This paper proposes a novel joint channel estimation and beamforming approach for multicell wideband massive multiple input multiple output (MIMO) systems. With the proposed channel estimation and beamforming approach, we determine the number of cells $N_{c}$ that can utilize the same time and frequency resource while mitigating the effect of pilot contamination. The proposed approach exploits the multipath characteristics of wideband channels. Specifically, when the channel has a maximum of $L$ uncorrelated multipath taps (or correlated multipath taps satisfying modest criteria which is valid in most practical scenarios), it is shown that $N_{c}=L$ cells can estimate the channels of their user equipments (UEs) and perform beamforming while mitigating the effect of pilot contamination. The proposed approach can also be applied for general correlated multipath taps, and achieves good performance for this scenario as well. In a typical long term evolution (LTE) channel environment having delay spread $T_{d}=4.69\,\,\mu \text{s}$ and channel bandwidth $B=5$ MHz, we have found that $L=36$ cells can use this band. In practice, $T_{d}$ is constant for a particular environment and carrier frequency, and hence $L$ increases as the bandwidth increases. All the analytical expressions have been validated, and the superiority of the proposed design over the existing ones is demonstrated using extensive numerical simulations both for correlated and uncorrelated channels. The proposed channel estimation and beamforming design is linear and simple to implement.