Obviously this is a bit silly to argue about in the abstract, but if there are a number of different domains where 1% problems threaten the business (AWS, security, random other service is down...) and a senior engineer isn't expected to know about all of them, then you'll need some multiple of 1% of the engineers to feel resilient.

Rough amount: number_of_domains * max(ceil(1% * workforce), bus_factor_min))

At small companies, that may be most engineers. At larger companies, it may (or may not?) level out to a more reasonable ratio.