There is only one good reason to get a PhD: because you want to further our understanding of the field.

So do you want to further our understanding of the field, or do you want to build cool things and make lots of money?

I'm not saying you can't build cool things and make lots of money with a PhD, but I AM saying it's probably not the optimal path for someone with those goals to take. Your lifetime earnings will likely be higher if you just do ML engineering rather than ML research, and you'd spend more time building useful things that way too.