In this paper, we propose a novel pilot beamforming and CSI acquisition strategy for machine-Type-communication systems to achieve reduction in pilot overhead and improve the channel estimation performance. Key idea of the proposed beam-forming technique is to sparsify the time-domain channel vector using the beamforming. To be specific, using the deliberately designed beamforming weight, we minimize nonzero taps and thus sparsify the (beamformed) channel vector. Due to this, only a few samples are required to perform the channel estimation of beamformed pilots so that whole CSI can be acquired with partial pilot symbols in time and frequency. From the numerical evaluations, we show that the proposed scheme outperforms conventional channel estimation schemes, and achieves NT-fold reduction in the pilot overhead. Also, the proposed scheme can be applied for beamforming weight of energy harvesting.