My appreciation for the STK deal starts with the network physics: "Never underestimate the bandwidth of a station wagon full of tapes hurtling down the highway." Frequently used in talks by Andrew Tanenbaum, it's also in his Computer Networks book. When you're dealing with bulk data transfer and multiple (required) locations, tape is good.
Storage is a hierarchy in time and space. It's not all disk or disk cache or optical; there's a spectrum of latency that fills the time axis. The space axis covers capacity as well as location. As long as natural disasters, mechanical wear, and government regulation remain real world factors, every enterprise needs to keep its data in more than one place, and usually in more than one medium. Tape is good.
Yes, customers in time-sensitive business like investment banking require completely synchronized redundant data centers, and they'll have disk to disk replication in place. You still want tape backups, to provide a longer-term data memory and possibly provide an insurance of last resort if there's corruption introduced into one of the primary sources. Data corrupts, and absolute data corruption corrupts absolutely, while thoroughly ruining your system administration staff's weekend.
This need for less speed (but more capacity at a good price) is going to affect the consumer market. Online backups of your digital pictures? How about your iTunes library? Latency matters less than certainty in this case. If I'm going to trust a third party with my personal bits, they had better be available to me for as long as I'm paying the bill. I've gotten to the point where various family records - financial, photographic, video, and audio - exist only on disk. I'm less than comfortable with this, even though I perform regular backups, because the backups are located in the same physical location as the primary media. Tape is good, offsite tape is better.