Tech has this weird, generational semi-imperialism, where a particular company seizes control of the platform everybody else needs and becomes "king" for a while, before fading into relative irrelevance when a new platform emerges. IBM and Microsoft both fit this pattern perfectly. Google and Facebook have arguably been contenders in more recent decades, except neither was really ever essential to developers.
Google search rankings have been crucial to businesses, and Facebook's got a somewhat frightening control over social interactions — and the business implications of that were enough to terrify Google executives into acts of pitiful desperation — but neither Google nor Facebook was ever actually essential to developers. They have both been arguably essential to businesses, but attempts to paint Google or Facebook as hegemonic tyrants in the 1980s/1990s Microsoft style don't really work, in my opinion, because while businesses do have an equivalent level of dependence on these platforms, developers don't.
So look at Amazon in that light. How many startups run on AWS? In 2015, AWS made Amazon almost as much as Amazon's retail operation did, and in 2016, Jeff Bezos expects it to hit $10B (about three times as much as Amazon retail).
Right now, being able to run all your infrastructure on Amazon is kind of awesome, although not without challenges. But if the last decade or two have disproved (or at least provided a counterexample to) this idea that tech's history consists mostly of cycles of platform domination, the 2020s might be a strong example in favor of the theory, with Amazon in control.