(I have a list of six predictions about what I expect to see happen in academic library technology over the next 5-10 years. This is prediction #1.)
Libraries, publishers and vendors will continue to push more content, tools and services onto the cloud.
I think the content part of this prediction is reasonably uncontroversial. (But I have been wrong before.) As more content goes digital, more content will go nonlocal, because really, who has the server space or wants to deal with the rights complications? If you're the Ontario Council of University Libraries, you might be able to negotiate for "local load" of monographs (see slide 4)--but for all but one of the member libraries, and maybe just plain all of them, that content is still going to be living on one or more external servers.
As the content goes nonlocal, the tools will follow even more than they already have. If I go to netLibrary through my library or Google Books on the open web, I don't have to download any software. I've given up on using e-books or e-audiobooks from my local public library precisely because there's software to download, and it is a royal pain. Online journal databases might require me to download the latest version of Adobe Reader, but that's such a common tool that I won't even hyperlink it. Meanwhile, my library's catalog is hosted remotely, as was its predecessor. I don't know much about WorldCat Local, but it seems to be trying to move the "get it" function onto the cloud along with the "find it" function. In general, it seems to me that "the thing" and "the way you find and use the thing" are going to increasingly live in the same place.
All that's left is services. Services will follow the tools. Services are part of finding and using "the thing." Technical services folks seem okay with this concept. Public services folks, less so, though I could be listening to the wrong conversations. I'm not even talking about in-person vs. online or SMS (text) reference, or staffed vs. automated circulation. I'm saying that if in-person reference involves two people sitting in front of a computer using networked resources, that's a networked service, and the in-person aspect will increasingly become a boutique experience, as with, say, travel agents. If circulation is fundamentally a series of barcodes or RFID tags interacting with a networked ILS, that's a networked service. Contactless smart cards + robot book fetchers + library RFID tags = probably nothing any time soon, but just the idea of it will inspire various smaller changes in various academic libraries.
I still remember the day I was meeting with our director of distance learning, and we were discussing our electronic reserves system and whether it had any advantages over how they were using the content management system, and I blurted out, "Of course, now that IT has moved all the students to gmail accounts, really we should just do all this with Google Docs and Google Groups." I still think we should try it. It's only a matter of time before patrons Google Chat with reference librarians who then "check out" identified materials to them by sending a "share document" notification. Maybe a "check this item out" link in the local catalog will automatically generate the same kind of notification, and "return this item" will delete their identification from the document sharing list. For all I know, this kind of thing is already happening.
Obviously, there are serious digital divide questions here. For this reason, I don't see this trend as either total or super-fast. But 5-10 years? I really think there will be a lot more of it. (For one look at the same trend in higher ed in general, see the Tenured Radical.)