[microarchitect]

  These companies can afford to set up big labs with lots of PhDs and 
  pay them to do whatever the hell they want with little accountability

As an anonymous commenter has already pointed out, with the possible exception of some parts of MSR, there aren't any industrial labs where you can "do whatever the hell you want". I was offered a position at IBM research and they were very clear from the outset that they had a specific set of projects they wanted me to work on. In fact, these projects had less to do with my research work at grad school and more to do with what I'd done in my previous life as a software developer. These sense I got from talking to the researchers at IBM was there is a clear expectation that IBM research would directly impact the company's bottomline. The group I was to be recruited into was helping solve some of the hard problems that the rest of the organization simply doesn't have the expertise to solve.

I dunno about other fields but in my area of work, which is computer architecture, MSR, Intel and IBM have all written important papers that have directly influenced the design of many current-generation microprocessors. So, I'm a bit hesitant to suggest that the labs have failed or that the model is broken. The model could do with some improvement for sure, but then again, what couldn't?