All the evidence I've seen points to electromigration as a cause of this degradation, and IMHO excessively aggressive automatic overvolting by Intel's microcode is to blame.
There is actually a simple experiment which can determine whether that is true --- remove the fan from the heatsink, or even let the CPU run without a heatsink. As the CPU will automatically throttle once it reaches its designed maximum temperature (and AFAIK that is a hardcoded limit), it will lower its frequency and voltage to maintain that temperature. If this results in a stable CPU, while the one that has great cooling becomes more unstable, it confirms the hypothesis.
There are numerous stories of machines where the heatsink was not in contact with the CPU for some reason, yet they remained perfectly stable (but slow) for many years. I can also say that I've had an 8th-gen i7 running at 100% 24x7 with all power and turbo limits disabled, with its temperature constantly at the design limit of 100C, and it has also remained stable for over 5 years.
I once had a laptop, which came from the factory with the four screws which hold the heatsink to the CPU missing. It was very slow, and shut down after a few minutes (the reason being thermal shutdown in the BIOS event log helped diagnose the issue). After the four screws were replaced (each screw came in its own large individual cardboard box), it worked fine for many years, BUT after a couple of years (still under warranty), the motherboard failed with a short in the power input. I suspect that all the extra heat from when the CPU was without a working heatsink went to the power supply components through the motherboard ground plane, and cooked them, significantly shortening their useful life.
Turbo Boost (and Thermal Velocity Boost if applicable) frequencies are according to specifications, it's not an overclock.
Most Raptor Lake "server boards" right now are W680 with client CPUs because the C266/Xeon E-2400 took a long time to come out. The one intended for workstations typically has overclockable settings or is even overclocked by default, which means it's likely to get hit with the failure (1). The one intended for servers do have more conservative settings, but can still be hit with failure (2) under some conditions.
Buildzoid released a video on the Supermicro W680 blade a bit ago that were having issues after running a single-core load 24x7, which is essentially 24x7 boost[1] (aka issue (2)). Xeon E-2400 _could_ be affected in this scenario, although even the highest clock E-2400 SKU (E-2488) is only running at 5.6GHz without Thermal Velocity Boost, and most others are ranging from 4.5 to 5.2 GHz boost (rather than the 5.8 to 6 GHz boost some client SKUs do). I feel like the actual B0 Xeon E-2400 would be a lot less prone to both failures (1) and (2) due to this (but it could happen, though there's no reports of such).
But then the conversation gets muddied enough that "even servers and Xeons are affected" becomes the common narrative (while the former is true, the circumstances needs to be noted; and for Xeons, it's a _maybe_ at most, since right now there's no report of Xeon E-2400 failing).
These are gaming CPUs clocked right up to the threshold of stability (and in some cases, past it)
Server CPUs with ECC RAM are significantly better.
However, if you haven’t experienced much CPU instability, you may not have operated at scale where it appears. Get into the scale where operations occur across 100,000s of CPUs and weird things happen constantly.