Amazon Web Services
In 2003, Amazon was an e-commerce site growing rapidly. The servers struggled to keep up with the pace of orders, and every activity spike revealed the limitations of an infrastructure hastily built in 1994. Jeff Bezos and his teams sought ways to handle the load without endlessly multiplying hardware investments.
The company had launched a service meant to enable retailers like Marks & Spencer to build their own stores on the Amazon platform. Except no one had anticipated the architectural chaos awaiting developers. Code had piled up over the years without any real master plan. Extracting the software components cleanly to offer them to third parties was impossible. Everything had to be rebuilt from the ground up.
Amazon therefore began breaking down its infrastructure into distinct services, each accessible through well-documented programming interfaces. This reorganization made the service functional, but it had an unexpected side effect: internal teams began working differently. Each service became autonomous, reusable, standardized. A new discipline took root in software development.
Meanwhile, Andy Jassy, Bezos’s chief of staff, noticed a troubling paradox. Amazon was hiring engineers extensively, yet development speed stagnated. Digging deeper, he discovered that teams were losing three months setting up their technical environment before starting to code. Database, compute servers, storage: everyone was rebuilding the same foundations on their own. A tremendous waste of time and energy.
In summer 2003, Bezos gathered his executives at his home for a strategic retreat. What began as a routine session on Amazon’s core competencies stretched well beyond the planned thirty minutes. Beyond e-commerce and logistics, one thing became clear: Amazon knew how to manage complex IT infrastructures. The constraint of thin margins had forced the company to optimize its data centers to excellence. This expertise was invaluable.
The idea then emerged to transform this infrastructure into an operating system for the Internet. All the components already existed at Amazon, but no one else had access to them. Developers worldwide were reinventing the same building blocks, losing the same three months, hitting the same problems. Amazon could sell them these ready-made components.
The vision took time to materialize. Amazon Simple Queue Service launched in 2006, closely followed by Simple Storage Service and Elastic Compute Cloud. These three services formed the foundation of AWS. Developers could now rent computing power, storage, message queues, without buying a single server. Everything billed based on usage, like water or electricity.
The business model disrupted established practices. Gone were the massive investments in machine rooms sitting idle half the time. Resources adjusted to demand, scaled up during peaks, scaled down during lulls. Technical complexity disappeared behind simple interfaces. Companies could finally focus on their business rather than managing servers.
AWS entered a virgin market. Surprisingly, competitors took years to react. Microsoft, IBM, Google watched the train pass before launching their own efforts. Too late: AWS had set the standards for cloud computing and captured over 30% of the market by 2016.
Revenue climbed to 10 billion dollars annually. From startups to giants like Netflix or Dropbox, everyone adopted Amazon services. An obsession with operational efficiency partly explained this success. Amazon built massive data centers in strategic locations, where electricity and cooling were cheap. Economies of scale allowed it to regularly lower prices while remaining profitable.
AWS marked the transition from ownership computing to consumption computing. Like public utilities during the industrial revolution, the cloud became a commodity available on demand.