For instance, when people can print out custom handset chassis and form factors to go with easily swappable electronic internals. Custom input devices. The printable stuff will have to integrate with electronics (or print them anew) as to make it useful in order for it to outgrow the current market of prototyping enthusiasts.
I'd really like though to be able to print a paperback for reading and then tear it down in the machine when I'm done so the material can be fully reused. That's the ebook revolution for me.
Actually, I agree with you. When 3D gets fast to the point of instantaneous and cheap and with fully reusable material, we could have really interesting things. Imagine a 'physical' computer. I issue some command in bash, and a little physical representation of a file pops out from my workstation, and it has built in sensors and input so I can interact back with my workstation by interacting with it. I could have actual printed pseudo-ebook-readers tied up as tabs in my web browser.
There are many things that are impractical to print as a result of how slow 3D printers currently are. But if they were faster, then you can make things as you need it, you can make things where you don't know what you need to pack beforehand, and you can make disposable things on demand.
Same with computers. Many of the software we use regularly now were considered impractical with much slower computers with smaller storage (spam filters, any 3D graphics game). By making them faster, a larger class of things were available to be made more useful.
In the far far future (say 30-40 years), if 3D printers were fast enough with good enough materials, some startup can make a physical dropbox. Never use storage again. Digitize what you want to store, and recycle the object. When you need it again, print it out.
You can do that right now. Buy recycled paper, print a book, read it, and put it into the recycle bin. Just think of paper in the same way you think of ink.