Learning Curve Breakdown (work in progress) March 17th 2010.xlsx

The spreadsheet is set up to model the price of photovoltaics over the next 10 years. It assumes a growth rate of 25% per year and an initial learning rate of 18% that steps down year by year to simulate the effect of learning saturation.


I added additional columns that crudely model the change in wafer thickness over the next 10 years. I linearly scaled today's wafer thickness ~ 180 um and today's kerf loss ~140 um to 100 um each in 2020 which I've seen listed as a common 10 year goal. The polysilicon price trend is an educated guess based on FBR silicon production costs of around $20/kG.


The idea of the wafer and poly columns is to sketch out an additional cost projection that is independent of the learning curve's prediction. This allows a crude comparison of the two. So far the comparisons don't line up very well because my sketch of polysilicon's price future has a sharp drop in the near term due to the glut of poly on the market.


This is a work in progress. Suggestions and critiques are welcome.