Published • loading... • Updated
Google in Talks with Marvell to Build New AI Chips, The Information Reports
Google is adding a third design partner as it shifts TPU spending toward inference, a market projected to grow 45% in 2026, the report said.
- Google is in talks with Marvell Technology to develop two new chips for running AI models, according to The Information. One is a memory processing unit; the other is a new Tensor Processing Unit built specifically for inference.
- The strategy focuses on diversification rather than substitution, allowing Google to avoid depending on a single supplier. Adding Marvell to a supply chain that already includes Broadcom and MediaTek reduces pricing and supply risk.
- Inference has become the dominant compute cost as AI products reach hundreds of millions of users. Unlike training, which is a one-time event, inference runs continuously, scaling costs with user demand.
- For Marvell, a successful contract would validate its position as a top custom AI chip designer. Marvell's stock rallied approximately 50% year to date, with a 30% gain in April following the talks.
- The custom ASIC market is projected to grow 45% in 2026 and reach $118 billion by 2033. Google aims to build millions of Ironwood TPU units this year to meet demands in this expanding sector.
Insights by Ground AI
Podcasts & Opinions
39 Articles
39 Articles
Reposted by
The Economic Times
Google in talks with Marvell to build new AI chips: The Information
Alphabet's Google is reportedly in discussions with Marvell Technology to co-develop two new chips designed for more efficient AI model execution. One chip will be a memory processing unit to complement Google's TPUs, while the other is a new TPU specifically engineered for AI workloads.
·India
Read Full ArticleCoverage Details
Total News Sources39
Leaning Left5Leaning Right3Center9Last UpdatedBias Distribution53% Center
Bias Distribution
- 53% of the sources are Center
53% Center
L 29%
C 53%
R 18%
Factuality
To view factuality data please Upgrade to Premium


















