Strong Demand for AI Infrastructure: Google, Amazon, and Meta Each Secure Their Position

中美三大出行平台携手,在伦敦启动无人驾驶出租车路测
Published on: Apr 13, 2026
Author: Amy Liu

Recent data indicates that the competitive landscape of the AI chatbot market is changing. According to data from BNP Paribas, Anthropic’s Claude saw its average daily user rate more than double from February to March, while Google’s (GOOGL) Gemini also experienced a continuous increase in usage. Specifically, Gemini’s share of website visits and its share of monthly average daily active users in March rose to 28.0% and 16.5%, respectively, up from 26.2% and 15.1% in February. Claude’s two metrics also surged in March, rising to 6.6% and 1.8%, respectively, from 3.6% and 0.8% in February. Analysts led by Nick Jones at BNP Paribas pointed out that although ChatGPT remains the leader in this field, it continued to lose share in website traffic and mobile applications in March.

Google and Amazon: Deep Layout from Chips to Computing Power

Google not only holds the second position in the chatbot market but has also made substantial progress in the AI computing field. Jones stated that Google’s agreement with Broadcom (AVGO) to build 3.5 gigawatts of AI computing capacity for Anthropic highlights the strong and sustained demand for AI infrastructure, reinforces Google’s strong positioning, and could add tens of billions of dollars to its backlog of orders. Meanwhile, Amazon (AMZN) has also secured a favorable position in the AI market. Last week, Uber (UBER) announced it would expand its use of Amazon’s Graviton4 and Trainium3 processors. Jones added that Uber’s use of Amazon’s custom chips strengthens AWS’s AI value proposition, supporting incremental demand and monetization. Previously, Amazon’s CEO revealed that AWS’s AI-related annual recurring revenue had reached $15 billion, and its chip business’s annual recurring revenue had reached $20 billion.

Meta and Alphabet: Model-Driven Strategy and Chip Strategy Transformation

Furthermore, the Muse Spark AI model released by Meta (META) has driven a surge in downloads of its AI applications. Jones believes the release of this model indicates that Meta’s efforts in the field of super-intelligence are making progress, helping to justify its massive spending in the AI space. Another significant development comes from Alphabet. The company announced an expansion of its partnership with Anthropic, allowing it to deploy more of its self-developed Tensor Processing Units. Previously, Anthropic had agreed to purchase $21 billion worth of TPUs from Broadcom, while the new agreement involves 3.5 gigawatts of TPU computing capacity to be delivered starting in 2027. Having Broadcom sell complete TPU racks directly to Anthropic so that the latter can manage its own servers marks a major strategic shift for Alphabet. Morgan Stanley estimates that for every 500,000 TPUs Alphabet sells, it would generate approximately $13 billion in additional revenue and $0.40 in earnings per share. Analysts believe that as a company possessing both a top-tier foundational AI model and its own leading AI chips, Alphabet’s shift toward selling TPUs to customers will create a substantial revenue stream, leveraging the cost advantages of its proprietary chips and its powerful distribution network.

AI Financial Service Growth Stocks Technology