It’s not impossible to switch engines on new projects lots of devs have stated this. Devs have switch engines for far less or made their own.
It depends on a lot of factors though. Creating your own engine is by far not an easy task. The more feature rich it shall become, the more work it will need. Especially if it should have high 3D graphics quality while also running performant. That alone can cost a good team at least 2 to 5 years.
Switching engines also depends on how portable your work from the old engine is with regard to the new engine. It may not be impossible but can still be a lot of work. The earlier that decision is made, the better.
If the devs are determined enough they can surely do a switch. But they might sweat a lot. And especially for smaller studios, or studios without sufficient funding, this quickly becomes a matter of financial survival.
So it’s not impossible, yes. But don’t take that lightly as well.
Switching engines also depends on how portable your work from the old engine is with regard to the new engine. It may not be impossible but can still be a lot of work. The earlier that decision is made, the better.
Not to mention I’m guessing a good amount of indie devs are not abstracting every detail of interacting with the engine from the getgo in the chance they want to swap engines down the line. I’m sure some more experienced studios due for that just incase measure or to make migrating past breaking changes a bit easier when they crop up. But generally speaking I can’t imagine that’s a common tactic. But even if it did your still going to have to recreate every new implementation for your interfaces and there are bound to be differences here that are gonna take some time.