Abstract: In some isolated but significant incidents, such as during the COVID-19 pandemic, bandwidth demand suddenly surges to much higher levels from the normal demand levels. A sudden surge in demand may impel users to lose their access or having a poor online experience due to bandwidth throttling by their Online Service Provider (OSP). Earlier work on optimal user demand throttling mechanism under stochastic demand has overlooked demand surges. In this paper, we recast the user demand to adequately capture the surges in demand, such as home bandwidth demand surges during the COVID-19 pandemic; specifically, we model the user demand as a geometric Levy (jump—diffusion) process. Within our general setting, we show that it is optimal for the OSP to follow a modified control-band policy that encompasses the existing results as a special case. Our numerical study not only enhances the current insights about the optimal throttling mechanism by including demand surges but also provides new insights concerning the nature of demand surges. For example, our study suggests that OSPs should initiate demand throttling at relatively lower usage levels if the likelihood of having demand surges is high; furthermore, we find that when OSPs are exposed to demand surges with higher intensities, it is optimal for them to wait until the usage level reaches a relatively high level before throttling it.