New competition for Nvidia in AI chips as Meta tests Google’s TPUs
Nvidia’s shares dropped 4.56% in pre-market trading on Tuesday, following a report suggesting that Meta may switch to using artificial intelligence (AI) chips from Google.
According to The Information, Meta is exploring the use of Google’s tensor processing units (TPUs) in its data centres starting in 2027. Moreover, Meta is reportedly considering leasing TPUs via Google Cloud as early as next year.
Google has confirmed surging demand for its custom TPUs and Nvidia GPUs, emphasising its commitment to supporting both technologies. The TPUs were originally designed in 2018 for internal use within Google’s cloud operations, but have since evolved into more advanced versions capable of handling complex AI workloads.
Experts note that the customised nature of TPUs is a significant advantage for Google, as it allows them to offer clients highly efficient solutions for AI applications.
If Meta starts using TPUs, it would mark a major win for Google and validate the company’s technology. Despite this development, Nvidia remains the dominant force in the market, with its GPUs serving as the cornerstone of large-scale AI infrastructure.
While it is unlikely that Nvidia’s position will be threatened in the short term, the rise of Google’s TPUs is introducing new competition into the AI semiconductor landscape.
Companies investing heavily in AI infrastructure are actively seeking alternative chip sources to reduce their dependence on Nvidia. Meta, one of the leading investors in this space, expects capital expenditures this year of between $70 billion (about €60.5 billion) and $72 billion (€62.3 billion).
The share price fluctuations coincide with ongoing discussions about a possible ‘AI bubble’ and inflated valuations in the tech sector. Nvidia, which sits at the heart of this debate, recently exceeded expectations with its revenue forecasts for the current quarter, but tech stocks subsequently declined.
Business AM
Subscribers 0
Fans 0
Followers 0
Followers