Was cloud computing a fad? The verdict

So, about that fad.

Amazon spent the early 2000s doing something less exciting than launching a startup: it rebuilt its own internal infrastructure into standardized, automated, on-demand computing so its retail business could move faster. In 2006, it started renting that infrastructure to outsiders — storage first, then compute. This time, the balance sheet behind the “fad” belonged to one of the largest retailers on earth, not a two-year-old company that had just spent $5 million on office furniture.

So, was it a fad?

Technically, yes — plenty of respected analysts called it one, in print, with specific dollar figures attached to prove it. Gartner's $25.3 billion ASP forecast for 2004 wasn't wrong because the idea was bad; it was wrong because it was pricing a twenty-five-year infrastructure buildout on a two-year hype cycle. The underlying pitch — rent the computing, skip the ownership — never actually changed between 1999 and 2006. What changed was the balance sheet behind it, the broadband under it, and who was patient enough to keep building through the part where everyone else called it done.