Key Highlights
- Shares of Marvell Technology rose 6.3% during premarket hours following reports of partnership negotiations with Google for AI chip development.
- The collaboration involves two distinct chip projects: a memory processing unit to complement Google’s TPU architecture and a dedicated TPU for inference applications.
- Google plans to complete the memory chip architecture by next year before proceeding to prototype testing phases.
- This partnership supports Google’s strategic objective to establish TPUs as competitive alternatives to Nvidia’s GPU offerings.
- Google’s first-quarter 2025 financial results, scheduled for April 29, could provide additional insights into AI chip infrastructure spending.
Marvell Technology (MRVL) experienced a significant premarket rally on Sunday following a report from The Information revealing that Alphabet’s Google has entered discussions with the semiconductor company to jointly develop two advanced AI processing chips.
Marvell Technology, Inc., MRVL
Shares climbed 6.3% around 4:38 AM ET, delivering an encouraging signal to investors heading into the trading week.
Based on information from two sources with direct knowledge of the discussions, the companies are developing a memory processing unit (MPU) engineered to function in tandem with Google’s established tensor processing unit (TPU) infrastructure. The partnership’s second focus involves creating an entirely new TPU optimized for AI inference operations.
The timeline calls for finalizing the memory chip architecture within the next year, paving the way for test manufacturing runs.
Google Expands Its Semiconductor Strategy
Google has been methodically constructing a comprehensive chip development network, collaborating with industry players such as Intel and Broadcom in addition to Marvell.
Throughout most of its existence, Google maintained TPU technology for internal applications exclusively. This approach transformed in 2022 when the cloud business unit assumed responsibility for external chip distribution and began marketing TPUs to third-party clients.
Following that transition, Google has accelerated both manufacturing capacity and commercial sales. Last year marked another milestone when the company introduced direct TPU sales into customers’ private data center facilities — extending beyond cloud-only distribution. This represents a substantial evolution in commercial strategy.
Earlier this month, Google unveiled TorchTPU, an initiative designed to enable native compatibility between its processors and PyTorch — the prevailing AI development framework. This integration reduces migration barriers for engineers who have constructed their systems around PyTorch and are evaluating alternatives to Nvidia’s ecosystem.
TPU revenue has emerged as an increasingly significant component of Google Cloud’s financial performance as the organization works to demonstrate that AI infrastructure investments are yielding tangible returns.
The Nvidia Competitive Landscape
Nvidia maintains its leadership position in AI computing infrastructure, yet Google’s strategic initiatives are intensifying competitive dynamics.
The Marvell collaboration strengthens Google’s capabilities in the inference chip segment — an area where Nvidia has also been pursuing aggressive expansion. Nvidia is developing new AI inference processors incorporating technology from Groq, according to reports.
With Google, Marvell, Intel, and Broadcom all advancing similar initiatives, competition in the inference chip sector is intensifying rapidly.
Google’s quarterly financial report arrives on April 29. Market observers will scrutinize the disclosure for indications regarding TPU production scaling plans, cloud division revenue trajectory, and strategic implications of the Marvell collaboration for future chip development initiatives.
