How We Analyze Amazon

How we analyze Amazon: the business first

How we analyze Amazon is the way we analyze any company: business quality first, then the balance sheet, then valuation, then the long-term record. The order is deliberate, and none of the four steps is optional. Amazon looks like one company but behaves like two: an enormous retail and logistics operation that runs at razor-thin margins, and Amazon Web Services, a cloud computing business with software-like economics that earns most of the profit. You can follow along in the live Amazon report on Tenet, which pulls the same figures used below.

The single most important idea in analyzing Amazon is to stop treating it as a retailer. The retail business exists to build scale, habit and a customer relationship. The profit engine is elsewhere. Get that split right and the rest of the analysis follows, because almost every mistake investors make with Amazon comes from valuing the whole thing as if it were one low-margin store rather than a retailer wrapped around a cloud and advertising business.

The retail-versus-AWS profit split

Amazon's profit comes overwhelmingly from AWS and, increasingly, advertising, not from selling goods. In fiscal 2025 total revenue was about $716.9 billion, and AWS contributed roughly $128.7 billion of that, under a fifth of the top line. Yet AWS supplies the majority of Amazon's operating income, because cloud computing carries margins that online retail cannot approach. Selling a book or shipping a package earns pennies; renting server capacity to enterprises earns dollars, the same software-like cloud economics we examine at its closest rival in how we analyze Microsoft.

This is why the pieces have to be valued separately. A thin-margin retail machine and a high-margin cloud franchise are different businesses with different multiples, bolted together under one ticker. The retail side gives Amazon scale and a favorable cash cycle, since it collects from customers before paying many suppliers. The AWS and advertising side gives it earnings. Advertising, sold against Amazon's own shopping traffic, has quietly become one of the most profitable parts of the whole company, built on the same query-and-intent economics we examine at the largest ad business in how we analyze Alphabet. The competitive edge of a business like AWS, embedded in a customer's operations, is the switching-cost dynamic covered in identifying competitive advantages (moats).

The retail moat is different in kind but real. Amazon's advantage in retail is not margin, which is thin by design; it is scale and speed. Its fulfillment and delivery network is something rivals would need many years and tens of billions of dollars to copy, and every additional package that flows through it lowers the cost of the next one. Prime, the paid membership that bundles fast shipping with video and other perks, locks customers into that network the way Costco's card locks in its members. The third-party marketplace adds another layer: millions of outside sellers pay Amazon fees to reach its customers, which turns competitors into a high-margin revenue stream. None of that shows up as fat retail margins, but it is a durable position all the same.

Why reported earnings understated the value

For most of Amazon's history, its reported profit was a poor guide to the value it was creating. This is the part traditional analysis got wrong for two decades. Amazon deliberately plowed nearly every dollar it earned back into warehouses, delivery networks, devices and the buildout of AWS. That reinvestment crushed reported net income, so on a simple price-to-earnings basis Amazon always looked absurdly expensive or unprofitable.

The spending, though, was building durable competitive assets: a logistics network rivals could not replicate and a cloud platform that would become hugely profitable. Low earnings were a choice, not a weakness, and they understated the intrinsic value being compounded underneath. This is the classic case for why cash flow and asset-building can matter more than accounting profit, explored in why cash flow is more important than profit. An investor who dismissed Amazon on its P/E for twenty years missed one of the great compounding stories, and the lesson is that reported earnings and economic value can diverge for a long time.

The free cash flow lens

Because reported earnings are noisy, free cash flow is the clearer lens for Amazon, but it comes with its own catch. Free cash flow is operating cash flow minus capital spending, and Amazon's capital spending moves in large cycles. In fiscal 2025, operating cash flow was strong at about $139.5 billion, but capital expenditure was enormous, around $131.8 billion as Amazon built data centers for AWS and artificial intelligence, leaving free cash flow of only about $7.7 billion.

That small free cash flow figure does not mean the business got worse. It means Amazon is in a heavy-investment phase again. When the spending cycle eases, free cash flow can rebound sharply, as it did after previous buildouts. The point is that you cannot read a single year's free cash flow and conclude much; you have to understand where Amazon sits in its capital-spending cycle, the mechanics of which are explained in understanding capital expenditures (CapEx) and what is free cash flow?. This is a business whose reported cash generation is deliberately understated in investment years.

Balance sheet and valuation discipline

Amazon's balance sheet is sound. At the end of fiscal 2025 it held about $123 billion in cash and short-term investments, against total debt near $153 billion, much of which is lease obligations tied to its warehouses and data centers. Given the size of its operating cash flow, the debt is comfortably serviceable, and Amazon has the capacity to fund its buildout without strain. This financial strength is what lets Amazon keep spending through a downturn while weaker rivals pull back, which is often when it extends its lead. A company that can invest counter-cyclically has an edge that never shows up in a single quarter's numbers.

Amazon does not pay a dividend and has bought back very little stock, because it still sees higher-return uses for its cash inside the business. For a company at this stage that is a defensible choice, but it does mean shareholders are betting entirely on reinvestment rather than direct returns of cash. Whether that reinvestment keeps earning good returns is the central judgment, and it is a capital-allocation question as much as a valuation one.

The harder call is price. In July 2026 Amazon shares traded around $244, giving a market value near $2.6 trillion and a trailing price-to-earnings multiple of about 34. On earnings that figure is high, but earnings understate the profit power of AWS and advertising, so the effective multiple on Amazon's cash-generating core is lower than 34 suggests. A common approach is to value AWS and advertising as the high-multiple growth businesses they are, value retail as the low-margin operation it is, and add the two, which usually produces a very different picture than the blended headline ratio. That is the whole difficulty: the headline multiple is misleading in both directions, and the valuation depends on assumptions about AWS growth and future free cash flow that each investor has to make. We keep the quality judgment and the price judgment separate, and never let one answer the other.

The long-term record: growth through a stumble

Amazon's track record over the past five years shows both its power and its volatility. Over the fiscal years from 2020 through 2025, revenue grew from about $386.1 billion to $716.9 billion. Net income is choppier: Amazon earned $21.3 billion in 2020, posted a $2.7 billion loss in 2022 as costs and a soured equity investment hit results, then recovered to $77.7 billion by 2025.

Figures are from Amazon's fiscal 2025 10-K and Tenet data as of July 2026. Past results do not predict future returns; the record shows what the business has done, not what it will do. The 2022 loss is a useful reminder that Amazon's reported earnings can swing hard on non-operating items, which is exactly why the free cash flow and segment lens matter more than any single year of net income. The longer arc of that transformation is told in Amazon's evolution.

Where to go from here

Amazon is a clear example of the Tenet lens in action: a two-part business where a thin-margin retailer builds scale and a high-margin cloud earns the profit, a company whose reported earnings understated its value for years, best judged on free cash flow and segment economics, and offered at a full price. To pressure-test the valuation, read whether to invest in Amazon, then open the live Amazon report on Tenet and check the current figures against the record above for yourself.