Microsoft spends a significant amount on basic research, like Bell labs used to and Microsoft research publishes a lot of publications. 15% of the research budget goes to universities for research, that could keep the profits lower than Apple.
I've always felt they spent so much on "research", in part, as a way to reduce their reported profits. Because Microsoft was notorious for mostly just copying what others did first (see Apple --- and yes I know about Xerox PARC), or acquiring/rebranding/integrating (Spyglass -> IE, etc.), and only a small percentage of so-called original/in-house stuff. When the original version of C# and .NET came out it was so hilariously obvious what they were ripping off at that time (cough-Java-cough), they didn't quite shave off all the boilerplate for those in the know. But They Capitalized Differently. Oooooh. ;)
On a related note, I once knew someone who officially worked in software R&D. And I noted they mostly spent a lot of time surfing the web and then parroting some sexy sounding thing they read about recently. Ironically (or not?) I think that's a not un-legitimate way, and certainly can be an effective way, to introduce new thinking inside a particular company, or introduce a new technology, like say as a pilot project. But to describe this process as requiring heavy spending (on software R&D mind you, not talking about hardware stuff) sounds ridiculous and an unnecessary money sink. Find a smart creative engineer. Take them off the leash. See where they go, what shiny nuggets they bring back. That's how you do practical software R&D, outside of academia. Doesn't require billions. Just brains, and it can be a side product of what your staff is already doing anyway -- if you have the right staff.