Ultrasound medical imaging systems typically follow standard delay and sum (DAS) beamforming at the reception for image reconstruction. In DAS, the echo signals that are returned to the transducer are aligned in time and summed to form the beamformed signal. To improve the image quality and the signal to noise ratio of DAS, a non-linear beamforming named delay multiply and sum (DMAS) has been proposed in the literature, where, the signals arriving at the transducer are aligned in time and are pairwise multiplied in all possible combinations before summation. This provides better coherence, a correlation-based data-driven apodization, and consequently result in better contrast and resolution. However, the computational complexity of DMAS is higher than DAS thus restricting its real-time implementation. This paper presents two novel VLSI architectures for the implementation of DMAS, whose complexity is independent of the number of transducer elements. The proposed architectures are implemented on xc7z010c1g400-1 FPGA and the results clearly show the channel independency of the proposed architectures. © 2020 IEEE.