I had 16GB in my work M1 Mac Pro, it was barely enough when I was mainly doing remote work via SSH and web-browsing (4-5GB of Swap constantly), but would have issues with any decent sized local data analysis (R, Jupyter notebooks etc). Switched to a project that required a lot more software development (Visual Studio Code, Scala, Rust and Python) and local data analysis (Python, Polars and PyTorch) and it was not sufficient was getting "you have run out of memory errors every couple of days (maxed out swap)", or when ever I ran a decent sized analysis.
Switched to a 32GB M2 Pro, and that is sufficient for most things, I've not hit the out of memory error since, but there are points with moderate data analysis, a couple of IDE's and Jupyter notebooks open were it is tight (10-16GB Swap).
Desktop is 64GB + 12GB VRAM and that's much nicer, never any memory problems and if I'm looking at datasets that need more than 64GB of RAM they're typically hungry for more CPU/GPU compute than my 7900 can deliver, so at that point they get transferred to to a AWS VM (128-256GB RAM) or one of our HPC servers (512-1024GB RAM, 64+ cores).
Partner was running a 32GB desktop, but that wasn't sufficient for GIS workloads and heavy stats modelling, so bumped her to 96GB which has solved the problem (probably only needs ~64GB, given the typical workloads).