For Microsoft, at least, take a look at work done by Leslie Lamport and Simon Peyton Jones while working for Microsoft Research. Look at the papers they've publ

For Microsoft, at least, take a look at work done by Leslie Lamport and Simon Peyton Jones while working for Microsoft Research. Look at the papers they've published, and also what they work on - things like TLA+ for Lamport and Haskell for SPJ. I'd argue that both of these guys do interesting and important CS work that doesn't have an immediate commercial payoff for Microsoft. They're far from the only ones doing good CS research at MS; they're just the first that came to mind.