Anthropic has agreed to spend US$200 billion on Google Cloud over five years, according to Reuters. The commitment represents more than 40% of Alphabet’s disclosed cloud revenue backlog, and Alphabet shares rose about 2% in extended trading following the announcement.
The five-year commitment underscores Anthropic’s escalating demand for computing capacity. Alphabet is separately investing up to US$40 billion in the startup, strengthening the relationship between the cloud provider and the AI firm.
Anthropric also signed an April agreement with Google and Broadcom for multiple gigawatts of tensor processing unit (TPU) capacity starting in 2027. This arrangement complements the broader Google Cloud commitment and signals long-term infrastructure planning.
Despite the Google commitment, Anthropic maintains a diversified cloud approach. Amazon remains Anthropic’s main cloud provider and training partner. The startup spreads Claude across Google’s Tensor Processing Units (TPUs), Amazon’s Trainium AI chips, and NVIDIA’s graphics processing units (GPUs), allocating each workload to the hardware best suited for the task.
Anthropric’s computing needs are rising alongside its business growth. The company’s run-rate revenue increased from approximately US$9 billion at the end of 2025 to more than US$30 billion in 2026.
The Anthropic deal reflects intensifying competition for AI infrastructure. Large cloud companies including Microsoft, Amazon, Google, and Oracle plan to spend more than US$320 billion on capital expenditure in fiscal 2025. This spending is tied to substantial contracted future revenue: Microsoft has US$627 billion in commercial remaining performance obligations (RPO), while Alphabet lists a backlog of approximately US$462 billion.
The scale of required investment creates significant barriers to entry. Only the largest companies can fund the infrastructure AI firms require, typically binding them to extended contracts. Market research projects the AI infrastructure market will reach US$947.46 billion by 2035.