Urban air pollution is a major environmental problem affecting human health and quality of life. Monitoring stations have been established to continuously obtain air quality information, but they do not cover all areas. Thus, there are numerous methods for spatially fine-grained air quality inference. Since existing methods aim to infer air quality of locations only in monitored cities, they do not assume inferring air quality in unmonitored cities. In this paper, we first study the air quality inference in unmonitored cities. To accurately infer air quality in unmonitored cities, we propose a neural network-based approach AIREX. The novelty of AIREX is employing a mixture-of-experts approach, which is a machine learning technique based on the divide-and-conquer principle, to learn correlations of air quality between multiple cities. To further boost the performance, it employs attention mechanisms to compute impacts of air quality inference from the monitored cities to the locations in the unmonitored city. We show, through experiments on a real-world air quality dataset, that AIREX achieves higher accuracy than state-of-the-art methods.