How did an online store become the world's biggest renter of computers?

Amazon first built computing infrastructure to run its own business, then discovered that other people would pay to use it. Inside the company, engineering organizations were spending 70% of their time on what they called "undifferentiated heavy-lifting", such as IT and infrastructure problems, so Amazon created a shared IT platform. In July 2002 it opened parts of the platform to outside developers, and over a hundred applications were built on it by 2004. The interest took Amazon by surprise and convinced it that developers were "hungry for more".

The decisive idea came on paper. In late 2003 Chris Pinkham and Benjamin Black presented an internal paper on a standardized, automated computing infrastructure for Amazon's retail business. Near the end, they mentioned the possibility of selling access to virtual servers as a service. Pinkham, Willem van Biljon and developer Christopher Brown built what became Amazon EC2 with a team in Cape Town, South Africa. It launched in August 2006, a few months after the S3 storage service.

EC2 changed the deal. Amazon's announcement promised to cut the time to obtain and boot new server instances to minutes, and to let customers pay only for the capacity they actually use. That is the heart of cloud computing: computing rented on demand, scaled up and down as needed, and billed by use, instead of bought and installed in advance.

It worked. In November 2010 all of Amazon.com's own retail sites were reported to have migrated to AWS, and by the end of 2020 its revenue had grown to $46 billion. As of early 2023 AWS held 31% of the cloud infrastructure market, ahead of Microsoft Azure at 25% and Google Cloud at 11%.