Such has been the buzz around Google’s TPUs over the last few days that the world’s most valuable company has had to issue a clarification.
NVIDIA has seemingly commented on reports that Meta was in talks with Google to use its TPUs for its datacenters. Thus far, Meta had been using NVIDIA’s GPUs for its AI efforts, but a partial switch to TPUs was seen to be a big win for Google, and a bit of a blow for NVIDIA. After the news broke, NVIDIA’s stock was down 5%, while Google’s was up 2%.
This seemed to have prompted NVIDIA to take to X to issue a statement. “We’re delighted by Google’s success — they’ve made great advances in AI and we continue to supply to Google,” NVIDIA’s Newsroom account posted on X. “NVIDIA is a generation ahead of the industry — it’s the only platform that runs every AI model and does it everywhere computing is done. NVIDIA offers greater performance, versatility, and fungibility than ASICs, which are designed for specific AI frameworks or functions,” it added.
The statement seemed to be a means to assuage fears in the industry that Google TPUs could take away marketshare from its own GPUs. NVIDIA’s GPUs have powered the AI revolution — it’s these chips that AI models are trained on, and on which AI companies run inference. But Google had set the cat among the pigeons last week by revealing that its latest Gemini 3 model — currently the most powerful AI model in the world — was trained on its own TPUs. Additionally, the news of Meta’s talks with Google on acquiring TPUs caused TPUs to be widely discussed on Wall Street, and led to the ensuing price movements in Google and NVIDIA stocks.
Now Google’s TPUs aren’t new. They’ve been around for more than a decade, but after a deal with Anthropic, and talks with Meta, there seems to be renewed interest in these chips. But NVIDIA’s statement is receiving some flak on X — some users have said that the statement shows that NVIDIA is worried about TPUs taking away its market share, and others have said that that a statement like this about a competitor could’ve been avoided. But while it remains to be seen if TPUs can actually compete with GPUs, they do seem to have made the industry — and NVIDIA itself — sit up and take notice.