Thinking about your average end-users, like most of my family, 8 to 16GB is about all they need for their systems (and if software were better written they'd probably need less). So is this specific work like machine learning? Video and image processing?
If you have a microservices architecture or are using Kubernetes then you will easily need more than 16GB. I have 128GB on my desktop purely so that I can have my full platform running in the background whilst also doing development.
Also if you're doing data engineering e.g. Spark then you will likely want more than 16GB.
I have 74GB on my system (10GB of which is GPU) so I can run my dev env (Kubernetes, PGSQL, Visual Studio), data science, machine learning and do 6K video editing. But then, there's also zero chance that I would consider doing this on a laptop.
Which is why I use Final Cut Pro. It was a little sluggish from time to time on my Air with 8GB of RAM, now on my mid-2015 MBP with 16GB of RAM it never stutters or slows down. Editing 1080p.
What are you doing that needs them all actually running simultaneously? Disk-backed VM pause/resume is extremely fast when your SSD's sequential read/write performance is measured in GBps.
I have a hard time imagining a workflow that actually uses multiple systems all at once, so to me what you're saying sounds like "Well of course my RAM gets filled if I fill it on purpose."
Often it is. And so whilst SSDs are fast they still aren't fast enough to not slow everything else doing under high load.
I've bought One Plus 3 (6GB RAM) in 2016, still strong. Computing power has not changed much, limiting factor is RAM.
I don't want to be forced to close my complete dev. environments to work on another project; or play with a ML project.
The cost of additional RAM is marginal, to me it's the way it should be. Your time matters, your productivity matters, why not max it out?
When I buy an Apple computer, I pay a premium but I plan to use that computer for several years.