Melissa Coates enumerates several reasons why you might want to use Azure Analysis Services:
Varying Levels of Peak Workloads
Let’s say during month-end close the reporting activity spikes much higher than the rest of a typical month. In this situation, it’s a shame to provision hardware that is underutilized a large percentage of the rest of the month. This type of scenario makes a scalable PaaS service more attractive than dedicated hardware. Do note that currently Azure SSAS scales compute, known as the QPU or Query Processing Unit level, along with max data size (which is different than some other Azure services which decouple those two).
Read on for more use cases.