Google talks with Marvell to add a third custom AI inference chip partner
Google is exploring a partnership with Marvell to design a memory processing unit and an inference‑optimised TPU, adding a third custom‑silicon partner as the AI market shifts toward inference‑heavy workloads.
