Nvidia stock has struggled to maintain its recent momentum. The reason might be a familiar threat -- Alphabet's Google and its artificial-intelligence chips.
Nvidia shares were up 0.8% at $219.62 in premarket trading Wednesday. The chip maker has dropped back from the record highs of more than $240 it recorded earlier this month as investors gauge whether it can maintain its prime position in AI chips amid the growth of agents.
The biggest threat to Nvidia's dominance is Google and its Tensor Processing Units, designed in collaboration with Broadcom. The growth of TPUs is nothing new -- Barron's looked at the impact they were having on the market late last year. But now Google seems to be marketing them more fiercely as an alternative to Nvidia's hardware.
The first sign of a stronger commercialization push was the announcement on Tuesday that Google's TPUs would be made available through software-and-cloud computing company Oracle's cloud infrastructure platform from 2027. This is the first time TPUs will be offered on a so-called hyperscale cloud platform outside of Google's own.
"The next generation of AI will require a diverse range of compute," said Amin Vahdat, senior vice president of AI & Infrastructure at Google, in a statement Tuesday. "TPUs give customers a powerful, purpose-built option for demanding training and inference workloads."
Google making its TPUs available on a major external cloud platform is one thing, customers opting to use them is another. But there looks to be some news there as well -- ChatGPT developer OpenAI plans to buy TPUs, technology-focused news outlet The Information reported late Tuesday, citing people familiar with the matter.
OpenAI and Google didn't immediately respond to Barron's request for comment on the report. Google-parent Alphabet will post earnings after the market close on Wednesday.
In the absence of any financial details or quantities involved, it's hard to gauge whether there is any risk to Nvidia's own substantial business with OpenAI. But it's a reminder that big clients are still aiming to reduce their dependence on Nvidia's hardware and TPUs are the most popular alternative.
Comments