%; International: $42.2 billion, up 15%.
- Advertising: $19.8 billion, up 26% — high-margin revenue that gets little attention.
The number under the number
Net income was $62.6 billion, against $18.2 billion a year earlier. Of that, $53.4 billion was non-operating pre-tax income, primarily a mark-to-market gain on Amazon's stake in Anthropic. It is an accounting revaluation of a private holding: no customer paid it, nothing shipped for it, no cash arrived.
That single item is why the reported EPS and the analyst estimate were describing different things: consensus estimates model the operating business, and the print contained a revaluation nobody forecasts. Strip it out and operating profit still grew 43%. That is an excellent quarter, and it needs no help.
The second tension runs the other way. Trailing-twelve-month free cash flow was negative $7.6 billion, down from positive $18.2 billion a year earlier. Capex in the quarter alone was $54.2 billion against $32.1 billion, and management raised 2026 capex guidance from $200 billion to $220 billion, blaming higher memory-chip prices. So a single quarter produced both the largest paper gain in Amazon's history and negative trailing cash generation. Neither number says much about how the operating business is actually performing.
What mattered in the operating business
Three things, none of them EPS.
AWS reaccelerated. Growth of 36.7% on a $42 billion base matters more than a higher rate on a smaller one, and it reversed a multi-year deceleration narrative. One asterisk: about $600 million of segment operating income came from mark-to-market gains on energy derivatives. Excluding those, management put underlying margin expansion at 520 basis points rather than the reported 650, and guided Q3 assuming no repeat.
The backlog. AWS's contracted, not-yet-delivered backlog reached $496 billion, up $132 billion in the quarter — close to three times its current annualized revenue run rate of about $169 billion. Not a pipeline or a forecast: signed commitments.
Capacity, not demand, is the binding constraint. Andy Jassy told analysts Amazon will not have enough capacity to meet 2026 demand, expects the same in 2027, and called the demand already visible for 2028 striking. The custom-silicon business is now above a $25 billion annual run rate, underpinned by Anthropic's April commitment — ten years, $100 billion, up to five gigawatts of Trainium.
Q3 revenue guidance of $197–202 billion came in below consensus of roughly $204 billion, which management attributed to Prime Day landing in Q2 this year (nearly 400 basis points of growth) plus 80 basis points of currency. Microsoft (MSFT) reported days earlier with a similar pattern — Azure up 43% alongside a raised capex outlook (full breakdown). In both cases the question an analyst asks of an enormous buildout is the same: is the spending visibly arriving as revenue?
What the quarter says about the moat
Here is the part the headline does not change. A quarter this good does not upgrade Amazon's moat rating.
MoatScan AI's AMZN moat analysis was re-run on August 1, 2026 — after this report — and still reads 75/100, Narrow Moat, with a Positive trend and a Strong management verdict. Microsoft, on the same five-pillar framework, reads 77/100 and Wide.
Two parts of the quarter are real moat evidence. The backlog is a switching-costs statement: $496 billion of contracted, undelivered cloud work is customers pre-committing years of workloads, data gravity and architecture to one provider. Switching costs is already Amazon's strongest pillar at 8.5/10, and a multi-year contract is the most literal form the pillar takes. Custom silicon is a cost-advantage statement: designing Trainium and Graviton lets AWS serve AI workloads on hardware whose margin it keeps rather than pays away. Cost Advantages scores 7.5/10.
What the quarter does not fix is the pillar that keeps the rating Narrow: Efficient Scale at 5.5/10, far below everything else in the profile. Efficient scale is the moat of markets too small to be worth fighting over, and Amazon's markets are the largest and most contested on earth. Retail contends with Walmart, Temu, Shein and every brand's own storefront; AWS with two rivals carrying comparable balance sheets, one of which just grew faster. Our analysis puts it as efficient scale existing in selected submarkets rather than across the enterprise. A three-way capex race is the opposite of a market whose economics deter entry: Amazon is guiding to $220 billion of spending precisely because standing still is not available.
That is the whole distance between 75/Narrow and 77/Wide — not a verdict on business quality, but on how much of the profit stream is protected by structure rather than by continuing to outspend everyone else. Our AI assessment reads the same way from another angle — Opportunity 7, Threat 6, a net of roughly +1. Amazon is among the clearest beneficiaries of AI infrastructure demand and among the most exposed if that demand disappoints, the two-sided dynamic covered in How AI Is Reshaping Economic Moats.
The financials, and two model assumptions worth examining
As of that same August 1, 2026 analysis, Amazon carries a Financial Score of 80/100 — Income Statement 8.5, Balance Sheet 8.5, Cash Flow 6.5, Key Ratios 8, Growth 8.5. Cash Flow is the honest grade there, and the June-quarter figures make it read worse, not better.
The seven-year discounted cash flow model behind the analysis runs on an 8.1% WACC and 4.0% terminal growth, and two of its other assumptions pull in opposite directions. Modelled revenue growth of 7.7% against the 20% Amazon just printed is genuinely conservative. Capex at 13.2% of revenue is not: this quarter's $54.2 billion of capital spending on $200.6 billion of sales was about 27%. The statement data underlying the model runs through the twelve months ended March 31, 2026 — neither the June-quarter step-up nor the raised $220 billion guide is in it yet. If buildout intensity stays near current levels rather than halving, the model's output falls — and the depreciation from today's servers lands on future income statements regardless. That sensitivity, more than any single output, is what the model is useful for; DCF Valuation Explained walks through why.
The scorecard
The open questions after this report are the ones that were open before it: whether AI demand holds at this intensity, whether depreciation compresses margins faster than revenue can outrun it, and whether three hyperscalers spending this hard against each other can all earn adequate returns. What the quarter settled is narrower — Amazon's spending is showing up as accelerating revenue and contracted backlog rather than as an unexplained bill.
The live version of everything quoted here — the five-pillar breakdown, the financial sections, the valuation model and its assumptions — is on MoatScan AI's AMZN analysis and will change when it is next re-run; this article deliberately freezes the August 1, 2026 snapshot. For how Amazon's moat compares with similarly-rated companies, see the narrow moat stocks list.
MoatScan AI is a free educational research tool. This article explains how a business and its competitive position work; it is not investment advice or a recommendation to buy, sell or hold any security. Figures are as of the dates stated and will change.