Text things are extremely data light. All of wikipedia's text is smaller than a 2k movie. There is absolutely data stuff happening in the background on the server that makes it more complicated, but the actual piped data that goes from a google search result is actually quite small (though larger than it used to be).

Video is at the other end. There's only so many things you can do to a video to reduce the amount of stuff you send to the user (and a lot of the things you do put more strain on the user's computer to interpret what you've sent).

Music, singular images, video game data, and mass data tend to be somewhere in the middle, though context does matter for each of them.

Comparatively, sending videos and storing videos for later use is many times a more resource-expensive task than sending an image, forum post, email, weather updates etc.

It doesn't have to be ads before videos, but it does have to be something (subscription services, the page itself being littered with ads, state backing etc).