SSAS is a service for data warehousing.

Pre-aggregation of measures along dimensions makes querying the data faster. Dimensional cubes instead of normalized tables make analyzing the data more efficient and flexible.

Yes, some data analysis is done from more normalized relational databases, but a lot of serious BI work is done in analytical cubes. SSAS happens to be Microsoft's version of that.

Ever worked with Excel pivot tables/charts? Analysis Services makes that kind of thing efficient even on some very large datasets. Allows BI analysts and/or managers to look at all kinds of slices of the data, often in ways nobody ever anticipated, and find useful information that way.

For example, one company I worked for, in a very competitive market, we did some serious analysis of our order processing line, using cubes generated from our CRM, our ERP system, our order-tracking system, and a few other systems just for the heck of it. Found ways to cut order processing time by 3 days (from 7 to 4). Gave us a huge competitive edge over the whole market. 25% increase in sales just because of that one factor.

Would have been possible to do this from the disparate data sources, but would have required a lot more time and effort to accomplish, and with less likelihood of noting a few key measures if we hadn't had the right SCDs and KPIs in those cubes.

Same cubes allowed us to cut a few significant overhead costs per-order, without anything but changing a few minor details of the workflow. Nothing obvious to even pros in efficiency, etc., but easy enough to find through the right SSAS cubes. Measurable increase in profitability per order. The owners invested that back into the business, to very good results.

That's why SSAS. More companies want that kind of result.