In brief

Amazon is several businesses sharing infrastructure, customers, data, and capital. Its stores and marketplace create selection and purchasing frequency; Prime can increase loyalty; fulfillment and delivery improve convenience; advertising monetizes seller and brand demand; and AWS sells computing infrastructure and services to outside organizations.

The combination is powerful but analytically difficult. Consolidated revenue can hide very different margins and reinvestment requirements. I would focus on segment operating income, cash generation after capital spending, fulfillment efficiency, AWS competitiveness, and whether newer services deepen the ecosystem without consuming capital indefinitely.

The businesses inside Amazon

Amazon’s filings describe revenue from product sales, third-party seller services, advertising, subscriptions, and AWS. The company also identifies intense competition, demand variability, inventory and fulfillment risk, data security, regulation, international exposure, and infrastructure investment as material risks. Amazon 2025 Form 10-K

It is useful to separate the engines:

Engine How it works What to watch
First-party retail Amazon buys inventory and sells it to customers Gross margin, inventory turns, markdowns, shipping and fulfillment costs
Marketplace Third-party sellers provide selection and pay service fees Seller health, fee pressure, regulation, counterfeit and quality controls
Prime Members pay for a bundle of shipping, media, and other benefits Retention, engagement, content cost, and benefit inflation
Advertising Brands and sellers pay for visibility near purchase decisions Ad load, relevance, measurement, seller economics
AWS Organizations rent computing, storage, databases, and other services Growth, margins, capital intensity, competition, and customer optimization

The flywheel—and its limits

Amazon’s retail logic is often described as a flywheel: more selection and convenience attract customers; customer traffic attracts sellers; seller participation expands selection and can improve asset efficiency; scale can lower unit costs; and lower prices or better service attract more customers.

Flywheels are not perpetual-motion machines. Delivery speed costs money. Marketplace fees can strain sellers. Dense logistics networks can improve unit economics, while excess capacity can destroy them. Regulators can challenge conduct that advantages Amazon’s own services. The framework is useful only when supported by current evidence about service quality, customer behavior, and cash flow.

Why AWS changes the analysis

AWS gives Amazon exposure to an enterprise infrastructure business with economics unlike retail. Customers use its computing, storage, database, analytics, security, and AI services. Scale, service breadth, technical expertise, and customer switching costs can support durable relationships.

Competition remains formidable. Microsoft and Google invest heavily, customers can use multiple clouds, and large buyers continually optimize workloads. AI can expand demand while requiring accelerated infrastructure investment. AWS growth should therefore be studied with operating income and capital spending, not as revenue alone.

Advertising is economically important

Amazon sits close to a purchase decision. A seller advertising within Amazon can connect spending to product discovery and sales more directly than many general media channels. This supports an attractive business, but excessive ad density could reduce the customer experience, and rising seller costs can eventually affect price or selection.

Advertising also illustrates why segment disclosure matters. The revenue can be high margin, but the ecosystem that produces the advertising opportunity depends on expensive retail, marketplace, and logistics infrastructure.

Risks I would not treat as footnotes

Capital allocation. Amazon has a history of entering new categories and investing ahead of demand. That can create valuable infrastructure or prolonged losses.

Retail margin sensitivity. Small changes in shipping, labor, fuel, inventory, and mix can matter in a high-volume business.

Cloud competition. AWS must keep pace in infrastructure, data, security, and AI while defending customer economics.

Regulatory exposure. Marketplace rules, labor practices, privacy, competition, and acquisitions face scrutiny in multiple jurisdictions.

Stock-based compensation and dilution. Cash-flow analysis should account for compensation paid with shares rather than treating it as economically free.

What I would monitor

I would read segment revenue and operating income together; compare operating cash flow with purchases of property and equipment; examine lease obligations, stock-based compensation, fulfillment commentary, paid-unit and seller-service trends, AWS growth and margin, and advertising expansion. I would also distinguish temporary efficiency improvements from a durable change in the cost structure.

Compare Amazon’s cloud economics with Microsoft and Alphabet, and its physical membership model with Costco.

Bottom line

Amazon’s advantage comes from several reinforcing systems rather than one product. That diversity creates optionality, but it also makes consolidated results easy to misread. A sound analysis separates retail, marketplace, Prime, advertising, and AWS; accounts for capital intensity and dilution; and values the stock only after deciding which economics are durable.

Sources