One piece of infrastructure trips offline, causing an abnormal situation at another, which then trips etc.
That pipeline/powerline is your single point of failure.
It is still a single network, but you need multiple failures to manifest into a cascading issue like this.
What you're probably thinking of is more akin to distributed system which fails when any one component fails, i.e. modern microservice architectures, aka distributed monoliths. But that's not the case here, because you constantly have minor issues on the grid. They're just continuously being handled. What becomes the issue is the cascade, with each failure increasing the likelihood of a following failure etc.
In web developer terms, this is as if a production k8s cluster fails because a node went offline, which rebalanced too many containers to another node which ran out of memory, causing it to crash and starting the cascade, ultimately ending with adjacent clusters starting to crash because of the error quotas etc
It starts with a small thing at 12:15 p.m. and takes a while for other things to happen. But then bam 4:05:57 p.m. to 4:13 p.m. results in:
End of cascading failure. 256 power plants are off-line, 85% of which went offline after the grid separations occurred, most due to the action of automatic protective controls.