For next-generation applications like holographic communication, immersive extended reality, and dense Internet of Things (IoT) deployments, sixth-generation (6G) wireless networks operating at terahertz (THz) frequencies promise extremely high speeds of data, immense connections, and very little latency. In this paper, a detailed overview of hybrid precoding architecture has been provided which encompasses analog digital decomposition, beamspace modelling, and wideband aspect of implementation. In order to justify above techniques, we simulate the DFT codebook based hybrid precoder and compare its performance with completely digital SVD precoding and the random analog baseline. In a 64x16 MIMO system using four data streams to transmit over a sparse multiple path channel the DFT codebook hybrid precoder has a average spectral efficiency of 6.34bps/Hz at 20 dB SNR whilst the fully digital SVD benchmark is 29.55bps/Hz and the random baseline is 16.12 bps/Hz. The DFT codebook hybrid scheme has demonstrated a 78 percent performance difference as compared to the fully digital bound, but it has a large improvement on the random analog beamforming because it is using structured spatial codebook and less RF chain. The findings show that hybrid precoding is an effective compromise of hardware cost and complexity with respect to spectral efficiency and it is therefore a viable choice in next generation wideband massive MIMO systems.