Meta Looks at Billions for Google’s Chips to Challenge Nvidia
Meta Platforms is engaged in discussions with Google regarding a potential investment of billions of dollars in chips produced by the Alphabet-owned company for deployment in its data centers beginning in 2027, as reported by The Information. This development positions Google as a formidable competitor to semiconductor leader Nvidia. The discussions reportedly include Meta renting chips from Google Cloud as soon as next year and are part of Google’s larger initiative to encourage customers to integrate its tensor processing units, which are utilized for AI workloads, into their own data centers, according to sources familiar with the negotiations. The move would signify a shift from Google’s existing strategy of utilizing TPUs solely within its own data centers and could significantly broaden the market for its chips, placing the company in direct competition for the hundreds of billions being invested in data-center processors to support AI services.
According to the report, some Google Cloud executives have indicated that the strategy could enable the company to capture as much as 10% of Nvidia’s annual revenue, a portion valued in the billions of dollars. Alphabet shares increased by over 4% in premarket trading on Tuesday, positioning the company to achieve a landmark $4 trillion valuation if these gains are sustained. Broadcom, which assists Google in the production of its AI chips, saw an increase of 2%, whereas Nvidia experienced a decline of 3.2%. Securing a chip agreement with Meta, a leading Nvidia client projected to spend as much as $72 billion this year, would represent a significant victory for Google. The company has already emerged as one of the foremost beneficiaries of the generative AI surge, driven by an increase in demand for its cloud services from businesses embracing this technology.
Alphabet, Meta, and Nvidia have yet to provide a response to requests for comment. Reuters was unable to confirm the report. In recent years, demand has surged for custom chips like TPUs as businesses seek alternatives to Nvidia’s expensive and supply-constrained graphics processors. Last month, Anthropic announced the expansion of its partnership with Google, aiming to utilize up to one million of the tech giant’s AI chips, a deal valued at tens of billions of dollars. In recent months, Google has gained significant traction by attracting Warren Buffett’s Berkshire Hathaway as an investor, transforming its previously marginal cloud unit into a robust growth engine, and receiving positive early feedback for its latest Gemini 3 model.
Renting Nvidia chips to customers serves as a significant revenue source for its cloud unit. Challenging Nvidia’s supremacy necessitates that Google surmount almost twenty years of exclusive Nvidia code, which has entrenched the company’s ecosystem and made it difficult to displace. Over 4 million developers across the globe depend on Nvidia’s CUDA software platform to create AI and various applications.
