Today all of this stuff is pretty well resolved, but at a cost. Easy sharing and interactive collaboration over network is built on protocols like TLS that would be very slow even on the Quadra. The networks that runs over use an amount of compute that rivals supercomputers of the 90s (for one example just look at channel correction in MIMO radios). The disk footprint of all the fonts, languages, etc, is mitigated by storing them server-side, but this is all usable due to how fast networks are now. Machine translation means you can work across a language divide. Dictation is a win for accessibility and avoiding RSI. All of this is available on a Chromebook or iPad that is more reliable than the Quadra, runs on a battery, and is 50x cheaper.
We're not done, just as today spell check and grammar check are widely accepted as table stakes in another ten years features like built-in writing coach, editing, and research assistant features will likely be considered basic requirements with all the hardware needed to support that.
And that's just word processing.
Obviously, things that move boatloads of bits (mostly media) have well benefited from the vast amounts of RAM and incredible bandwidth of modern machines. And, of course, video games, which will always push limits.
But outside of that, word processing, data collecting, "printing checks from AP", those peaked a long time ago. And, of course, I appreciate how businesses collect vast amounts of data today (because they can, because storage is free), but core business functions, transaction processing, that's pretty well plateaued as well.