The recent story of Dropbox leaving AWS brought about a wide mix of reactions from my friends. Some thought it was only logical that Dropbox needed to get more control of its solution stack while others saw the change as a waste of time and money. What is the real right answer?
Most disruptive technologies seem to exist in hyperbolic bi-polar cycles of emotion. Markets either want to believe that a new technology is irrelevant or a complete world domination. Companies delivery new technology are either DOA or the next “Facebook.” The problem is this; most of the time even the most disruptive events aren’t binary. For example, mobile devices clearly have had an impact on the PC market but they didn’t eliminate it. Things have just settled into a new normal.
Let’s look at Public Cloud and AWS. When I was at EMC back in 2006 and saw what AWS was delivering, I was concerned. If they could get people to adopt their approach to infrastructure-as-a service, they would be able to deliver a flexible product at a great price. This, to me, was a big deal. At the time, some folks understood the potential but most chose to believe is was “just a toy” and could only be used for a few applications and would be only adopted by startups that weren’t “real” businesses. These folks were proved very, very wrong.
The situation 10 years later in 2016 is just the reverse. AWS has run away with the show and taken a big lead in public cloud technology. Now, all I hear in the VC community is that they are unstoppable and the no individual company will have any IT infrastructure in 10 years. The Dropbox article was disturbing to many because it went against the current hyperbolic thinking that all private infrastructure is dead.
When it comes to clouds, the fact is that we are in the early stages of a not just a long game but a long season. AWS has made Public Clouds a major market segment that is here to stay but it will not be the only way IT infrastructure is delivered. I am pretty sure that companies like Facebook and Google will continue to build their own infrastructure stacks because it gives them competitive advantages. While I acknowledge that Private Clouds are not competitive today, that is due in large part because they are mostly built with legacy solutions marketed with a “private cloud” label. As real cloud technology moves into private environments, the game will change – again.
In my next blog I will talk about the critical factors that I think should drive a rent, build, buy infrastructure decision.