I don't know that there is a specific name for that "model", but it's a reasonably common approach. In addition to Microsoft (your example), AT&T*, HP, IBM, and quite a few others follow the same approach. Even Dell and others are starting to embrace this approach.
The reasons should be fairly obvious.
R&D activities need to be run in a different manner than a typical profit / loss model. R&D is explicitly for activities that may never pan out. The corporation may push the R&D towards more likely to be practical / applicable efforts, but there isn't a guarantee that a product will result. As an example, IBM used to heavily fund basic research. As of the mid-2000's, researchers were encouraged to focus on aspects that were more likely to be applicable to market needs.
Services divisions are a typical example of a normal profit / loss model. A Services division can measure utilization rates, profit margins (or losses), service backlogs, etc... All of which are in alignment with selling a product (a service).
*I'm not certain about AT&T having a separate R&D division anymore, but they certainly used to.