I don't think anyone's banking on Dart unseating JS, and it doesn't need to. It compiles to JS to work across modern web browsers while smoothing a fair deal for you in the way of optimizations and cross-browser issues. But that's saying nothing of the real reason folks choose Dart, or JS or any transcompiler or framework for that matter: the ways in which it may benefit you and your projects, relative to your other choices.
In Dart's case relative to JS, I think the language is more sane, debugging easier, the core libraries really powerful, the tooling more uniform, and the ecosystem more organized. I like the idea of one language on client and server, and yet the way it's done in JS-land with Node feels wilder and less approachable than I liked when I researched it some years back. That's just me, and admittedly I don't have all that much experience in the world of JS before I dove into Dart.
There are some tradeoffs of course, the biggest being that you sort of block yourself off from a world of JS libraries and even community to some extent. There's dart:js interop to wrap any JS library, and work being done to make that even easier with js:interop, but I think there'll always be some disconnect. Still, it's about costs versus benefits, and there are many benefits that I think ultimately outweigh the costs.
The point is there are a whole host of reasons to choose Dart, and I think it's foolish for any one of those reasons to be because you think it'll unseat JS. It won't, but there might be merit to it unseating JS in your own workflow.