🔥Google is in talks with Meiman Electronics about a partnership to develop two AI inference chips to reduce dependence on Broadcom

According to Beating Monitoring, Google is negotiating a partnership with Meiman Electronics to develop two custom chips, aiming to improve the efficiency of AI model operations. This move accelerates Google’s efforts to diversify its AI hardware suppliers, reducing reliance on its long-term partner Broadcom and cutting costs. The two chips include a new tensor processing unit and a memory processing unit. Google plans to produce approximately 2 million MPUs, and the design work is expected to be completed as early as next year. Inference chips are responsible for running pre-trained models and are core components driving commercial products such as autonomous agents. Broadcom is Google’s only long-term design partner for TPUs and charges high fees for each chip. As TPU demand surges, Google is spreading risk by bringing in Meiman Electronics and MediaTek. Meiman Electronics previously helped startups…