Google in Talks with Marvell to Develop AI Chips, Report Says
The proposed designs include a memory processing unit to complement Google’s existing accelerators and a next-generation TPU for AI workloads.
Topics
Alphabet’s Google is in talks with Marvell Technology to develop two chips aimed at improving the efficiency of running artificial intelligence models, The Information reported.
The reported effort involves two designs. One is a memory processing unit intended to work alongside Google’s tensor processing units (TPUs).
The other is a new version of the TPU tailored for running AI models.
The chips are expected to improve how data is handled and processed in large-scale AI systems, an area that can limit performance.
Google has been expanding its use of custom silicon as it seeks to reduce dependence on Nvidia, whose graphics processing units are widely used for AI workloads. The company has positioned its TPUs as an alternative within its cloud business.
Demand for AI computing infrastructure has increased as companies deploy generative AI systems and large language models.
Cloud providers are investing in specialized hardware to manage costs and improve performance.
TPUs have become a contributor to growth in Google Cloud revenue, as the company looks to show returns on its AI investments.
According to the report, the companies aim to finalize the design of the memory-focused chip next year before moving it into test production.
Reuters said it could not independently verify the report. Google and Marvell did not immediately respond to requests for comment.
