Alphabet, Google's Parent Company, Plans to Invest Up to $40 Billion in Claude AI Developer Anthropic
Multiple English media outlets cite market sources stating that Alphabet, Google's parent company, plans to invest up to $40 billion in Claude AI developer Anthropic: an initial investment of $10 billion in cash at a valuation of approximately $35 billion, with a potential additional investment of up to $30 billion contingent on performance and collaboration conditions, aimed at significantly expanding the computing power and infrastructure behind the Claude model. This investment, combined with approximately $3 billion previously invested in equity and cloud/TPU collaborations worth several billion dollars, positions Anthropic as one of Google Cloud's most critical external partners in the AI model layer. In the coming years, Google will provide computing power equivalent to several gigawatts, even up to 3.5 gigawatts, and will work with Broadcom to customize the next generation of TPU chips.
Meanwhile, Anthropic recently disclosed that its annualized revenue has exceeded $30 billion and has opened up explosive demand on the enterprise side through products like Claude Code and other "programming agents." Google and Amazon have each committed to multiple rounds of investments worth billions of dollars and additional options, creating a tripartite balance in Anthropic's model, cloud resources, and capital structure: it is a heavy customer of Google TPU and cloud while maintaining a multi-cloud layout on AWS and NVIDIA GPUs to reduce dependence on a single infrastructure.
Source: Public Information
ABAB AI Insight
This "up to $40 billion" arrangement is essentially not just a financial investment but a deeply binding dual cooperation framework of computing power and applications: Google bundles cash with large-scale TPU resources into a whole, locking in Anthropic's capital structure, technology roadmap, and cloud deployment for the long term. For Anthropic, this secures "sovereign-level" computing power guarantees during a high-intensity capital expenditure cycle; for Google, it firmly ties the leading model Claude to its own infrastructure through equity and cloud contracts, preventing it from fully leaning towards AWS or a pure NVIDIA stack.
From an industry structure perspective, this effectively establishes a new AI alliance chain of "Google-Anthropic-Broadcom" outside the OpenAI-Microsoft axis: Microsoft binds with OpenAI and NVIDIA, while Google attempts to create an alternative computing power-model closed loop through TPU and Broadcom. In this landscape, top-tier cutting-edge models are no longer just software assets but "computing power engines" driving trillion-dollar CAPEX in chips, data centers, and cloud services; whoever secures long-term contracts for leading models will have more pricing and bargaining power over the next generation of data center architecture.