Bei Jing Shang Bao·2025-11-26 14:55Core Insights - Google's Gemini 3 series models are reshaping the AI landscape in Silicon Valley, challenging Nvidia's long-held dominance in the GPU market with the introduction of Google's self-developed TPU [1][3] - The stock performance of tech giants is being affected by this shift, with companies in the "Google chain" like Google and Broadcom seeing stock price increases, while the "OpenAI chain" represented by Nvidia, SoftBank, and Oracle is experiencing declines [1] Google TPU's Impact - Google's TPU technology is gaining traction, prompting Nvidia to publicly assert its continued industry leadership and capability to run all AI models across various computing scenarios [3] - The release of Gemini 3 has led industry experts to claim it surpasses OpenAI's GPT models in several aspects, utilizing TPU for training instead of Nvidia's GPUs [3][4] - The demand for ASIC chips, which are tailored for AI training and inference, is increasing, potentially eroding the market share traditionally held by GPUs [4] Stock Market Reactions - Following the release of Gemini 3, Nvidia's stock fell over 7% at one point, closing down 2.59%, leading to a significant valuation adjustment with its forward P/E ratio dropping from approximately 34 to 25 [5] - In contrast, Google's stock rose over 3% initially, closing up 1.53%, with Alphabet's market capitalization reaching $3.9 trillion, nearing the $4 trillion mark [6] Market Dynamics - Analysts predict that if TPU adoption continues successfully, Google could capture about 10% of Nvidia's annual revenue, translating to billions of dollars [6] - Meta, traditionally reliant on Nvidia GPUs, is reportedly negotiating with Google to use TPU chips in its data centers, indicating a potential shift in supplier dynamics [7] Competitive Landscape - The Gemini 3 model has received overwhelmingly positive reviews, with industry experts noting that Google has regained strategic control in the AI competition [8] - Google's vertical integration, including self-developed TPU chips and cloud services, positions it favorably against competitors who rely on rented computing power [9] - OpenAI acknowledges the strength of Google's new AI model, indicating a competitive pressure in the market [9] Future Trends - The ongoing demand for AI computing power suggests that tech giants will continue to develop in-house AI chips or support new suppliers, with Nvidia also adjusting its strategy by forming an ASIC division [10] - Nvidia's recent partnership with Microsoft and Anthropic highlights a strategic move to secure computing resources for AI model expansion [10]