I think more failures than successes can be interpreted as a healthy appetite for product experimentation. The other option in todays world would be not to innovate except through purchases of established product families from other companies and acquisitions.

Amazon phone, lol.

So yes the phone failed, but how about the 50 services they offer on AWS? Failures on their part are heavily publicized, but successes move along silently.

Or AWS itself. Doesn't seem like a good fit for a company like Amazon.