Several people, including Roland and Patrick, have pointed out that physicists from Boston University have fabricated nanomechanical switches which promise significant advances in data storage densities (to much greater than 100GB/in2).
The technology for these high storage densities is great but the problem is they are doing it with E-beam lithography which has never been an inexpensive manufacturing technology.
There is also the problem that the press release saying that current data storage devices are read at the kilohertz rate. My hard drives, which are 7+ year old IDE and SCSI drives, read data at megabytes/second rates so this sounds like the person responsible for the press release doesn't really understand how data storage technologies work. Can anyone figure out what they are talking about here?
Finally, if what they have are real "switches", then how close are we to building a rod-logic like computer that could capture Part 2 of the Feynman Grand Prize? Given the current detailed specifications one probably could produce 32 devices using E-beam lithography and capture the prize but one might have a situation like the X-prize where it cost more to do it than the prize award amount was.