Here's a soil map of Africa. Focus on the Sahara desert. Entisols, Yermosols (Yermic Andosols) and aridisols abound. It wasn't always like this, and there was clearly a process that made it less viable for agriculture than it currently is, going by testimony relating to the history of Mauritania, and their ample supply of Gum Arabic. From what I've read, Nitrogen-fixing bacteria is already present, and I've got an extra excerpt that should apply to this.

>Most Arenosols in the dry zone are associated with areas of (shifting) sand dunes. Evidently, soil formation in such dune sand is minimal until the dune is colonized by vegetation and held in place. Then, some humus can accumulate in the surface soil and a shallow, ochric surface horizon can develop; `Aridic' Arenosols contain less than 0.2 percent organic carbon and show evidence of aeolian activity. The sand grains of Arenosols in the dry zone may acquire a coating of (brownish) clay and/or carbonates or gypsum. In places, desert sand is deep red by coatings of goethite (`ferrugination', a relic feature according to some). Where the parent material is gravelly, sand is blown out of the surface layer and the coarser constituents remain behind at the soil surface as a `desert pavement' of polished pebbles and stones. `Yermic' Arenosols may be found in such situations. Depending on parent material and topographical situation, `Gypsiric', `Calcaric', `Hyposalic' and `Hypoduric' Arenosols, or combinations of these, occur as intergrades to Gypsisols, Calcisols, Solonchaks and Durisols. High permeability, low water storage capacity and low biological activity all promote decalcification of the surface layer(s) of Arenosols in the dry zone, even though the annual precipitation sum is extremely low.

What would it take to make this land usable for large-scale agriculture? Money's no object. No "Silicate-based trees" answers unless there's actually some reasonable precedent.