The way this works in the server world is “95th percentile” billing. They track your bandwidth usage over the course of the month (probably in 5 minute intervals), strike off the 5% highest peaks, and your bill for the month is based on the highest usage remaining.

That’s considerably more honest than charging you based solely on the highest usage you could theoretically use at any time point in a 24 hour period (which is how ISPs define the “max bandwidth”) and then charging you again or cutting off your service if you use more than a certain amount they won’t even put in writing.