There's a lot that goes into this and it's not Apples (heh) to Oranges.
a normal Intel machine has some kind of southbridge, this means a traditional CPU that has 16 PCIe lanes total (this is actually the standard for intel) can have more than just 16x lanes, usually a PCH can add another 20 with an internal PCI switch.
Great in theory, but it does mean you can be constrained in bandwidth since that switch is limited to 16x on the upstream, in fact usually there's only 4! upstream PCIe connections to the CPU and there's some natively attached items (like the USB controller and some lanes of a GPU).
Apple's M-series CPUs do not have any southbridge or PCI switch, you get the full bandwidth of the lanes; this is why Apple M1 can run all of its ports at native speed at the same time, (where most laptops can't).
It also means that when you dedicate PCIe lanes for graphics, they can't just take away bandwidth from inactive ports.
It's only when they increase the die area and add more lanes that they dedicate more lanes to attachable graphics.
It's not a soft limitation, it's a trade-off.
They're banking on the fact that casual consumers won't use more than 2 displays (1 internal, 1 external); and that casual consumers would prefer more ports.
