Marvell (MRVL) Stock Soars on Report of Google AI Chip Partnership

TLDR

• Marvell stock saw a significant increase, climbing as much as 6.3% in premarket trading following a report indicating discussions with Google for the development of two new AI chips.
• One of the proposed chips is a memory processing unit intended to complement Google’s existing tensor processing unit (TPU), while the other is a novel TPU designed specifically for AI inference tasks.
• Google’s objective is to finalize the design of the memory chip by next year, with plans to proceed to test production thereafter.
• This collaboration is a component of Google’s broader strategy to establish its TPUs as a competitive alternative to Nvidia’s GPUs.
• Further insights into Google’s AI chip investment strategies may be revealed when the company releases its Q1 2025 earnings on April 29.

(SeaPRwire) –   Marvell Technology (MRVL) experienced a surge in premarket trading on Sunday after The Information published a report stating that Alphabet’s Google is in discussions with the chip manufacturer to jointly develop two new AI chips.

Marvell Technology, Inc., MRVL

The stock climbed by 6.3% around 4:38 AM ET, providing an early boost to traders at the start of the week.

According to the report, individuals with knowledge of the matter indicated that the companies are collaborating on a memory processing unit (MPU) designed to work in conjunction with Google’s current tensor processing unit (TPU). The second chip is a completely new TPU engineered for AI inference workloads.

The aim is to have the memory chip’s design finalized as early as next year, after which the companies will move towards test production.

Google’s Chip Ambitions Take Shape

Google’s efforts in this area are not isolated. The company has been systematically building its semiconductor ecosystem, engaging with partners such as Intel and Broadcom, in addition to Marvell.

Historically, Google primarily utilized its TPUs internally. This approach shifted in 2022 when its cloud division began handling external chip sales and promoting TPUs to external clients.

Since then, Google has significantly increased both production and external sales. Last year, it began offering TPUs directly for deployment in customers’ own data centers, extending beyond its cloud platform. This represents a notable change in its market strategy.

Earlier this month, Google officially introduced TorchTPU, an initiative to ensure its chips are natively compatible with PyTorch, the leading AI software framework. This move is intended to reduce the barrier for developers who have already established their workflows with PyTorch and are considering migrating away from Nvidia’s hardware.

TPU sales have become an increasingly important contributor to Google Cloud’s revenue, as the company seeks to demonstrate to investors that its AI investments are yielding positive returns.

Where Nvidia Fits In

While Nvidia continues to dominate the AI compute market, Google’s recent actions are intensifying the pressure on its leading position.

The partnership with Marvell enhances Google’s capabilities in the inference chip sector, a market where Nvidia has also been aggressively expanding. Reports suggest Nvidia is developing new AI inference chips utilizing technology from Groq.

With Google, Marvell, Intel, and Broadcom all pursuing similar objectives, the inference chip market is rapidly becoming more competitive.

Google’s first-quarter earnings are scheduled for release on April 29. Investors will likely be looking for indications of the company’s plans for scaling TPU production, its cloud revenue growth, and the implications of the Marvell discussions for its future chip development roadmap.