There are programming limitations with higher numbers unless you are only fine with two or three decimal places. Otherwise I actually agree with your points with the supply.
What is the programming limitation in total places?
There aren't any, so I'm not sure what limitations he may think there are.
I'm assuming the poster isn't a developer and has made an assumption based on the reading of some potentially incorrect document.
BigInteger/BigDecimal math solves any ranging "issues" you might encounter quite easily, for any number you might dream up.
So long as you aren't wanting to perform many operations on very high power numbers (10^1024 etc) the performance hit is negligible.