In this correspondence paper, we design a new computationally efficient blind carrier frequency offset (CFO) estimator for multiuser orthogonal frequency division multiplexing (OFDM) uplink. We consider that massive receive antennas are configured at the base station (BS), such that channel matrices of different users tend to be orthogonal to each other. The proposed method first performs multiuser interference (MUI) cancellation, after which the multiuser OFDM uplink model can be decomposed into a set of parallel single user signal models. The conventional single user CFO estimation can then be carried out individually. Moreover, considering the number of antennas at the BS is limited in practice, an iterative procedure is further developed to gradually improve MUI cancellation as well as CFO estimation. The provided numerical results validate the proposed studies. As compared to existing multiuser CFO estimators, the proposed method can achieve much better estimation performance with a lower order of computational burden.