Ensuring the effective placement of firebreaks across the landscape is a critical issue in wildfire prevention, as their success relies on their ability to block the spread of future fires. To address this challenge, it is essential to recognize the stochastic nature of fires, which are highly unpredictable from start to finish. The issue is closely linked to the wider problem of climate change, which is causing more frequent and severe wildfires worldwide due to rising temperatures and changing rainfall patterns. Determining the optimal placement of firebreaks in a landscape is a stochastic combinatorial optimization problem that involves the interplay of different management options with the possibilities of a random variable representing the spread of fires, which is currently not well understood. To tackle this issue, our research presents a two-stage stochastic programming approach to model uncertainty in the spread of fires. We thus propose a mixed-integer linear programming formulation to determine the placement of firebreaks, taking into account both the minimization of the expected loss due to wildfires and the expected loss in worst-case scenarios measured based on the Conditional Value-at-Risk function (CVaR). We assess the effectiveness of our proposed solutions by comparing their performance with random plans, where our preliminary numerical results indicate an average reduction of 5% and 9% in the expected burned area and the average of the 10% most intense wildfires, respectively.