It's lots of things, which added together means it's not going to happen.
Firstly, the initial calculation of area required was just to illustrate how small the world's energy use is, compared to the amount of energy we receive in sunlight, even after compensating for the 20% efficiency of PV modules. So it wasn't actually intended as an engineering solution.
Secondly, there is little political interest globally in becoming completely dependent on a handful of North African countries for electricity supply: most countries made that mistake with oil and OPEC, creating decades of geopolitical mess.
Thirdly, it's not a great renewables scenario, because the amounts of energy generated in any two spots of that 1.2% of the Sahara are very closely correlated over all timescales from hours to decades, which means the whole output of the theoretical farm would not show a lot of smoothness. It doesn't span that many timezones, so there'd be several hours every day with no output, for the entire world. Now, that can be addressed with back-up power - just use excess PV to generate hydrogen which is stored, then burnt when its dark to generate electricity then - but that's when it starts getting expensive.
It's cheaper to diversify both the types of energy harvested (wind, tidal, geothermal, hydro, biomass), and the geographic area over which it's harvested (both latitude and longtitude), which significantly reduces the storage requirements needed, and reduces the risks associated with intra-annual and inter-annual variability of supply. This also hugely reduces the geo-political implications, and means that all countries get a decent chunk of internal investment from the project.