Stage 1 — Solve a Selection Constraint
Amazon's starting point was books.
Books were unusually well suited to early internet retail because the number of titles was enormous.
A physical bookstore had to choose which books deserved shelf space.
Amazon could display a much larger digital catalog.
The 1995 launch announcement explicitly argued that a physical bookstore carrying Amazon's selection would be economically impossible. [S8]
The first Amazon advantage was therefore not “cheap books.”
It was:
internet catalog
+
aggregated demand
+
centralized fulfillment
=
selection beyond local shelf-space limitsThat insight shaped the company.
Stage 2 — Reuse the Commerce Infrastructure
Once Amazon had built:
- customer accounts;
- payments;
- catalog software;
- search;
- recommendations;
- order management;
- fulfillment;
- customer service;
it became easier to add new product categories.
The marginal challenge of adding music or electronics was smaller than building a new company from zero.
This is the first example of Amazon's reusable-infrastructure pattern.
book infrastructure
↓
commerce infrastructure
↓
multi-category retailStage 3 — Stop Owning All the Selection
A retailer can only stock products it is willing to finance and manage.
Amazon's marketplace changed that constraint.
Independent merchants could bring their own inventory.
Amazon could provide:
- traffic;
- checkout;
- trust systems;
- payments;
- discovery;
- later, fulfillment and advertising.
The company's role changed.
Amazon as merchant
↓
Amazon as merchant + platformThis was one of the most important structural transitions in its history.
Stage 4 — Turn Shipping Into Membership
Prime launched in 2005 with unlimited two-day shipping for a $79 annual fee. [S11]
The product changed the customer's unit of decision.
Before:
Is this one order worth the shipping charge?After:
I already joined Prime; should I order this from Amazon?That shift encouraged Amazon to improve the physical network because every improvement increased the value of the membership.
Stage 5 — Turn Fulfillment Into a Seller Product
Amazon had already built warehouses for its own inventory.
In 2006 it launched FBA and opened much of that infrastructure to sellers. [S12]
That was a second major platform transition.
internal capability
↓
external serviceFBA made Marketplace stronger because independent inventory could now use Amazon-style fulfillment.
It also made Prime stronger because more third-party offers could fit fast-delivery expectations.
Stage 6 — Turn Computing Infrastructure Into AWS
The same abstract pattern appeared in a different domain.
Amazon had struggled with internal technology infrastructure. [S14]
Rather than keep every capability private, it exposed standardized infrastructure to outside developers.
AWS launched in 2006.
internal computing capability
↓
external cloud services
↓
independent developer ecosystemThis was more radical than FBA because the customer no longer needed any retail relationship with Amazon.
Stage 7 — Turn Traffic Into Advertising
Marketplace created more products and more competition for visibility.
Amazon's shopping traffic also contained strong purchase intent.
Those conditions created a large advertising opportunity.
shopper intent
+
many competing products
↓
advertising demandAmazon Ads later expanded beyond shopping pages into Prime Video and live sports.
The same customer attention could now generate:
transaction revenue
+
seller-service revenue
+
advertising revenueStage 8 — Add Attention and Membership Depth
Prime Video gave Amazon a consumer relationship that was not tied to buying a physical product.
MGM added owned entertainment IP.
Alexa added a voice and smart-home interface.
Ring added home-security devices.
These businesses broadened the company's surface area inside the household.
Stage 9 — Add New Infrastructure Frontiers
Amazon's 2020s infrastructure bets include:
- AI data centers;
- custom silicon;
- robotics;
- Amazon Leo satellite connectivity.
The company continues to apply a familiar pattern:
build capability at large scale
↓
standardize
↓
turn access into a product or serviceNot every new bet will necessarily work.
The pattern is still visible.
The Company Did Not Become “Randomly Diversified”
Some Amazon businesses clearly reinforce one another.
Marketplace, FBA, Prime and Ads are tightly connected.
Other businesses are more independent.
AWS does not need a retail transaction.
Amazon Leo may develop its own customer base.
Healthcare has different regulation and economics.
A useful company model therefore has two categories.
Reinforcing ecosystem layers
Marketplace
Prime
FBA
Ads
Prime Video
Alexa for ShoppingAdjacent / independent infrastructure businesses
AWS
Amazon Leo
HealthcareThis avoids forcing every Amazon activity into one simplistic “flywheel.”
Why the Story Matters
The defining transition was:
sell products
↓
own customer demand
↓
open demand to sellers
↓
open logistics to sellers
↓
open computing to developers
↓
open media inventory to advertisersAmazon repeatedly found ways to turn a system built for itself into infrastructure other people could pay to use.
That is the deeper company story.
Common Misconceptions
“Amazon became successful by selling more product categories.”
Category expansion mattered, but the platform transitions mattered more.
“AWS was a random diversification project.”
AWS emerged from Amazon's internal infrastructure experience. [S14]
“Prime is just a loyalty program.”
Prime changed the economics and strategic importance of fulfillment.
“Marketplace reduced Amazon's need for infrastructure.”
It reduced the need to own every product, but FBA and fast delivery increased the need for fulfillment infrastructure.