谷歌与Marvell洽谈开发新型AI芯片

Google is reportedly in talks with Marvell Technology to develop two new artificial intelligence chips, expanding its custom silicon supply chain as the company shifts more spending toward inference, the increasingly expensive stage of AI where trained models respond to users at scale. The discussions, first reported by The Information and later summarized by Reuters, involve a new memory processing unit designed to work alongside Google’s Tensor Processing Units and a separate TPU built specifically for running AI models more efficiently. Google and Marvell have not publicly confirmed a final agreement. The talks show how quickly the AI chip market is changing. Google has spent years building TPUs as an alternative to Nvidia GPUs inside its own cloud infrastructure. …