Nvidia panics? Rare statement triggers selloff as Meta turns to Google’s TPUs
Nvidia (NVDA) broke with its usual practice on Tuesday by issuing a rare reactive statement following news that Meta Platforms is using Google’s TPU chips rather than relying solely on Nvidia GPUs. The market’s reaction was swift and negative.
In its statement, Nvidia said it is “delighted by Google’s success” and emphasized that it continues “to supply Google.”
The company also sought to reassure investors, noting that it remains “a generation ahead of the industry” — a reminder that its GPU platform is still widely viewed as the standard for running AI models.
The move comes as Meta’s adoption of TPUs raises questions about whether large tech companies might eventually reduce their dependence on Nvidia’s hardware.
Market commentator The Kobeissi Letter called attention to the unusual nature of the response, writing that “the AI wars are heating up.”
Despite Nvidia’s attempt to stabilize sentiment and avoid a broader narrative shift, investors were unconvinced. NVDA shares fell roughly 6% in after-hours trading on Tuesday following the statement.
The development also raised the prospect of a deeper pullback for NVDA shares, which are already down nearly 15% from last month’s peak.
How serious is the Google threat?
Google’s TPUs do compete with Nvidia GPUs in certain segments of the AI compute market, particularly in large-scale model training and inference inside hyperscale data centers.
Several observers have highlighted growing momentum behind TPUs. Rohan Paul, who hosts one of the largest AI shows on X, said the chips are “on a serious winning streak,” citing enterprise adopters including Salesforce, Safe Superintelligence, and Midjourney.
Google’s hardware also gained mainstream attention in October when Anthropic reached a long-term agreement to use Google Cloud and TPUs to train future versions of its Claude models.
The deal signaled that leading AI labs are willing to diversify away from Nvidia’s GPUs when alternative compute platforms offer competitive performance and deeper cloud partnerships.
Seaport analyst Jay Goldberg said Anthropic’s decision represents “a really powerful validation of TPUs,” adding that the move will likely encourage more companies to consider adopting Google’s chip platform.
The key caveat is that TPUs do not compete across all of Nvidia’s product lines. Nvidia’s platform supports general-purpose AI compute, while TPUs are available only through Google Cloud and cannot be purchased as standalone hardware.
Nvidia, by contrast, maintains a dominant position in enterprise servers, laptops, robotics, and automotive systems.
Nvidia remains by far the most influential company in the AI sector, reporting record demand and a multiyear pipeline of commitments from enterprise customers in its latest financial filings. Industry estimates suggest that combined commitments for 2025 and 2026 could reach $500 billion.