This is bull. It's not zero cost to fork a cryptocurrency. It's not zero cost to mine. The supply of a digital good might well be very large. However, it will most certainly be finite.
Then you're limited to "What's the biggest number that a miner can name in the finite amount of memory available?", which has the ultimate limit of a Busy Beaver function.
BB(n) for n > 1500 behaves like an infinity in many important ways.
Then just apply my original argument. It's not zero cost to mine. It's not zero cost to fork. The supply of cryptocurrency will be very large, but it will be so many x orders of magnitude smaller than BB(kerjillion), that x itself might be larger than the number of every cryptocoin ever made and every cryptocoin that ever will be made.
The useful thinking is about the computational limits of our current civilization, not infinities or redonkulous numbers.
My point was that other traditional "store of values" have inherent value based on their utility per amount, which means that the finite supply has implications for people who want to use it. Bitcoin doesn't have that - it's purely useful in terms of what you can trade it for (much like a dollar.) Hence the fact that there's only so many "units" is kind of uninteresting, because a fractional unit is no less useful.