The standardization of the flexible dense wavelength division multiplexed (DWDM) grid and the ability to explore higher order modulation formats have been key to simultaneously improve the spectral efficiency and reduce the cost per bit transmitted in optical transport networks. These developments will be complemented by the availability of next-generation line interfaces capable of operating at higher symbol rates, which will further augment the capacity per line interface. However, maximizing the benefit of deploying next-generation line interfaces in existing optical transport networks requires taking into account the consequent coexistence of optical channels at different symbol rates in the network design process. Accomplishing this demands efficient spectrum management combined with an optimized provisioning of hardware throughout the entire network life-cycle. This paper investigates the resulting network optimization problem. In detail, it proposes a multi-period design framework based on integer linear programming (ILP) models to optimize the different types of line interfaces deployed with the aim of minimizing the capital expenditures (CAPEX) and spectral resources’ usage. Simulation results highlight that the proposed framework is able to recover from the initial overprovisioning of spectrum inherent to using a virtual grid and achieving spectral-savings in the medium-term of the network lifecycle. Moreover, it also shows that this is attained while simultaneously carrying more traffic and without increasing the number of the line interfaces.