Given that starter sizes for the average homebrewer range from 0.5 to 4 liters, and that all these calculations can vary by as much as +/-30% in the real world anyway. The numbers between the two formulas are not that far off from each other.
The real difference, with regards to smackpacks and vials, is in the way the viability of the initial cells pitched affects the final cell count.
According to Kai's model it doesn't really matter if you pitch 100 million cells or 100 cells you still end up with a crap load of cells at the end of fermentation.
Here is where I feel we need to be careful though, because the purpose of making a starter isn't to just propagate a crap load of cells. The purpose of making a starter is to propagate a crap load of healthy cells; and in order for the cells to have an environment conducive to healthy reproduction there needs to be a balanced ratio of cells to food.
This is where the importance of inoculation rates come into play. Too much wort (low inoculation rate) and the yeast become stressed from excessive budding, (this is talked about in Boulton & Quains "Brewing Yeast and Fermentation") There is a limit to how many times a yeast cell can bud before it becomes senescent, something like 24 to 40 generations.
Yes, theoretically you could toss 100 yeast cells into 2 liters of wort and end up with 281 billion cells, but how many of those 281 billion cells are still up to the task of fermenting 20 liters of wort?
Keeping the inoculation rate of your starter in the range of say 25 to 100 million cells per ml will help ensure that the cells you have propagated are strong, healthy and vital to the task of fermenting that recipe you spent so much time deliberating over.
If your going to use Kai's formula (or Jamil's for that matter) I'd suggest trying to keep the inoculation rate for your starter close to 25 to 100 million cells per ml.