Won't be, until it's out of the lab and into the fab:
but it promises to provide yet another sidestep to Moore's law. I'm hoping it works out as practical as opposed to another lab curiosity that can't hack it in the real world. So far they cite a room-temp on- off current ratio of 100. That's OK for a lab curiosity, but not so much for power or signal switching. Also, impurities seem to do more harm that good; how do they plan on doping these things?
Doping isn't needed to demonstrate a MOSFET, you can glue a gate onto a CdS photocell (i.e., undoped) and 'enhance' it (to the tune of +200V since the insulator is rather thick). In this case, the trouble is increasing the effective bandgap, because as they say, it's too conductive.
Carbon can be doped by functional groups (oxidation, reduction and various kinked bonds, particularly in nanotubes), but the trouble with graphene (not -eme as they spellcheckified...ugh!!) is it's particularly inert over the surface. Reactions only occur on the edges, so you get edge effects. For a device of the scale shown in the figure, that would probably do a good job. A wider channel would effectively be doped and undoped channels in parallel, which is no good (unless you want a "remote cutoff" characteristic!).
Speaking of carbon based materials, I'd love to see a bacteria which is engineered to shit out conductive protein structures. Make biodegradable FETs, then start building wetware with it.
Deep Friar: a very philosophical monk.