I've been in IT for decades, and it's been a hobby for longer.
Long ago, I realised that "future proofing" is largely a mug's game. Or something very similar to FOMO. Outside of a small number of areas, it's usually not worth doing. Specs change so fast, gear gets cheaper so fast, tech goes in different directions than expected. It's often better to go middle of the road then buy again in 2-3 years after a generation or two has passed.
The only areas I can think of where "future proofing" is really relevant is perhaps cable installations, where it's the cost of labour that's high, not the materials. Even then... Cat 5e is still 100% fine at home. I guess because despite some expectations, 10Gb never got anywhere for home deployments. Buying cases and PSUs for the long term can work out, but these are static products.
Motherboards, CPUs, storage, giant amounts of RAM, these things either come and go so fast, or drop in price so fast, that it's not worth trying to buy for five years down the line.
In a corporate environment, we don't 'future proof', we buy for long term product support, physical scalability (does the DC have enough rackspace for future expansion, if there's a doubt we buy the option for extra racks now). We don't buy for hypothetical workloads in 3 years time because we just buy more equipment.
Oddly enough, talking to my tech friends last week, we came across an area where future proofing worked out by accident. They tend to buy the Intel K CPUs, test overclocking ability, then promptly set them back to base and forget about them.
Turns out that Intel have stalled so long, and that lack of progress on mutli-threaded games has been so consistent, and AMD went for core count rather than clock speeds, that their generations-old systems will match anything available today if some tweakng is done. This was pure luck.
