You had bank switching. So for example the MP/M system I used (Altos) would allow several terminals (serial) of which each would get a command processor in a bank. I think it had 256KB, maybe 512. You know, I never ever remember thing thing crashing, even with multiple terminals/sessions.

Multiuser machines could get away with surprisingly little memory; They had some major advantages over home computers:

Users were all connected via CRT terminals over serial, so you didn't need to waste any memory on screen buffers. This also reduced CPU usage as the user wouldn't get any CPU time until they pressed a key, and the scheduler might queue up multiple key presses before even switching to the user's task (while still showing immediate feedback)

They also had superior IO. Home computers were connected to slow floppy drives (at best, some were still using tape) so they typically tried to keep the whole working set in memory. But multi-user machines had relatively fast hard drives and could afford to shuffle the working set in and out of memory as needed, with DMAs handling the actual transfer in the background.

As others have pointed out, these machines had bank switching. But I doubt such machines were ever configured with more than 256k of memory, with 128k being more typical. Wikipedia claims 32k was the absolute minimum, but with very little free memory, so I suspect 64k would be enough for smaller offices with 4-6 users.

I had after school computer classes on a shared 8080 system running MP/M. Worked very, very well.

It would usually be lightweight data entry, and it would work as fast as people could type (which was pretty fast for dedicated data entry workers).

Often they had bank switching to access more memory.

It wasn't unusual for larger-scale microcomputers to have bank switching in the late 1970s to the mid-1980s.

RAM chips usually have a "chip enable" pin, you might have chips that have 4k of addresses that are 8 bits wide [1] and fill out the 64k address space by having 16 RAM chips, feeding the least significant 12 bits to the RAM chips and the most 4 bits to a multiplexer that goes to the 16 RAM chips. All of the RAM chips are on the bus but only the one with CE set responds.

The same kind of thinking could be applied to extend the address space past 16 bits, for instance you poke at some hardware register and that determines which chip enable pin get sets, there is really no limit on how much RAM you could attach to an 8-bit machine.

A really advanced bank switching scheme for an 8-bitter was on the TRS-80 Color Computer 3

https://www.chibiakumas.com/6809/coco3.php

where the 64k address space was divided into 8k blocks and which might be backed by 128kB (minimum), 512kB (max from radio shack) or more RAM and you poked into a table which mapped virtual blocks to physical blocks. That wasn't too different from a modern memory management system greatly scaled down with the exception that systems like that rarely if ever had a true "executive" mode so nothing stopped user mode software from poking to change the memory map. The CoCo for instance had a multitasking OS called OS-9 that did muiltitasking like described in the explained if you had the orginal Color Computer, you could get Level II that supported more memory and if you never poked at those registers, some memory protection.

[1] at least you did in 1979.

"Mechanically" is funnier expression. But kinda true because it switched also trunk lines in some small PBX. 8080 was bloody expensive in 1976 and RAMs were too. I only heard about because I was employed to design something similar for Telenokia.

I was thinking more about the sharp sm83 chip. Gameboy Advanced is just too modern by comparison...