A while ago Si Li and I suggested that the string-theory operation of holography can be captured mathematically by Koszul duality. I'll explain some abstract arguments for this, and describe several examples where this proposal has been checked explicitly, using the material of the previous lectures.