Google X / Moonshots seem like a viable model, though I can't comment on how well they've done. I think the economic incentives are too different for the likes bell labs to exist anymore. It's hard to justify basic R&D to investors. The days of Bell inventing transistors only to have intel benefit is over, research need to be justified and monetized now. Basic research at Bell / Xerox / IBM labs existed in a time where there was some sense of charity.