AMD (NASDAQ:AMD) shares sank 9% by late morning on Tuesday, while Nvidia (NASDAQ:NVDA) slid 6.5%, after reports emerged that Meta Platforms (NASDAQ:META) may shift part of its future AI hardware spending toward Google’s custom processors — a move that could meaningfully disrupt the current GPU landscape.
The Information reported that Google (NASDAQ:GOOGL) has begun actively pitching its Tensor Processing Units (TPUs) to major enterprise customers, including Meta and several large financial institutions. Meta — one of Nvidia’s biggest GPU buyers — is said to be evaluating the possibility of investing billions of dollars into Google’s TPUs for deployment across its data centers starting in 2027, and could begin renting TPU capacity from Google Cloud as soon as next year.
Following the report, Google shares gained 3%, while Broadcom (NASDAQ:AVGO), which manufactures Google’s TPU hardware, traded flat after initial premarket strength.
The prospect of Meta leaning into Google-designed AI chips rattled the broader semiconductor sector. SoftBank Group dropped 10% on concerns that Google’s Gemini model could intensify rivalry with OpenAI, one of SoftBank’s key AI bets. Nvidia and AMD both supply chips for OpenAI.
Bernstein analyst Stacy Rasgon wrote that the developments appear “incrementally negative to the AMD narrative,” noting that the company has positioned itself as “the viable second source to NVDA, and hitched their wagon to OpenAI’s horse.”
Rasgon questioned whether investors will maintain confidence in AMD’s long-term goals if “Gemini hypothetically dominates over ChatGPT, and TPUs prove to be viable second sources for a broader swath of AI customers.”
Even so, Rasgon reiterated a preference for owning both Nvidia and Broadcom, writing that while “AVGO has the narrative at the moment,” Nvidia’s valuation “is looking increasingly attractive and the sell-offs seem overdone.” He added that he “would be a buyer of both stocks.”
The sharp market reaction highlights the escalating competition in the AI hardware race, as Nvidia’s long-held dominance is increasingly tested not only by AMD but also by hyperscalers developing their own specialized silicon.