AWS (the closest thing to software Amazon owns) will converge to 0 profitably as competitors copy the easily replicable service.
This is not to say there are some advantages in a few others (say, Rackspace appears to be simple to grasp, excellent human support) and GCE (load balancers don't need to be pre-warmed, etc).
But if anything, I see it increasingly harder for the other providers to keep parity with Amazon, because of their velocity and market share.
Now, for many people using cloud as just a easy way to get IP addresses, it's all the same, but once you start using a lot more of their services, they provide a lot of good stuff in what is basically "middleware" for applications to latch onto.
The question is whether an organization is doing this themselves (i.e. run their own Foo on their nodes) or is using the value added services or not.
What keeps on making money is the Amazon marketplace and their physical distribution infrastructure.
And if they make more money, the advertising parts of the competitors will make less.
And as people move away from 'raw cloud metal' to whatever the layer du jour is, what's underneath will matter less and less.
Of course, they were upset that she wouldn't respond to the funny sort of queries you could make of Siri. And my son really wishes her ability to do conversions extended to questions like "how many miles are in a light second" (Siri uses Wolfram Alpha to great effect here. Alexa - Amazon Echo - just apologizes and says she's added a bing search to the echo app.)