Medical industries are using MRI as the major method of acquiring brain tumors. It also requires an automatic brain tumor segmentation model from 3D MRI images to diagnose and plan the right treatment based on the severity level of the disease. Manual diagnosis is not recommended nowadays because medical practitioners cannot provide more accuracy in tumor detection; it is time-consuming and costly. Earlier MRI brain tumor segmentation methods follow standard image processing methodologies, including multiple algorithms, that take more time and are semi-automatic. Most of the medical industries are using manual segmentation. The main objective of this paper is to provide an automatic 3D MRI brain tumor segmentation with more accuracy. Thus, this paper proposed a U-Net model for automatically segmenting the MRI brain tumor. The proposed U-Net architecture can be divided into three sub-layers, each with different amounts of feature maps for a deep study of the Brain MRI images, which helps detect a brain tumor in patients. The proposed U-Net is compared with an existing algorithm in terms of various performance metrics, and it can be seen that the proposed algorithm outperforms other existing algorithms and provides efficient brain tumor segmentation.