The problem of time delay in earthquake early warning (EEW) systems restricts their effectiveness. Shortening the delay time can significantly improve the efficiency and emergency response ability of EEWs. In this study, we analyze the delay factors, decompose the period from earthquake occurrence to the warning information release to many links, then calculate the specific delay time in each link. We then compare the warning level and radius of warning blind area in autochthonous, allochthonous, and mixed EEW models, and analyze each model's delaying effect. The results show that the time delay in EEWs mainly comprises delay in data transition and processing; with a minimum theoretical delay time of Δt=3.9 s, an average delay time of Δt=14.3 s in the mainstream EEW, and Δt=8.7 s in the most advanced EEW. Moreover, the time delay exerts influence on the radius of the blind area in a shallow earthquake and a positive relationship can be witnessed. For example, when Δt=8.7 s, the blind area radius is 52.2 km and when Δt=3.9 s, the largest blind area radius is only 23.4 km. The results indicate that effective shortening of delay time can not only improve the timeliness of earthquake early warnings, but also reduce the blind area.