The Maze: Amazon generated $716.9 billion in 2025. Direct online stores—the business most people still picture first—contributed $269.3 billion, or 37.6%. Everything else added up to 62.4%. That does not mean retail is shrinking: online-store revenue grew 9.0%. It means retail has become the front door to a machine that earns from sellers, ads, subscriptions, cloud infrastructure, physical stores, and other services. The interesting story is not diversification for its own sake. It is how Amazon monetizes the same pool of demand several times.
The store is still the largest line, but no longer the whole identity. Amazon's 2025 filing breaks revenue into seven groups. Online stores led at $269.3 billion. Third-party seller services followed at $172.2 billion, AWS at $128.7 billion, advertising at $68.6 billion, subscriptions at $49.6 billion, physical stores at $22.6 billion, and Other at $5.9 billion. The first three lines alone represented 79.6% of revenue. Yet only one of them was Amazon selling products directly. The rest show how far the company has moved from a single-margin retail model.
Marketplace scale hides an important accounting asymmetry. Online-store sales include product revenue recorded gross when Amazon owns and sells the inventory. Third-party seller services record Amazon's commissions, fulfillment fees, shipping fees, and related services—not the seller's full merchandise value. So 37.6% versus 24.0% is not a clean first-party-versus-third-party merchandise comparison. It is more revealing as a monetization comparison: Amazon earned $172.2 billion from enabling other merchants without booking the full value of their goods. The marketplace can be economically larger than the revenue line suggests.
The faster-growing engines sit around the transaction. Advertising grew 22.1% in 2025 and AWS grew 19.7%, while online stores grew 9.0%. Subscriptions rose 11.8% and third-party seller services 10.3%. Amazon's full-year results also put AWS operating income at $45.6 billion, against $29.6 billion for North America and $4.7 billion for International. Amazon does not disclose profit for ads, subscriptions, or seller services separately, so broad claims about every service being high-margin would be guesswork. But the direction is clear: several non-retail lines are compounding faster than owned retail.
One demand pool now supports several toll booths. A shopper can trigger product revenue, a seller commission, fulfillment fees, an ad auction, and a Prime renewal inside the same ecosystem. A business can then rent the computing infrastructure behind the stack through AWS. That is the strategic advantage. Amazon does not need each business to win customers independently. Retail supplies intent. Prime increases frequency. Marketplace sellers expand selection. Advertising sells visibility. Fulfillment sells logistics. AWS monetizes infrastructure far beyond commerce. The pieces reinforce the front door even when their accounting sits in different categories.
Why it matters: Amazon's moat is not simply scale in ecommerce. It is control of a high-intent environment that can charge at multiple layers: supply, discovery, loyalty, fulfillment, and infrastructure. Retailers usually earn once when an item sells. Amazon can earn before, during, and after the same transaction—and again from the technology that enables it. The operator lesson is not to copy seven businesses. It is to identify which customer interaction can support a second revenue model without weakening the first. Amazon's online store still matters enormously. Its bigger achievement is turning that store into distribution for everything else.
