I've heard this use case come up in hybrid new/legacy products where there is a chasm between legacy scalability (probably a monolith in the compute or storage layer, or both). The "new" side needs to be able to self-regulate and if you can put that into the service-side it would it would be a helpful transition enabler (and, let's face it, given these transitions often never finish your product would be effectively sticky).

I don't think the cost of queues is a problem anywhere (I'm sure it is somewhere, but not a market's worth). The problems created by queues, on the other hand, are myriad and expensive.