Anecdotal data point: The other day I was toying around with an eGPU, with the card sporting 4GB VRAM (system has 32GB Ram installed). When connected to a 4K monitor, the GPU‘s VRAM filled up to ~2/3. Attaching a second monitor (3440x1440) would bring the VRAM to full load and the system got noticeably slower. No gaming - just working in finder (Big Sur).
On a mini there‘s no dedicated VRAM, therefore all required video memory is taken from main Ram. 4K monitors are nothing unusual anymore, so running one with an 8GB mini could easily bite quite a chunk off the system‘s main Ram.
For comparison: 8GB of dedicated VRAM is considered only average in the gaming community nowadays - and 4GB is seen as lower boundary for anything remotely modern and “only” Full-HD resolution. I don’t know what Apple may be storing in Ram when it comes to Quartz Extreme, but the finder easily filling the 4GB VRAM on the eGPU seems to indicate that 8GB Ram on a Mac may quickly become restrictive in the near future, especially on high-res and/or multi-monitor setups.
Sure - the M1 machines are said to handle Ram better than Intel machines, but you can’t upgrade the Ram after purchase (except for buying a new machine) and swapping to an SSD is still by magnitudes slower than direct Ram access: The most modern SSD’s in a Mac have read/write speeds around 4-8 GB/s, whereas the M1Pro chip boasts 200GB/s. Go figure …
Call it fear-mongering, if you like, but I know for sure that my next mini will have 16GB Ram as minimum and I’m probably even going for 32GB, as I’m working a lot with 3D objects.