Google in Talks with Marvell for New AI Chip Development

Google, a significant player in AI and cloud computing, develops its Tensor Processing Units (TPUs) for internal workloads and external clients. This strategy reduces reliance on third-party vendors and customizes hardware for specific AI demands. Google has historically collaborated with Broadcom on custom AI chip development. The reported discussions with Marvell suggest a strategic move to broaden its partnerships and enhance supply chain resilience.

Google is in talks with Marvell to co-develop two new custom AI chips. These chips reportedly include a new TPU, designed to run AI models, and a memory processing unit, intended to support Google's AI accelerators. This collaboration marks a potential shift in Google's chip design ecosystem.

Marvell's market performance reflects growing investor confidence. Its stock rose 95% year-to-date in 2026 and 227% over the last twelve months. The company projects its data center revenue could reach nearly $19 billion by 2028, based on a $94 billion addressable market for custom silicon and data center solutions.

A partnership with Google, a leading hyperscaler, would represent a significant win for Marvell. It expands Marvell's footprint in the custom AI processor market, an area where it aims to increase its market share. This collaboration positions Marvell to capture a larger portion of the expanding AI chip demand, creating new revenue streams and strengthening its competitive standing. Observers will monitor official announcements from both Google and Marvell regarding this reported partnership, along with its potential impact on future chip development timelines and market dynamics.