Selective Cloud Usage
My thought is that persistence of large volumes of data in the cloud is more expensive than on premise, especially if the growth in data is consistent, that is you know in general how your data is growing. With many large organizations having their own on premise clouds, or fabrics, offering the same elasticity - this is even more valid. I also realize that processing, that is ETL/ELT plays into the picture as well.
Where I see a strong advantage for cloud is in serving up summary data, for spinning up quick data marts, and in cases where the volume is small and change is uncertain.
The idea of a logical data warehouse really comes into play when you have a mixed persistence solution. If all your data warehouse data is in one place, be it an old school RDBMS, or RDBMS in the cloud, then the extisting tools around metadata and data governance should be fine.
At some point, we'll all drive electric cars, and all computing will be in the cloud, but we're not there yet, and we need to be judicious about our resources, and our approach in adopting technology.
It seems to me, our technology experts are marketers as much as well, anything else. We need more informed articles that help us data and enterprise architects drive out the right decisions.