While PCIe 4.0 is extremely short-lived as a product, and I suspect Intel might even launch 5.0 with their next platform if they decide to hurry (which would go especially well if their GPUs launched as PCIe 5.0 cards, which they should be aiming for if they want to market them as superior products), one has to look at AMDs decision with future-proofing in mind. I think it was a fabulous move in favor of consumers, especially those who got older 3.0 boards that will now get 4.0 speeds enabled.
In the mainstream, where GPU support is the most relevant factor, 3.0 has not been maxed out yet, and 4.0 offers twice that speed. Even if cards supporting 5.0 speeds launch in the upcoming years, they will still not even reach anything close to actual 4.0 bandwidth usage (twice the speed of current 3.0), meaning they will work on 4.0 boards as well as they would on 5.0 boards. I actually suspect those 4.0 boards will be able to support max speeds of GPUs for many, many years to come.
While 3.0 might approach bottleneck area for highest-end GPUs in 2 or 3 generations, 4.0 won't for a long time. 5.0 simply has an insane additional reserve, considering it's quadrupling 3.0's bandwidth. 5.0 cards won't work faster on 5.0 boards than 4.0 boards unless 4.0 can't provide enough bandwidth though, which isn't likely to be the case for a very long time with the 3.0 not even bottlenecking the 2080Ti (it actually seems to be the very first card that merely shows any advantage in full 16x3.0 vs 8x3.0). Again, 4.0 is twice as fast.
While at first it might seem that there's no point to bother with 4.0 and we should've jumped straight to 5.0, we can actually get 4.0 right now, and for most users it will be as good as 5.0 for the lifetimes of their platforms.