Google plans up to $40 billion investment in Anthropic to secure compute capacity

Google is planning to invest up to $40 billion in Anthropic, the AI safety-focused startup, according to reporting from Bloomberg. The investment is structured in two phases: Alphabet will commit $10 billion immediately at a $350 billion company valuation, with an additional $30 billion contingent on Anthropic achieving unspecified performance milestones.

Beyond cash, Google is providing 5 gigawatts of computing capacity through Google Cloud over the next five years. This builds on an announcement earlier this month where Anthropic, Google, and chipmaker Broadcom unveiled a partnership granting access to 3.5 gigawatts of TPU-based compute beginning in 2027. The new arrangement expands that capacity pool and locks in additional resources as Anthropic scales its operations.

Anthropic is pursuing a multi-pronged strategy to secure compute infrastructure as demand for AI training and inference accelerates. The company received a separate $5 billion investment from Amazon tied to a broader commitment under which Anthropic will spend up to $100 billion for approximately 5 gigawatts of compute over time. CoreWeave, a data center provider, also inked a deal with Anthropic this month to supply additional capacity.

The compute race reflects structural competition in large-scale AI development. OpenAI has pursued aggressive deals with cloud providers, chip suppliers, and energy firms to underpin its expansion. Anthropic's latest deal underscores the shift: capital deployment now centers on securing the infrastructure—chips, power, data centers—needed to train and run powerful models rather than on model development alone.