You may be right, I agree that it's bound to happen at some point. I think one big difference is Amazon is rolling out revenue-producing products and Google was rolling out non-revenue-producing "experiments."
The problem is the cost that comes with lack of focus. Exploding headcount. Unclear corporate vision. When only a few of your products make nearly all of your money, why spend a bunch of energy on a lot of things that lose money, or make practically no profit?
I'd love to see Amazon spend real effort on making AWS easier to use [for people who are not deeply steeped in AWS already, it's hard]. For the love of God, I can't believe AWS console still makes you globally switch on region for everything. Can't there just big a dashboard that lets you see and control, ya know, everything? It's things like that. Things that have been bad for a long time. Instead of pushing out redshift or cognitio or zocalo or workspaces or elastic transcoder, how about they make AWS as a whole a better product? Something easier to user. Something more consistent.
That's the beauty of the API. There are already various vendors that do just this on top of it. The funny thing is, if you're a big guy too, say 10K instances, you also need to use 3rd party tools since the AWS UI can't handle that case either. But since everything is an API call you can build your business by making AWS better for hire. Then again, better hope your idea isn't killed off overnight when AWS does add it as a new feature.