I agree that MSR is probably the best CS industrial lab out there right now, but I think Moshe's point is that even the rest of the MSR is probably a bit wary right now given what (and how it) happened to MSR SV. I don't think HP, IBM, and Bell Labs (with a few specific and localized exceptions) are doing basic research as they were doing a few years ago (or in Bell Lab's case, a few decades ago). Maybe you are right about Google, but generally speaking, there are very few papers from Google Research (relative to their strength).