Microsoft Research has a great stable of researchers -- could you imagine just casually having Leslie Lamport being an email away while working on some distributed systems problem/design? Or being able to pull on
This might be one of the only benefits to monopolistic presences -- the amount of great minds in the same place, set free to mingle and create can produce amazing results when a benevolent sponsor exists and it's tunnel-visioned on profit.
I'm quite cynical towards Microsoft but I am very grateful that they publish at least some of their papers/insights and their links (to existing papers) generally don't die.