Twenty years ago, my University used 2 CDC supercomputers to support most all of the research needs of Thirty Thousand students and Thousands of academics.
Yep, so did mine, thirty years ago. And the closest I could get to 'em was the window through which I passes the card deck... Today, I have three computers sitting on my desk, and I use them all in different ways -- but I use them directly.
Twenty years ago, engineers wrote proposals and reports longhand, and made rough sketches and graphs; secretaries typed them; draftsmen and illustrators did pen-and-ink renderings of the graphics. The engineers proofed these and redlined them, and the corrections were often done directly on the originals. Design work was mostly hand-work, with lots of extrapolation and interpolation of graphical data; the few computer runs were expensive in both time and dollars, when they were done at all.
Today, the engineers write their proposals and reports on their desktop computers; edit them there; produce the graphics there and refine it themselves; assemble the graphics and text into a final document; and generally print it out themselves, unless they distribute it electronically -- which they also do themselves. And the bar has been raised for the final product: corrections-by-hand aren't acceptable, and the graphics really need color. The engineering itself involves multiple iterations, with much (even most) of the detailed design actually being done on computer models instead of physical prototypes. And the engineer does most of this work directly, too, unless they truly need a supercomputer run.
It's gotten to where I can work as a single individual and replace an office full of support staff -- which is exactly what I do. Is that wasteful of computers? I don't think so, even though one of my computers basically acts as a file, fax and printer server for one person. It's worth what it costs me to use it for nothing else... it keeps the load of my workstation, and saves me the few minutes a day that it cost me over a few months to buy.
The other point I want to make is this: the analysts who say productivity isn't going up with computer use, are are missing the points I've made above -- particularly the one about the bar being raised.