Amazon's Margin Machine Is Working. The $131 Billion Question Is What Happens Next.

Amazon is three distinct businesses operating under one balance sheet: a retail and logistics network, a cloud infrastructure platform, and a digital advertising exchange. Each has different economics, different competitive dynamics, and different growth trajectories.

Retail generates volume and customer relationships but thin margins. AWS generates the cash. Advertising is a high-margin business scaling faster than most investors model. The interaction between these three segments is where the real analytical work happens.

Understanding Amazon in 2026 requires holding all three simultaneously. Investors who treat it as a retailer miss the AWS story. Investors who treat it purely as a cloud play underestimate how advertising and logistics network effects reinforce each other.

Start with the raw numbers. Revenue grew from $514 billion in 2022 to $716.9 billion in 2025, a 39% increase. Operating income over the same period grew from $12.2 billion to $80 billion, a 556% increase. That is not a small improvement. That is a structural shift in the economics of the business.

Gross margin expanded from 43.8% in 2022 to 50.3% in 2025, reflecting the growing contribution of AWS and advertising, both of which carry margins well above the company average. The mix shift is structural, not cyclical.

EBITDA grew from $54.2 billion in 2022 to $165.3 billion in 2025. That is more than a tripling of cash earnings in three years. At current enterprise value, the EV/EBITDA multiple of 13.8x looks conservative against that trajectory.

The earnings beat pattern reinforces the picture. Amazon beat consensus EPS estimates by 22.3%, 25.4%, 25.4%, 16.9%, 26.3%, and 26.6% across six consecutive quarters through Q3 2025. These are not marginal beats driven by share count games. They reflect genuine operational outperformance.

AWS is the profit engine that funds Amazon's everything-else ambitions. Cloud infrastructure is a business with high switching costs, compounding scale advantages, and margins that retail cannot approach.

The competitive dynamics in cloud favour the incumbents. AWS, Azure, and Google Cloud have established such deep integration with enterprise workloads that displacement requires not just a better product but a credible migration path and significant organisational will. For most enterprise customers, neither condition is easily met.

AWS's strategic position in the AI infrastructure buildout is particularly important. In April 2026, AWS confirmed it is simultaneously invested in both Anthropic and OpenAI, a dual-bet structure that positions AWS as infrastructure provider regardless of which foundation model architecture ultimately wins. It is a capital allocator's hedge, not a conviction call on AI outcomes. That distinction matters.

The $131.8 billion capex program is substantially directed at AWS data centre expansion, custom silicon development, and the physical infrastructure required to serve generative AI workloads. Amazon estimates demand for cloud capacity is growing faster than their ability to build it. If that assessment is accurate, the current investment phase should produce a multi-year capacity absorption cycle with improving returns.

Amazon's advertising business is growing faster than AWS and carries comparable margins. Sponsored product placements, display advertising across Amazon properties, and streaming ad inventory on Prime Video represent a category that most sell-side models still underweight.

The structural advantage is intent. Advertising on Amazon reaches consumers at the moment of purchase consideration. That intent signal makes Amazon advertising inventory structurally more valuable per impression than social or display advertising. Conversion rates are measurably higher. That premium will not erode quickly.

Prime Video's shift to an ad-supported model in 2024 added streaming inventory at scale. Amazon has avoided discussing specific advertising revenue figures within its broader segment reporting, but the directional picture from gross margin expansion is clear: high-margin advertising is growing as a share of the total revenue mix.

The advertising business also reinforces the retail flywheel. Brands that advertise on Amazon get better placement and discovery, which drives more sales, which generates more customer data, which makes Amazon's advertising targeting more effective. The loop compounds.

Capital expenditure reached $131.8 billion in 2025. To put that in context: Amazon spent $52.7 billion in 2023. In two years, annual capex has increased by $79 billion. That is a number without precedent in corporate history.

The consequence for free cash flow is severe. Operating cash flow grew from $115.9 billion in 2024 to $139.5 billion in 2025, a strong result. But capex consumed $131.8 billion of that, leaving free cash flow of just $7.7 billion, down from $32.9 billion in 2024. Amazon's reported FCF yield has collapsed.

The market is being asked to treat this as a temporary distortion rather than a structural problem. That framing is reasonable but not guaranteed. History contains examples of large technology capex cycles that produced excellent returns and examples where they did not. AWS's own founding required sustained heavy investment before returns materialised.

Amazon does not pay dividends and is not currently running buybacks at scale. The entire capital return story depends on capex eventually normalising and FCF recovering. The question is timing.

At 13.8x EV/EBITDA, Amazon trades below where pure-play cloud businesses have historically been valued. That discount reflects two things: the retail drag on blended margins and genuine uncertainty about the capex payoff timeline.

On a trailing P/E of 29.8x, the stock looks expensive relative to the S&P 500 average. That framing is misleading. Amazon's earnings have been deliberately suppressed by reinvestment, and the earnings power embedded in AWS and advertising at normalised capex levels would imply a materially different earnings figure.

Analysts have a strong consensus: 47 strong buy ratings, 19 buys, 4 holds, and zero sell or strong sell recommendations. The average price target of $281 implies approximately 33% upside from recent trading levels near $214. That target distribution is not typical sell-side optimism — it reflects genuine conviction that the capex cycle resolves positively.

The EV/EBITDA multiple is the cleaner lens. If EBITDA continues expanding at its current trajectory and capex normalises modestly over the next two years, even a modest multiple re-rating produces substantial returns.

Amazon's competitive position is not a single moat. It is three overlapping advantages that reinforce each other in ways that make the overall position more durable than any one layer alone.

The logistics network is the most capital-intensive piece and the hardest to replicate. Amazon has spent decades building fulfilment infrastructure in proximity to population centres. The speed and reliability that comes from that network is not something a competitor builds in two years with a favourable funding environment.

AWS's moat is enterprise integration depth. Once a company's data, applications, and development workflows are built on AWS, migration is expensive, disruptive, and risky. The moat compounds with each new service layer that enterprise customers adopt.

The advertising moat is the least discussed but arguably the fastest-compounding. Purchase intent data, at scale, across hundreds of millions of active buyers, is not something that can be replicated. The longer Amazon has been collecting that data, the better its targeting, the higher its conversion rates, and the more it can justify charging per impression.

The Q1 2026 earnings consensus estimate sits at $1.73 EPS. The prior four quarters produced beats ranging from 17% to 27%, which has understandably made analysts cautious about getting caught under-modelled again. The setup favours outperformance, though the magnitude of recent beats may moderate as models have been revised upward.

Operating margin should continue expanding if AWS mix keeps growing. Advertising will add incremental high-margin revenue regardless of macroeconomic conditions, because the performance advertising market is tied to sales outcomes, not sentiment.

Capex may have peaked or is close to it. Amazon has signalled in recent commentary that the current build rate reflects a period of extraordinary demand response, not a permanent operating model. Any moderation in capex will show up immediately in FCF improvement, which remains the stock's most important catalyst in the next 12 months.

Sentiment has been broadly positive over the trailing 30 days, with average scores above 0.65, though there was a notable dip to 0.385 on March 27 likely related to broader tariff concerns. The sentiment picture has since recovered. Fundamentals and sentiment are aligned, which is a less interesting setup than divergence but a more comfortable one.