my spreadsheet, page-32

FF your previous post was ill informed; totally inappropriate model thrown at totally inappropriate numbers.
Yes quadratic curve fitting is better than linear, if you have the data. Parabolic curve fitting is a nonsense in this type of context. (It's quadratic minus the linear component by the way.)
Finally throwing ramp-up numbers at a poor model then drawing conclusions about steady state operating expenses is simply ignorant of the engineering and financial realities.
The company's guidance has remained remarkable steady throughout the last 12 months on this; perhaps they know the plant better than either of us.