How can I estimate monthly server costs for N customers?*

I am trying to put together a feasibility study for a python/django powered webapp I am developing. How can I provide an estimate of my monthly hardware/bandwidth server costs in the feasibility study? I assume it would have the form: *M->thousand 1) 0-2M users: $200/month 2) 2M-5M users: $500/month 3) 5M-10M users: $X/Month 4) 10M+ users: $Y/Month with cost of $Z per thousand users.

These are just random guesses at numbers and ranges. What are some practical numbers to use for back of the envelope calculations? It doesn't matter to me what technology would be used for each, just what some practical dollar costs would be in those ranges.