The Internet of Things (IoT) and particularly Machine-to-Machine (M2M) communications are considered as major enablers for future smart cities' initiatives. While offering a wide range of applications and services, supporting such devices constitutes, however, one of the most important challenges to be faced by Network Operators (NO). Indeed, the expected huge number of devices requesting to connect to the network at the same time may result in severe congestion in the access network with a high risk of congestion collapse. Different schemes were proposed in the literature to solve the congestion problem by regulating the M2M devices' opportunities of transmission. Nonetheless, as revealed in this paper, these schemes turn out to be ineffective in case of heavily congested M2M networks. In fact, in such a condition, the unpredictable and increasingly accumulated number of devices cannot be blocked. This augments the risk of M2M devices' synchronized access, which may result in a congestion collapse. We focus in the following on showing the impacts of this phenomenon while highlighting its main roots. The paper also investigates different key directions for a more efficient management of M2M devices.