Billions, it would appear. But that doesn't necessarily mean that billions of acres can be taken out of service.

While local conditions don't perfectly extrapolate to the world, remember that 100% of my acres grow human food when observed over the span of more than one year. The 35% animal food production seen in a single year is the outcome of the crop rotation.

Certainly, I'd be far more profitable if I could reliably grow beans for human consumption over and over and over, but Mother Nature is a harsh mistress.

I think the rest of what that post is saying is that 35% is an important part of soil maintenance (ie avoiding another dust bowl). Though I do think with testing we should still be able to drastically reduce the amount of land needed, though likely not exactly 1:1 with previous animal consumption.