Google parent Alphabet is reportedly in discussions with Marvell Technologies to develop new versions of its artificial intelligence chips. The planned partnership, as detailed by sources including The Information and Funda AI, aims to enhance AI “inferencing,” a process that focuses on executing workloads rather than training AI models. Marvell is expected to create a specialized AI memory chip that will integrate seamlessly with Google’s processors.
Google’s AI accelerators, known as TPUs, have emerged as competitors to Nvidia’s dominant position in the market. In parallel, Nvidia is working on its own AI inferencing chips, leveraging technology from Groq. Google has an existing partnership with Broadcom, which is set to run through 2031, for the development of future generations of TPUs.
Both Broadcom and Marvell are key players in providing custom chip design services, a demand that has surged with the growing adoption of AI tools in advanced data centers. Reports suggest that Google may seek to diversify its supply chain away from Broadcom in light of increasing demand for its chips. Companies are actively looking for alternatives to Nvidia’s high-cost offerings. Notably, Anthropic utilizes a mix of chips, including Google’s TPUs, for its AI software and chatbot, Claude.
“It should be no surprise that rivals of Nvidia will want to grab a piece of the market and the apparent growth on offer by developing their own product,” said Russ Mould, investment director at AJ Bell. “It also makes sense for customers to diversify their sources of supply, if they can, so they can spread technological and supply chain risk.”
In a related development, Meta recently extended its agreement with Broadcom to produce several generations of custom AI processors, having previously paid the firm $2.3 billion for AI chip design and associated services last year.
Marvell’s stock has seen a remarkable increase, gaining approximately 64% year-to-date after a prior decline of around 23% in 2025. This uptick is underpinned by Nvidia’s recent $2 billion investment in Marvell, aimed at facilitating the use of custom AI chips alongside Nvidia’s networking technologies and central processors. Marvell anticipates its revenue to approach $15 billion by fiscal 2028, with an expected market value increase of over $6 billion, pending current gains. As reported by Reuters, Marvell currently trades at 33.35 times its estimated earnings for the next 12 months, compared to 27.84 for Broadcom.
The ongoing rivalry in the AI chip sector highlights the accelerating shift towards specialized processors as businesses increasingly adopt AI technologies. With the market landscape evolving rapidly, partnerships like that between Google and Marvell could play a pivotal role in reshaping the industry’s competitive dynamics.
See also
Alphabet Advances Talks with Marvell to Co-Develop Custom AI Chips for Cloud Services
a16z Reveals 5 Ways Blockchain Enhances AI Agent Infrastructure and Trust
Victory Giant Technology Surges 60% in HK Debut, Raising $2.2 Billion for Expansion
Intel Launches AI-Enhanced Laptops in Malaysia with Core Ultra Series 3 Processors
Artificial Intelligence Market to Surge from $185B in 2026 to $744B by 2032 at 26.1% CAGR
