The problem with future proofing is that you can never predict what new feature will end up becoming the next big thing or the next bottleneck. Sure, if everything works out exactly as predicted, if what the companies said was going to be the next big thing, really did end up being the next big thing, then you'd be future proof for up to 10 years. But then what if it turns out that the next big thing wasn't what anyone predicted it was going to be? Think about the whole VRAM issue that came up last year. Almost overnight, supposedly high end GPU's got instantly re-evaluated to be not worth the money. DLSS, too - the newest version only supports the newest generation. And mesh shaders - even some GPU's from a couple years back don't support it. Out of all the things that people predicted was going to be the next big feature, how many actually ended up being the next big feature? You end up just as worse off as if you had just went for a lower-tier option.

I'm not advocating for replacing your GPU frequently. Far from it. But the argument of splurging on an excessively overkill tech part for futureproofing is just marketing mumbo jumbo

Just wait until 2031 when they end support for Windows 11 and require you to have an 18th Gen CPU to run Windows 12.

Let me put it this way, you could buy a 4090 now for 2k, then get prolly like 7-10 years out of it, or you could get something more midranged for like 300-400, and if you upgrade every other gen for the same price that means you spent at most 1200 which is more than 50% less than if you'd gone for the 4090 to be "future proof"