How AWS Became Amazon's Second Business Engine

The Internal Bottleneck

As Amazon grew, software teams needed common capabilities.

Every team required some combination of:

• storage;
• servers;
• databases;
• networking;
• authentication;
• deployment systems.

If each group rebuilt those capabilities independently, engineering became slower.

team wants to build customer feature
↓
team first builds infrastructure
↓
feature delayed

Standardization could reverse that order.

The Platform Insight

Amazon began treating infrastructure as reusable building blocks.

The deeper insight was:

These building blocks may be useful to companies outside Amazon too.

That transformed an internal engineering problem into a commercial opportunity.

S3 Made Storage a Service

Amazon S3 launched in March 2006. [S15]

The value proposition was simple but profound.

A developer did not need to buy and operate the physical storage system first.

Storage could be consumed through an internet service.

EC2 Made Compute a Service

EC2 added on-demand computing capacity.

The customer could provision servers in software rather than purchasing hardware and waiting for installation.

old model
buy machine
install machine
manage machine

cloud model
request compute
use compute
release / scale compute

That dramatically reduced infrastructure friction for many workloads.

Why Startups Loved the Model

A startup rarely knows future demand precisely.

Before cloud:

buy too little capacity
→ site fails if demand spikes

buy too much capacity
→ waste capital

Cloud made capacity more variable.

That reduced one of the largest early infrastructure barriers for software companies.

Why Enterprises Eventually Followed

Large companies had additional concerns:

• security;
• compliance;
• reliability;
• migration;
• procurement;
• legacy systems.

AWS gradually added enterprise capabilities across these areas.

The market therefore expanded from developer utility to enterprise platform.

AWS Became a Portfolio, Not a Product

The early platform centered on storage and compute.

Modern AWS spans:

compute
storage
networking
databases
analytics
security
developer tools
machine learning
generative AI
custom silicon

This makes the platform harder to compare with a single software product.

The Business Model Was Different From Retail

Retail requires merchandise throughput.

AWS requires infrastructure utilization.

The cost structures are different.

This matters because AWS can generate very different margins from Amazon's commerce businesses.

By Q2 2026 AWS generated approximately $16.6B of operating income on $42.2B of quarterly revenue. [S2]

AWS Changed How Investors and Competitors Saw Amazon

Before AWS became financially visible, Amazon could be described primarily through retail economics.

AWS revealed that Amazon had built a large technology-platform business inside the same company.

This changed the company's competitive set.

Amazon now competed simultaneously with:

Walmart / broadline retailers
AND
Microsoft / Google / enterprise technology providers

Andy Jassy's Role

Andy Jassy joined Amazon in 1997 and later founded and led AWS from inception. [S5]

He served as AWS CEO from 2016 until becoming Amazon CEO in 2021. [S5]

This makes AWS central to understanding his management background.

He spent years learning how to allocate capital into infrastructure before demand was fully visible.

That experience now shapes his Amazon-wide role during the AI infrastructure cycle.

The AI Cycle

Generative AI makes cloud infrastructure even more capital intensive.

The stack now includes:

power
↓
data centers
↓
chips
↓
networking
↓
training
↓
inference
↓
foundation models
↓
agents / applications

AWS can monetize multiple layers.

But doing so requires investment before future market share is known.

That recreates an old AWS problem at a larger scale.

Q2 2026 Reacceleration

AWS Q2 2026 revenue grew 37% year over year to approximately $42.2B. [S2]

Amazon described this as its fastest growth in 18 quarters. [S2]

The reacceleration matters because it occurred during heavy AI infrastructure investment.

It suggests demand was growing quickly enough to absorb at least part of the new capacity.

It does not guarantee the return on all future spending.

Why AWS Is Amazon's “Second Engine”

Retail and marketplace generate enormous revenue.

AWS contributes a disproportionate share of profit.

It also gives Amazon strategic relevance in enterprise technology.

The company therefore has two very different foundational systems:

COMMERCE ENGINE
customers + sellers + Prime + fulfillment + ads

CLOUD ENGINE
AWS infrastructure + developers + enterprises + AI

Neither fully explains Amazon alone.

Together they explain the modern company much better.