The Origins of Amazon's Cloud Computing

The widely repeated version of how Amazon built cloud computing is that EC2 came out of a plan to monetize spare holiday capacity. That is internet apocrypha. The actual story is Chris Pinkham wanting to keep working after he moved home to South Africa, Christopher Brown joining him, and the two of them building a virtualized server platform that the rest of us in operations did not understand at first.

I was running availability at Amazon when this was happening, and I gave Chris and Chris a hard time about it. The exact quote I gave Stacey is the one I still hear back from people: I was horrified at the thought of the dirty, public internet touching my beautiful operations. I meant it. I had spent years building operational discipline at Amazon, and the idea of letting the open internet run untrusted code inside that environment felt like a category error. So Pinkham and Brown went off and built EC2 in a different data center, away from my domain. Werner Vogels was the executive sponsor who created the space for them to do it.

What I had wrong was the lens. I was reading EC2 as an operations problem. It was a platform problem. The right question was not "is this safe inside our perimeter" but "what becomes possible if every developer in the world can spin up a server in ninety seconds." Once that question is on the table, the operational concerns are real but tractable, and the upside is the rest of the modern internet.

The pattern matters more than the anecdote. A company creates space for a small team. The team builds the thing the rest of the org cannot yet see. The operators with the most invested in the status quo resist, sometimes loudly. The build happens anyway. Years later, the operators are happy to have been wrong, because the platform they resisted is the one their work now runs on. That is the shape of every important infrastructure shift I have lived through.