Anthropic says it will spend over $100B on AWS over 10 years

In short: Anthropic says it will spend more than $100 billion on Amazon Web Services, or AWS, over the next decade to build and run future Claude AI models.

What happened

Anthropic and Amazon expanded their partnership. Anthropic committed to spending over $100 billion on AWS technologies over the next 10 years.

A big part of the deal is access to large amounts of computing. Anthropic says it secured up to 5 gigawatts of computing capacity on AWS to train and power advanced Claude models. That is roughly comparable to the power output of about five large nuclear power plants, which helps show the scale.

Anthropic plans to use Amazon’s in-house chips, including Trainium (specialized chips designed for training AI, like purpose-built engines) across multiple future versions, and tens of millions of Graviton CPU cores (general-purpose computing brains that run many everyday tasks). Anthropic expects close to 1 gigawatt of Trainium2 and Trainium3 capacity to be online by the end of 2026.

The expanded relationship also includes more investment from Amazon. Amazon will invest $5 billion in Anthropic immediately, with up to another $20 billion tied to business milestones, on top of the $8 billion Amazon previously invested.

Anthropic said the deal is driven by demand. More than 100,000 customers already run Anthropic models on AWS through Amazon Bedrock (a service that lets companies use AI models). Anthropic also said the full Claude Platform will be available directly within AWS, and that it will expand capacity in Asia and Europe to serve more users.

Why it matters

For regular people, this mostly affects reliability and speed. If more companies build customer support, search, writing, and other services on Claude, they need enough computing power to keep those tools working smoothly at busy times.

Source: NYTimes