To replace GIFs. I don't have a problem with users being able to opt out, etc. But GIF's are still useful, yet a huge drain on resources. If for nothing, I want my cat videos still available inline without having to click them.
Hardware video decoders mostly only handle one video at a time and if you could multiplex the one decoder the chip has, to keep up with multiple videos playing at once it just might take a milliwatt or two.
Because gifs already work that way on phones, and that's how users want it.
I never wanted any movable ads while trying to read something on the phone. Even less the movable the cat gifs. If something is not worth the click, it's certainly not worth to be inline in the text.