Key Moments
- Marvell Technology shares rose 6.3% in premarket trading by 04:38 ET (08:38 GMT) following a report of new AI chip talks with Alphabet’s Google.
- The reported collaboration focuses on developing a memory processing unit to pair with Google’s existing tensor processing units and a new TPU tailored for AI inference.
- The companies reportedly plan to lock in the memory chip design as early as next year before shifting into test production.
Premarket Reaction to Reported Chip Talks
Investing.com — Shares of Marvell Technology traded sharply higher in premarket action after The Information reported that Alphabet’s Google is negotiating with the chipmaker to co-develop two new semiconductors aimed at improving the efficiency of artificial intelligence model execution.
Marvell stock climbed 6.3% by 04:38 ET (08:38 GMT), reflecting investor enthusiasm about the potential expansion of the company’s role in the AI hardware ecosystem.
Details of the Proposed AI Chip Projects
According to the report, the discussions center on two distinct chip initiatives. The first is a memory processing unit that would operate in tandem with Google’s existing tensor processing unit (TPU) architecture. This component is intended to complement TPUs and support AI workloads more efficiently.
The second project reportedly involves a new generation of TPU designed specifically for AI inference, a phase of AI deployment focused on running trained models in real-world applications.
The report indicated that Google and Marvell are targeting completion of the memory chip’s design as soon as next year, after which the program would advance into test production.
| Aspect | Details |
|---|---|
| Parties involved | Alphabet’s Google and Marvell Technology |
| Number of chips | Two |
| Chip types | Memory processing unit; new TPU for AI inference |
| Timeline goal | Finalize memory chip design as soon as next year, then move to test production |
| Premarket move in Marvell stock | Up 6.3% by 04:38 ET (08:38 GMT) |
Google’s Broader TPU Strategy
The reported negotiations represent another step in Google’s effort to position its TPU technology as a viable option against Nvidia’s widely used graphics processing units. TPU sales have grown into a more meaningful driver for Google Cloud’s revenue as the company seeks to showcase financial returns on its AI-related investments.
Historically, Google largely reserved its TPUs for in-house workloads. That approach shifted in 2022 when the company’s cloud division successfully advocated for control of the team responsible for commercializing TPUs externally, which led to a significant increase in TPU allocation for outside customers.
As demand for AI compute capacity has expanded, Google has been working to ramp both manufacturing and external distribution of TPUs. Last year, the company also began offering TPUs directly into customers’ own data centers, rather than confining access exclusively to Google’s cloud platform.
TorchTPU and Software Ecosystem Integration
In a further move to broaden TPU adoption, Google earlier this month unveiled TorchTPU, described as an initiative to enable native compatibility between its chips and PyTorch, a leading AI software framework.
The project is designed to help developers move existing PyTorch-based workloads onto TPUs with minimal changes to their code. That objective targets a key obstacle that has hindered TPU uptake among users whose infrastructure is already built around PyTorch, and it adds to competitive pressure on Nvidia’s position in the AI computing space.
TradingPedia.com is a financial media specialized in providing daily news and education covering Forex, equities and commodities. Our academies for traders cover 
