This article first appeared on GuruFocus.
Nvidia (NVDA, Financials) came under pressure after reports that Meta Platforms is considering adopting Google’s custom-designed tensor processing units for its data centers. The potential move, reported by The Information and cited by Investing.com, marks a possible shift in Meta’s long-term infrastructure strategy, which has historically relied heavily on Nvidia hardware.
Meta is evaluating whether to use Google’s TPUs beginning in 2027 and may rent TPU capacity through Google Cloud as early as next year. The report triggered a selloff in Nvidia shares, which fell about 4% in premarket trading, placing the company on pace to lose approximately $180 billion in market capitalization.
The development lifted Alphabet’s stock, which rose around 4% and moved closer to a $4 trillion valuation. Broadcom, a manufacturer involved in producing Google’s AI chips, also traded higher amid expectations of increased demand tied to Google’s expanding AI ecosystem.
Analysts said a deal with Meta would mark a significant endorsement for Google’s chip strategy and represent a rare example of a major AI platform considering an alternative to Nvidia’s GPUs. Google’s TPUs, first introduced in 2018, are designed to manage the compute-intensive workloads of large AI models and have gained traction as businesses seek cost-effective options amid constrained supply of Nvidia processors.
