We have gone back and forth several times in history.

It started with single computers, but they were so expensive nobody had them except labs. You wrote the program with your data, often toggling it in with switches.

From there we went to batch processing, then shared computers, then added a networking, with file sharing and RPC. Then the personal computer came and it was back to toggling your own programs, but soon we were running local apps, and now our computers are again mostly "smart terminals" (as opposed to dumb terminals), and the data is on shared computers again.

Sometimes we take data off the shared computer, but there is no perfect solution so distributed computing and since networks are mostly reliable nobody wants that anyway. What we do want is control of our data and that we don't get (mostly)