It's because it's the best codec we have for moderate-resolution video, and it already has almost %100 of the market (played back via Flash). Making it available in <video> would allow people to drop the Flash part.
Dropping Flash for video playback would be nice, yes.
But making the shift from plugin-based playback to HTML5 video is an opportunity. And it's really the only chance that the web as a whole is going to get - if Flash video dies and H.264 through <video> becomes the dominant format, then the web is going to be stuck with that, and the MPEG-LA will rake it in from anyone who wants to make a product that involves web video. That is a terrible position to be in.
The best option is to make a clean break here and now, and leave patent-encumbered video codecs behind at the same time as Flash for video playback is left behind. Playing through a Flash container is fine for legacy videos, and long-term there's no real downside to pushing unencumbered formats for HTML5.
While free is generally better than not free, the cost of licensing from MPEG-LA is relatively cheap. Terms may have changed last I looked, but it's something like twenty five cents per paid subscriber per year, the first 50,000 subscribers are free with a $1M cap. The same goes for sold encoders/decoders.
So companies like Google, Adobe, and Microsoft have to pay $1M per year at most. A smaller site that charges, like PluralSight, probably pays $0.
This really isn't about money. It's about control.
Last time I checked it was capped at roughly $5million per year and had been increasing by 10% (the maximum allowed) per year, every year since the licensing began (though in two-year steps, i.e. just over 20% each two years).
I have seen multiple people claim that it's only allowed to go up by 10% every five years. You obviously are claiming a different scheme again. Just the fact that there's so much confusion about price is in itself a mark against this method of creating (so-called) RAND standards.
