Future cellular networks are expected to support new communication paradigms such as machine-type communication (MTC) services along with conventional human-type communication (HTC) services. This requires base stations to serve a large number of devices in relatively short channel coherence intervals, which renders allocation of orthogonal pilot sequences per-device in each cell impractical. Furthermore, the stringent power constraints, place-and-play type connectivity and various data rate requirements of MTC devices make it impossible for the traditional cellular architecture to accommodate MTC and HTC services together. Massive multiple-input-multiple-output (mMIMO) technology has the potential to allow the coexistence of HTC and MTC services, thanks to its inherent spatial multiplexing properties and low transmission power requirements. In this work, we first tackle the optimal non-orthogonal pilot design problem and demonstrate that the optimal pilot sequences are Welch bound equality sequences. In the second part, we investigate the performance of a single cell under a shared physical channel assumption for MTC and HTC services and propose a novel scheme for sharing the time-frequency resources. The analysis reveals that mMIMO can significantly enhance the performance of such a setup and allow the inclusion of MTC services into the cellular networks without requiring additional resources.