Don't be silly, all research done by companies has an ROI counter tied to it . . . the point is that some do much more forward thinking research into technologies that might not yet have a specific application quite yet, and others are focused 2-3 years in advance and that's it.
This is simply not true except perhaps in that any research center will have an aggregate ROI counter that some bureaucrats glance at annually. As somebody pointed out elsewhere, Bell Labs' ROI was capped. Much, and I would argue the best, corporate research is done with almost no consequential oversight from the host corporation. This was certainly the case with Bell Labs, XEROX Parc, and other great research centers of yore just as it is still the case to some extent at Microsoft, AT&T, IBM, and Intel today (between re-orgs).
At present, PARC (no longer under Xerox) operates on something of an agency model and Google runs research projects like startups. This appears to result in a higher rate of short-term commercial success, but at the cost of fundamental research. That said, I think we've done a tremendous amount of fundamental research since World War II and now that the conditions for producing more no longer exist, there's plenty of low-hanging fruit – new configurations of things that already exist – hence the prevalence of startups and the perceived excellence of hybrid research models.
Bell Labs was not setup like that, though. They were researching new materials to make their existing Monopoly run more efficiently. If ROI was the focus, IP would not have entered the public domain & people like Shannon & Shockley would have been shown the door before they changed the world.