Anthropic is reportedly set to spend a staggering US$200 billion (A$310 billion) with Google over the next five years in one of the largest AI infrastructure deals ever disclosed.

According to multiple reports citing The Information, the agreement will give Anthropic – creator of the Claude AI models – access to massive amounts of Google Cloud infrastructure, including next-generation Tensor Processing Units (TPUs) developed with Broadcom.

The deal is expected to account for more than 40% of Google Cloud’s disclosed revenue backlog.

The agreement comes as AI developers scramble for compute capacity to train and run increasingly complex models.

Anthropic has already secured multi-billion-dollar infrastructure arrangements with Amazon, CoreWeave and Nvidia-backed providers as demand for Claude surges globally.

The scale of these contracts highlights how the AI boom is moving beyond software and into a full-scale infrastructure buildout spanning cloud data centres, custom silicon, memory and storage.

The spending spree has also triggered a sharp rally in semiconductor and cloud stocks. Alphabet shares rose following the reports, while chipmakers including Micron, Intel and Broadcom also climbed amid expectations AI infrastructure spending will continue accelerating through 2027.

Samsung this week also crossed a US$1 trillion market capitalisation, driven largely by demand for AI memory chips used in hyperscale data centres.

But questions remain about the sustainability of the AI spending cycle. Analysts warn that soaring data centre demand is intensifying pressure on electricity grids, chip supply chains and global memory markets, with shortages already pushing up prices for storage and RAM components.

Neither Google nor Anthropic has officially confirmed the US$200 billion figure.