It's actually not very scaleable. The yield is incredibly low. Processing plants is also difficult because you have to deal with a bunch of stupid stuff like tough cell walls, etc. I'm not even sure that the glycosylation patterns that tobacco puts on the antibodies are human-compatible (i.e. problems could occur if you try to use tobacco-derived antibody therapeutics in a patient more than once).
The choice to use tobacco was very likely a symptom of "if you have a hammer," common in biology and chemistry, where some tobacco researcher (tobacco is a well-studied model organism) wrote a convincing grant with a handful of accurate but marginal improvements over contemporary systems (in this case glycosylated vs. unglycosylated) that would enable commercialization, and downplayed some of the setbacks (low yield). Meanwhile other technical approaches have matured that are really good at tackling those problems but without the other setbacks... but just weren't awarded seed money.
