You can only run a certain number of pods per node before the kernel craps out. Depending on your K8s version somewhere in the 250-500 range.

Admittedly, this is unlikely to be an issue at home lab levels. Therefore one must suspect the “because I can” methodology is at play here.