This line of reasoning can be applied to ANY change.
Also HTTP isn't going away.
Why isn't HTTP going away?
Because there's a lot of it.
Would it go away if there was less of it?
Can you imagine ways there could be less of it?
If there was something dramatically better for every use case (which is not IPFS) it would slowly displace HTTP in new deployments and there would still be HTTP for a very, very long time. The Web we have, in other words, isn't going anywhere.
We've had the web for twenty years now, and are just beginning to understand what it is and how it changes the world. If you're suggesting using IPFS (or any other system) as a way to expand and grow the web, I'm with you. But replacing the web means throwing out a lot of hard-won experience and intuition, in favor of a system that's bound to have its own issues down the line.