Rarefaction curve scale

Hi all, this is my first post so thanks in advance for any help.
I did a rarefaction curve using a Chao stimator and the resulting curve have mor OTUS than OTUS detected for that sample using the custering command in Mothur. The numer of OTU’s is really big (5,000,000 or so) is there any problem with the scale? If yes, how can it be solved?
Also I have to mention that I used excel to create the graphic.
Thanks again for your help.

The expectation is that the chao should be bigger than the observed number of OTUs. Please see the chao page and the example calculations for how this is derived.