The rise of Google's TPU chips: Does Nvidia need to worry in the short term?

AI chip market’s “Iron Throne” appears to be showing signs of wavering, as rumors that a key customer might “defect” have prompted a renewed scrutiny of Nvidia’s future. On November 25, news broke that Nvidia’s major client Meta is considering large-scale adoption of Google’s self-developed AI chips—Tensor Processing Units (TPUs)—in its data centers, and may begin leasing them as early as next year. This news hit the market like a bombshell and immediately ignited a frenzy. During the day’s trading, Nvidia’s share price plunged by up to 6%, while rival AMD dropped 10%. At the same time, Google’s parent company Alphabet saw its shares surge by as much as 4%. By the close, Nvidia’s shares were down by about 2.6%, while Alphabet rose by 1.6%, setting consecutive record highs. In response to market turmoil, both Google and Nvidia reacted swiftly. An Alphabet spokesperson said, "Google Cloud has observed that demand for our custom TPUs and Nvidia GPUs are both rapidly growing; we remain committed to supporting both products as we have for many years.” Meanwhile, a Nvidia spokesperson stated, “We’re happy for Google’s success… Nvidia leads the industry by a generation.” They also added that they believe their chips outperform ASIC (application-specific integrated circuit) chips, including the TPU. Citi: Competition intensifies, but Nvidia’s moat remains Amidst the commotion, a report from Citi Research released on November 25 offers a relatively calm perspective. The report acknowledges that custom AI accelerators like Google’s TPU and Amazon’s Trainium are continuously expanding and are bringing cost advantages to these platforms’ AI infrastructure. However, analysts including Atif Malik point out that since custom chip projects by hyperscale companies like Microsoft and Meta “appear to be facing delays,” they currently remain heavily reliant on Nvidia's platform. Citi maintains its “Buy” rating for Nvidia, with a price target of $270. The firm expects commercial AI GPU market share to remain high despite competition, only declining slightly from a predicted 90% in 2025 to 81% in 2028. In summary, while the rise of Google’s TPU and diversification of procurement strategies among top clients undoubtedly brings real competitive pressure to Nvidia, Citi’s analysis suggests that its short-term market dominance and technical moat remain solid. Still, Nvidia’s unusually defensive communication style does reflect that the throne of the AI king is not as secure as before. Tech giants fight back: Public messages and private memos Facing pressure on its stock price and market doubts, Nvidia has taken some rare defensive communication measures. A Wallstreetcn article wrote that on November 25, Nvidia publicly declared on social media platform X: “Nvidia is currently a generation ahead of the industry—we’re the only platform that can run all AI models and is general across all computing scenarios.” The company emphasizes that its GPUs offer “higher performance, versatility and interchangeability” compared to ASIC chips like Google’s TPU. Before this, Nvidia had taken more discreet actions. According to reports from The Wall Street Journal and other media outlets, over the weekend Nvidia distributed a seven-page memo to Wall Street analysts, aiming to “refute point by point” accusations from critics—including Michael Burry, the real-life inspiration for the movie The Big Short—about accounting fraud, circular financing, and an AI bubble. The memo clarified: - **Accounting practices:** Nvidia emphasized its business is financially sound, its reporting is full and transparent, and not comparable to any historical accounting fraud cases. - **Equipment depreciation:** Addressing accusations that AI chips last only 2-3 years, Nvidia responded that its customers typically set GPU depreciation periods at 4 to 6 years, in line with the actual usable life of the equipment. - **Circular financing:** To this point, Nvidia clarified that its strategic investments in Q3 amounted to just $3.7 billion, a small fraction of its revenue, stating that circular financing claims are “baseless.” Counterproductive? Defensive posture interpreted as “lack of confidence” Despite these emergency communications, Nvidia’s efforts appear to have failed to fully quell market concerns, and instead have sparked new interpretations. D.A. Davidson analyst Gil Luria commented to the media regarding the memo, “The memo itself makes Nvidia look passive… A company of this size doesn’t need to respond to every single question outside of earnings.” The analyst believes that a confident leader usually speaks through performance and products, and that Nvidia’s “lack of poise” in these moves “hits at existing fears.” Synovus Trust’s senior portfolio manager Dan Morgan also commented: “Google does indeed possess considerable strength—they are not a minor player standing on the sidelines.” He noted that Google’s most powerful AI model, Gemini 3, was trained entirely on its own TPUs, which increases the credibility of the TPU as a viable alternative to Nvidia’s products. Risk Warning and Disclaimer The market contains risks; investors should be cautious. This article does not constitute personal investment advice, nor does it consider the particular investment goals, financial status, or needs of any individual user. Users should assess whether any views, opinions, or conclusions in this article are suited to their own circumstances. Investment is at your own risk.