Two models to reduce potential risks in photovoltaic enterprises are introduced in this paper to demonstrate the problem that the price of silicon is dropping sharply today. Based on the portfolio theories, the first model is created by the use of Mean-variance model which is traditional and functional. As an improvement, the second model, the Mean-semivariance model takes not only the financial marketing risks but also the background risks into consideration. The data is collected by iFind from three different enterprises. Despite having a large production scale, each enterprise is separated from each other. The calculations indicate that the paper contains a figure that contains the weights of two stocks from two different enterprises. The figure displays the distinction between the effective and inefficient portfolios. Among the efficient portfolios, some of them own a high level of Sharpe ratio while having a low level of deviation, indicating that upstream industries could invest in both upstream industries, and midstream industries in order to weaken the risks brought by the price drop of silicon. This paper will assist investors in making better investment decisions using portfolio theories.