Google has committed up to $40 billion in Anthropic, the largest single investment in the AI lab to date. The initial $10 billion goes in immediately at Anthropic's $380 billion pre-money valuation, with a further $30 billion contingent on undisclosed performance milestones. Alongside the equity commitment, the pair have a separate 5-year infrastructure agreement to bring 5GW of data centre capacity online starting next year, a deal potentially worth $200 billion in its own right.
Google also supplies Anthropic with TPU chips and Google Cloud services, making it simultaneously a financial backer, an infrastructure provider, and a competitor.
The full picture on Anthropic's capital stack
This deal follows Anthropic's agreement earlier this week with Amazon, $5 billion upfront, up to $20 billion over time, and a $100 billion compute procurement commitment. Taken together, Anthropic has now secured anchor investment and infrastructure from both of the two largest cloud providers.
Anthropic has locked in roughly 10GW of total compute capacity across Amazon, Google, and Broadcom, enough to sustain frontier model development at scale for the foreseeable future.
Why Google is funding a competitor
Google's own Gemini models compete directly with Claude. The investment is not a vote of confidence in Anthropic over its own products, it is a hedge against being locked out of the frontier AI supply chain entirely. If Anthropic becomes critical infrastructure for enterprise AI, Google wants to be embedded in it, not watching from the outside.
The circular structure holds here too. Google invests in Anthropic, Anthropic buys Google's chips and cloud services, Google collects on both sides of the relationship. OpenAI and Microsoft ran the same dynamic for years before OpenAI diversified toward Amazon.
Why it matters now
Anthropic's annualised revenue has grown from $9bn at end-2024 to over $30bn. The broader rollout of Claude Mythos, held back from general release due to cybersecurity concerns around its vulnerability exploitation capabilities, will push that figure higher still. For private market observers, the more significant signal is what this concentration of capital around a single AI lab implies for the rest of the field. The infrastructure relationships are being locked in now. Latecomers will pay a premium for access, or build from scratch.



