The Alan Turing Institute, the British Library, London NW1 2DB, United Kingdom Determining design principles that boost robustness of interdependent networks is a fundamental question of engineering, economics, and biology. It is known that maximizing the degree correlation between replicas of the same node leads to optimal robustness. Here we show that increased robustness might also come at the expense of introducing multiple phase transitions. These results reveal yet another possible source of fragility of multiplex networks that has to be taken into the account during network optimisation and design.Multilayer networks [1-3] formed by several interacting layers have emerged as a powerful framework to analyse a variety of complex systems, including such classical examples as global infrastructures, economic networks, temporal networks and the multi-level structure of the brain. Other disciplines, as material science and chemical synthesis, are on the way to adopt the network science toolbox, less for analysis purpose, but mainly for its potential for optimisation and rational design [4][5][6][7][8]. Therefore, predicting the robustness of multilayer networks [9-11], assessing the risk of large avalanches of failures [9][10][11][12][13][14][15], and designing robust multilayer architectures [16][17][18] are the key theoretical questions entailing implications for engineering, economics, and biology. Recent works on percolation in multilayer networks revealed that the correlation between intra-layer degrees of replica nodes [16,17] and link overlap [19][20][21] have profound consequences in determining the response of a multiplex network to random damage. The case of positive interlayer degree correlation [1,16], when a hub node in one layer is likely to be interdependent on a hub node in another layer, is known to increase robustness of interdependent multiplex networks. It is widely believed that maximizing the inter-layer degree correlation is a good design principle for building robust infrastructures. The network optimization is used not only for engineered networks and infrastructures [22][23][24][25][26][27] but also for economic [28] and biological networks [29,30]. Therefore, it is of fundamental importance to understand how maximizing the inter-layer degree correlation may affect the properties of multiplex networks.