Thought experiment: Divorce AMP from Google. Google withdraws from being the AMP standards author and "prototype" AMP cache provider. The project becomes truly non-commercial and is handed over to a non-profit that users trust, let's say, hypothetically, the Internet Archive. The IA adopts as AMP's goal: making web pages less expensive to crawl (ideally, by parties other than Google) as well as making pages faster on mobile. In addition, the AMP standard is revised to require that AMP pages must allow equal access by all clients, whether "browsers", "bots" or otherwise. No preferential treatment for certain browsers, e.g., Chrome, or certain search engines, e.g., Googlebot.
Bias disclosure: I use a text-only browser and AMP pages look great in links. For a links user, the AMP version can be useful on some sites that have a large amount of cruft, e.g., excessive number of same site URLs, at the top of the page, with the content buried below it, and yet more cruft at the bottom. AMP eliminates the necessary scrolling on such sites.