Nvidia Shares Drop Premarket After Report Says Meta May Adopt Google TPU Chips
Nvidia shares slide after a report suggests Meta may adopt Google TPUs for future data centers.
Shares of Nvidia Corp. plunged more than 4% in premarket trading on Tuesday, November 25, 2025, following a report that Meta Platforms Inc. is in advanced discussions to integrate Google's tensor processing units (TPUs) into its data centers starting in 2027. The news, first broken by The Information, triggered a sharp market reaction, with Nvidia's stock dipping as low as 4.1% premarket, potentially erasing around $180 billion in market value if the losses hold at the open. In contrast, Alphabet Inc., Google's parent company, surged over 6% ahead of the bell, building on a more than 6% rally from the previous session and pushing its valuation toward $4 trillion. The developments underscore growing investor concerns about Nvidia's dominance in the AI chip market amid intensifying competition from specialized alternatives.
Meta, one of the world's largest AI infrastructure spenders, is reportedly eyeing a multibillion-dollar deal to purchase Google's TPUs for on-premises deployment while potentially renting capacity from Google Cloud as early as 2026. This would mark Google's first major push to sell its custom AI chips directly to third-party data centers, moving beyond its traditional cloud rental model. Analysts view the potential partnership as a significant validation for TPUs, which are application-specific integrated circuits optimized for machine learning tasks. Seaport Global Securities analyst Jay Goldberg described a recent Google deal with Anthropic—supplying up to 1 million TPUs in a tens-of-billions-dollar agreement—as a "really powerful validation" of the technology, noting it has spurred broader industry interest in alternatives to Nvidia's graphics processing units (GPUs).
Google's TPUs, first introduced in 2018 for internal use, have evolved into advanced versions like the seventh-generation Ironwood, released this month, tailored for AI workloads such as model training and inference. Unlike Nvidia's versatile GPUs—originally designed for gaming and visual effects but repurposed for AI due to their parallel processing prowess—TPUs offer specialized efficiency for tensor operations central to neural networks. Meta's planned capital expenditures of at least $100 billion in 2026, including $40-50 billion on inference chips, highlight the scale of its AI ambitions and the strategic need to diversify suppliers. This comes as hyperscalers like Meta seek to mitigate risks from Nvidia's supply constraints and pricing power, fostering a multi-vendor ecosystem for AI hardware.
The episode signals early cracks in Nvidia's near-monopoly, which commands over 80% of the AI accelerator market, but experts caution that its ecosystem lock-in and near-term dominance remain intact. Broader ripples included gains for Broadcom Inc., a key TPU designer, up over 2% premarket after an 11% prior-day surge, while shares of Advanced Micro Devices Inc. slipped around 1.7%. Bloomberg Intelligence analysts Mandeep Singh and Robert Biggar noted that Meta's move, following Anthropic's, positions Google as a viable secondary supplier for inference needs. As the AI infrastructure boom accelerates—with global demand projected to drive trillions in spending—these shifts could reshape pricing dynamics and innovation, compelling Nvidia to accelerate its roadmap while validating custom silicon's role in sustainable AI scaling.
Also Read: Rajnath Singh: India Emerging as Preferred Hub for Complex Naval Ship Refits in Indo-Pacific
