No, it is not about the numbers, once again. The numbers are just to illustrate the broad order of costs.

If it were about the numbers I would first research ways to cut the costs to sub $1m which is most likely possible by utilizing all sorts of unused hardware (instead of buying), bulk orders (instead of retail), software optimizations, etc. This doesn’t really matter though, I simply took the numbers directly from the whitepaper.

What really matters is that all “free transactions” systems have super strong centralizing tendencies. There is no free lunch. Not in the long run.