Sources: Google is in talks with Marvell Technology to develop a memory processing unit that works alongside TPUs, and a new TPU for running AI models (Qianer Liu/The Information)
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- Google is in talks with Marvell Technology to co‑design a memory processing unit that will pair with its TPUs.
- Marvell Technology will supply the custom memory chip expertise for the memory processing unit, expanding its AI‑focused product line.
- Google is also developing a next‑generation TPU to run AI models more efficiently, adding to its hardware portfolio.
- TPU and the memory processing unit together aim to cut AI inference latency by double‑digits and increase throughput, strengthening Google's AI infrastructure.
Why it matters: Google gains a hardware edge by co‑designing a dedicated memory processing unit with Marvell, cutting AI inference latency by double‑digits, while Marvell secures a high‑margin AI client, reshaping the AI chip market.
