It turned out that the generalized alternative version that takes the length of the string into consideration i.e. this one
Code:
=TRIM(MID(SUBSTITUTE(Text,Delim,REPT(" ",LEN(Text))),element*LEN(Text)-(LEN(Text)-1),LEN(Text)))
...was slightly faster on 1000 rows of sample data as the original formula.

And my alternative formula:
Code:
=MID(Text,FIND("|",SUBSTITUTE(Delim&Text&Delim,Delim,"|",element)),SUM(FIND("|",SUBSTITUTE(Delim&Text&Delim,Delim,"|",element+{0,1}))*{-1,1})-LEN(Delim))
...was slightly slower.

So my advices would be to use the generalized version, that not only is guaranteed to work with your data under greater string lengths, but is also likely to save Excel some effort. Not that you'll notice it: we are talking hundredths of a second over 1000 rows of sample data.