The global climate change is poised to increase the global temperatures in near future. It is estimated that the average global temperatures are expected to increase by 0.50 °F (0.28 °C) to 8.60 °F (4.78 °C) by the year 2100. Climate change is also responsible for the change of weather patterns and extreme events across the globe. This can have implications on preliminary HVAC systems design as it can be over/undersized resulting in discomfort or decreased energy efficiency. We purpose to create a Generative Adversarial Network (GAN) model to generate the HVAC energy usage profile of the residential dorms during the heating months. Multiple generative models for both before and after the building energy retrofit are created. The discriminators are trained using the real HVAC data obtained from seven residential dorms located in Syracuse, NY, both before and after the retrofit. The model also takes month of the year and outdoor temperature as inputs. It is essential to take month as an input variable as college dorms remain empty during holidays. The generated HVAC load is validated using the methodologies established in earlier studies. The future Typical Meteorological Year (fTMY) file developed by Oak Ridge National Laboratory (ORNL) is used as an input to the validated GAN models to generate the future HVAC energy consumption. The generated results from both models will help to quantify the potential energy savings from building retrofit in the future.