Amazon is ratcheting up the AI arms race, committing USD5 billion into Anthropic immediately and reserving a further USD20 billion for potential future investment. In exchange, Anthropic is pledging to route over USD100 billion of infrastructure spending through Amazon Web Services during the next decade. That commitment anchors the Claude developer’s core compute needs to AWS, turning a partnership into a long-term commercial dependency.
Under the expanded deal, Anthropic secures access to as much as 5 gigawatts of current and upcoming generations of Amazon’s Trainium chips to train and run its advanced AI models. The AI lab’s USD100 billion infrastructure tab will stretch across Trainium and tens of millions of Amazon Graviton CPU chips, according to Amazon. The new commitment builds on USD8 billion Amazon has already invested in Anthropic and shows how central the relationship has become to both companies’ generative AI strategies.
Amazon and Anthropic have been tied together since 2023, with more than 100,000 customers already deploying Claude models on AWS. The pair are also co-developing Project Rainier, described as one of the world’s largest AI compute clusters and a key backbone for scaling Claude.
As part of the new phase, they plan a major expansion of inference capacity across Asia and Europe to better support Claude’s growing global user base and deepen AWS’s footprint in those regions.
