NanoSemi, now part of MaxLinear, delivers its machine-learning powered, self-learning linearization solution to base station, smartphone and Wi-Fi markets. NanoSemi’s linearization solution is already in production with a Tier 1 base station OEM and multiple smartphone products. With successful realization of research to products, NanoSemi is expanding its solution to AI and DNN. NanoTransformer applies the similar mathematical augmentation technology to optimize neural networks. The augmentation is done at a mathematical abstraction level and, when combined with structural optimization such as quantization and pruning, produces NN that consumes up to 10x lower MACs for the same accuracy. This significantly reduces power consumption and latency for edge AI applications. NanoSemi is founded by researchers from MIT/MIT Lincoln Laboratory and has 10 Ph.D.s in mathematics, signal processing and RF.