Yeah, I saw that the other day.
For hobbyist purposes, my main reason for sticking with x86-64 other than performance to cost is convenience, since just trying to run a Linux desktop OS on an Arm SoC is likely to expose more untested (or unwritten) branches of code (whether in some library, build script, or kernel driver), to the point that even seeking out something like an Arm or MIPS based laptop is not even worth it. Also, a lot of the hobbyist single board computers using Arm SoC's you see on the market (like the so-called "plug computers") still have the usual plethora of hardware bugs in their peripheral cores, but which kernel developers seem NOT to have contributed patches for workarounds to (unless, of course, your Arm SoC happens to be used by Android, but in that case, say hello to various proprietary binary blobs in the kernel just to get things like Wifi and bluetooth working--although I could say the exact same thing about Realtek on x86-64).
But if Intel chips really do start to show general flaws in comparison to the performance that Arm can achieve with the same level of fabrication technology, then, well, server applications are probably the most trivial to repackage, and you don't even need to worry about all the drivers or whatever weird laptop issue you might have at all: just stick a Docker binary in an AWS instance and do the math for performance / cost, and maybe it'll be the beginning of the end for Intel....