Google and Meta join forces to disrupt the global AI chip market landscape

The chip rental agreement reached this time focuses on the commercial application of Google TPU and Meta's AI model research and development needs. As a dedicated AI chip developed by Google, TPU has always supported the development of Google's AI business with its core advantages of efficient execution of AI matrix operations and high cost-effectiveness since its internal deployment in 2015. From AlphaGo games to Gemini series model training, TPU's computing power support is indispensable. As a core player in global AI research and development, Meta has continuously increased its layout in fields such as big models and metaverse in recent years, and its demand for high-performance and low-cost AI computing power is becoming increasingly urgent. Choosing to rent Google TPU this time is precisely to match its core demand for new AI model research and development.

In fact, the cooperation between the two parties is not accidental, but an inevitable result of the development of the AI industry and the upgrading of market demand. For a long time, the global AI chip market has shown a pattern of "one dominant player". NVIDIA, with the versatility of its GPU products and mature CUDA software ecosystem, has occupied over 90% of the AI acceleration market share, almost becoming the "standard" choice for all technology companies to develop AI models. But with the explosive growth of AI large-scale models, the cost and power bottlenecks of NVIDIA GPUs are gradually becoming apparent. According to industry estimates, the cost of using NVIDIA GPUs for model training is almost five times that of Google TPUs, and TPU has a significant advantage in energy efficiency. When performing the same AI operations, its power consumption is only half or even one-third of that of GPUs. For tech giants like Meta, which have billions of users and huge computing power demands, choosing TPU is undoubtedly the optimal solution that balances cost control and research and development efficiency.

From the details of the transaction, the core value of this agreement lies not only in the 'lease' itself, but also in the deep collaboration between the two tech giants in the AI industry chain. Meta will leverage the computing power advantage of Google TPU to accelerate the development and iteration of new AI models, especially in the field of large model training and inference. The clustering advantage of TPU will be fully utilized - Google's latest generation TPU single cluster supports up to 9216 chips interconnected, far exceeding the scale of mainstream GPU systems, and can better adapt to the training needs of large-scale AI models. Through this transaction, Google officially promotes the transformation of TPU from "self use" to "commercial leasing", further expanding the market influence of its AI chips, breaking the previous supply model that only relied on its own cloud services, and gradually building a commercial path to compete with Nvidia.

The implementation of this transaction has quickly attracted widespread attention from global capital markets and industries, and its impact on the AI chip market is particularly evident. As early as November 2025, when the news of the two sides negotiating cooperation came out, it directly led to a 3% drop in Nvidia's stock price, while Google's parent company Alphabet rose 1.6% against the trend, with its market value approaching the $4 trillion mark. Industry insiders analyze that Meta's choice of Google TPU is essentially to diversify its dependence on Nvidia and ensure the stability of its AI research and development through diversified computing power supply. This trend may drive more technology giants to follow suit and break Nvidia's market monopoly pattern. It is worth noting that Nvidia has previously responded by stating that its products are one generation ahead of the industry and are the only platform capable of running all AI models and adapting to all computing environments, and reiterated that it will continue to provide technical support to Google, highlighting the intense competition in the industry.