Why do we have to provide the jacobians at the Non-linear MPC of Matlab?

Hello,

this is more of a theory question but I would greatly appreciate if you could cite good sources with the theory behind my following question:

Why do we have to define (either numerically or analytically) the jacobians for the NMPC in Matlab?

I can understand that the Jacobian for the A, B and C matrices are for the linearization of the prediction model, but I don't really understand why we need Jacobians for the Cost function and the Constraints as well.