Amazon Expands Anthropic AI Collaboration

Amazon and Anthropic said on June 20, 2024 they are widening their strategic collaboration on artificial intelligence, combining infrastructure commitments with new funding and product integration on Amazon Web Services (AWS). The companies framed the move as a step-up in joint work on AI development and the compute needed to train and run large-scale models.

At the center of the agreement is a long-term infrastructure commitment. Anthropic said it will allocate more than $100 billion to AWS technologies over the next ten years. The spending is set to cover both current and future versions of Amazon’s proprietary Trainium and Graviton chips, which Amazon positions as core components for AI training and processing workloads.

Anthropic also committed to securing up to 5 gigawatts (GW) of AWS capacity for training and operating its advanced AI models. The companies said a notable portion of Trainium3 capacity is expected to be available within the current year. Anthropic added that it plans to integrate Trainium2, Trainium3, Trainium4, and later generations of Trainium chips as they are released.

The partnership includes a geographic expansion tied to inference services. The companies said they will significantly broaden international inference capabilities into Asian and European markets. The stated goal is to support growing global demand for Anthropic’s Claude AI among customers outside the United States.

Amazon is also increasing its financial backing of Anthropic. The company said it is making an immediate investment of $5 billion. That amount could rise by an additional $20 billion if specific commercial milestones are met, according to the announcement. The new funding follows Amazon’s prior $8 billion investment in Anthropic.

For AWS customers, the companies said access will become more direct. AWS users are expected to get the full Anthropic-native Claude console through the AWS platform, which the companies said is intended to simplify how customers engage with the Claude Platform.

The announcement highlighted the scale of the AWS commitment and the potential size of Amazon’s total investment as factors that could influence competition in generative AI. The companies also noted that the size of the commitments could raise concerns about market concentration and potential regulatory scrutiny related to access to critical AI compute resources for smaller players.

They also pointed to potential downstream effects for developers and enterprises. By placing the Claude console more directly inside AWS, the companies said advanced AI tools could become easier to adopt across industries. They added that outcomes will depend on Anthropic’s ability to convert increased compute access into commercially viable models and on Amazon’s ability to deliver reliable, scalable infrastructure at the capacity levels described.