On April 21, 2026, Amazon.com Inc. and artificial intelligence safety and research company Anthropic announced a major expansion of their strategic partnership, centered on a $100 billion cloud computing commitment. Under the terms of the ten-year agreement, Anthropic will utilize Amazon Web Services (AWS) as its primary cloud provider for the majority of its workloads, including the training and deployment of its most advanced foundation models. This deal represents one of the largest infrastructure and service agreements in the history of the technology industry.
A primary pillar of the expanded collaboration is the integration of Amazon’s custom-designed AI silicon. Anthropic has committed to securing up to 5 gigawatts of compute capacity powered by Amazon’s Trainium and Inferentia chips. This hardware commitment is designed to facilitate the development of future iterations of Anthropic’s Claude models. By utilizing Amazon’s proprietary chips, Anthropic aims to optimize the performance and cost-efficiency of its large-scale training runs, moving away from a reliance on third-party GPU providers.
Matt Garman, Chief Executive Officer of AWS, described the agreement as a foundational step in the evolution of generative AI infrastructure. Garman noted that the scale of the $100 billion commitment reflects the intensifying demand for specialized cloud environments that can support the massive computational and energy requirements of frontier AI research. The 5-gigawatt power allocation is specifically intended to support the high-density data centers required for next-generation model training.
Dario Amodei, Chief Executive Officer of Anthropic, stated that the long-term nature of the partnership provides the structural stability required for multi-year research initiatives. Amodei highlighted that the collaboration involves deep technical integration, with engineering teams from both companies working to refine the software stack for Trainium. This effort is intended to ensure that Anthropic’s models are natively optimized for AWS hardware, potentially reducing the time required to bring new AI capabilities to the enterprise market.
The agreement also expands the availability of Anthropic’s models through Amazon Bedrock, a fully managed service that allows AWS customers to build and scale generative AI applications. Amazon confirmed that it will continue to invest in the underlying infrastructure, including power generation and cooling technologies, to meet the 5-gigawatt requirement. This announcement follows a series of previous investments by Amazon in Anthropic, including a $4 billion minority investment finalized in early 2024. The new $100 billion commitment is focused on operational expenditure and infrastructure utilization rather than additional equity funding.