First of all, it's a textbook exercise in topology to show that any rational decision process must have a utility function (at least for a countably infinite set of choices).
Let me repeat the statement for utility functions. Suppose I have a 1% chance of being CEO and increasing my utility to U(CEO pay) - U(my pay). Suppose this increases to U(10x CEO pay) - U(my pay).
It's simple arithmetic that 0.01 x U(10X CEO pay) + 0.99 x U(my pay) > 0.01 x U(CEO pay) + 0.99 x U(my pay).
Rearranging the arithmetic, this is merely the statement that U(10X CEO pay) > U(CEO pay) - i.e. I'll be happier as a CEO with 10M than with 1M. Do you disagree with this statement?
In the unlikely event you are actually interested in the correct formulation, you start to find the answer here [1].
[1] https://en.wikipedia.org/wiki/Marginal_utility#Quantified_ma...
There is a reason people are willing to buy $1 lottery tickets but not $1000 tickets. (hint: a certainty of misery or high chance of death is not compensated by a remote chance of enormous wealth of ever decreasing marginal utility)
You are simply misunderstanding the use cost benefit, I assume because it clashes with some ideological point.
But if you want to buy a one in a million chance for a trillion dollars, I have a bitcoin address you could send a million bucks too.