Is it possible to make regression models or time series forecast models where both the independent and dependent variables are datetimes?

I have a small dataframe where both columns are datetime datatypes, so if I were to plot, both the x and y are datetimes.

I basically want to make a model that predicts the delivery/due date of something, based purely on its start date.

Nothing I try seems to work. Does anybody have any ideas, would something like this be possible?

(I've tried LinearRegression via sklearn and ARIMA modelling but got stuck with my dataset.)