Amazon plans to expand Anthropic stake beyond $8 billion

Amazon is in talks to extend its investment in Anthropic beyond the $8 billion already committed, according to executives and investors familiar with the discussions.

Deepening the alliance would keep Amazon among Anthropic’s top shareholders and reinforce its position against rivals such as Google—another backer of the San Francisco-based model developer—and Microsoft, which has tied up with OpenAI.

Meanwhile, Amazon Web Services is building “Project Rainier,” a 16-data-center cluster in Indiana expected to draw 2.2 gigawatts when complete. The complex will rely on Amazon’s Trainium 2 chips, tuned in collaboration with Anthropic’s engineers, to train Claude-series models.

Both companies portray the arrangement as mutually beneficial. Anthropic secures compute capacity and a sales channel into AWS customers, while Amazon gains a showcase tenant for its home-grown silicon and reinforces its strategy of supplying infrastructure rather than chasing consumer-facing chatbots.

Amazon, however, is also training internal foundation models and could one day compete directly with Anthropic. Nevertheless, executives on both sides say the partnership is structured to avoid the governance frictions that have recently strained Microsoft’s ties with OpenAI.