Past all the considerations of content ownership and all, it seems to be quite the technical feat, the impact on page load is real (or maybe my testing was partial or biased).
I'd argue it is more of a social feat than a technical one. It is still HTML, after all.
The real underlying issue that AMP solves is ad bloat. Online newspapers tend to include that one more ad library that tracks a particular thing that one ad partner or one analytics team wishes to have, let alone visual bloat such as "share as", "like", "comment" and other social widgets all implemented as iframes with their own separate trackers, and the sum of it all has a large impact on both download sizes and performance, since they are seldom optimized. Oil on fire is the fact that there is generally no incentive to remove libraries that no longer have a use.
AMP forces newspapers and other content providers reliant on advertising to play nice. It gives them an incentive (through caching and better indexing) which costs them the bloat.
