I think the limiting factor is hardware.

Software video decoding is slow and causes devices' fans to spin up, which is why people use workarounds (such as visiting YouTube in Safari) to force hardware-accelerated H.264 decoding instead of using software VP9 decoding, despite VP9 using less bandwidth.

By the time VP9 and H.265 hardware video decoding is cheap and ubiquitous, I suspect the rest of the factors will fall into place.

The last TV buying guide I read was still suggesting 720p TVs at certain screen sizes ("it's a better value"), with 1080p for the remainder. But 1080p is starting to see wider adoption; even a Raspberry Pi can do hardware H.264 decode/encode these days.