Also, low-water-use plants tend to also be slow-growing. That's why they use so little water.
Hesperaloe produces 1.5 kg per cubic meter of water. This is comparable to the water use efficiency of forests used for paper, which is between 1.5 and 4 or so. So this actually requires as much water as trees, if not more. The only advantage, if you can call it that, is being able to plant it in places were trees won't grow.
But the problem is that this also means you'll be cultivating plants grown far apart, none of which produce very much mass per plant without taking so much that you kill the plant (whose roots are likely the only thing keeping the topsoil from eroding). That means you need an order of magnitude more land per unit of crop mass, and probably closer to two orders of magnitude. (Trees get tens of feet tall. Hesperaloe is a pile of long leaves that gets maybe a foot tall.) So the energy consumption involved in harvesting it is likely to also be one to two orders of magnitude greater because you might have to harvest an entire acre worth of red false yucca to get as much biomass as you'd get from cutting down one fully grown coniferous tree in a few minutes.
So you've gotten a little bit more crop land that nobody asked for, and/or saved a bunch of trees that were planted specifically to turn into paper, and in exchange, you've introduced a massively energy-intensive harvesting process followed by some other probably massively energy-intensive conversion process. This doesn't seem like a net improvement to me.
Either that or you harvest it including the root system, and yeah, you've gotten a lot of raw material, but now your topsoil erodes, and you can't plant anything there for a thousand years.
The process itself might be useful in some other context, of course, but it seems like a solution in search of a problem.