The White knight in front of his rook on the c-file should prompt thoughts of a discovery. Does the rook have a good target? It does: the bishop on c5 is loose. All that remains is to move the knight out of the way in a sufficiently threatening or profitable manner. The knight has no checks or other threats to make against Black’s king. But since White is going to take a piece at the end of the sequence, he can afford to sacrifice the knight for any lesser gain of material and still come out ahead. Thus 1. Nxd5, BxN, and now 2. RxB nets a pawn. (One can’t expect to win a whole piece every time.) Or Black replies to 1. Nxd5 with Nf6-e4, letting go of the pawn but protecting the bishop on c5.

Black also could try replying to 1. Nxd5 with 1. ...QxN, which protects his bishop on c5, but the protection doesn't last long: White plays QxQ, and then (after Black's recapture) still has RxB. All this now leaves Black with the kernel of a discovery against White on the c-file: his knight masks his rook. But the knight does not have a sufficiently threatening move to make, so the whole sequence is safe for White to play.