Since 2023, Amazon and Anthropic have partnered to facilitate the adoption of generative artificial intelligence across various industries. Currently, over 100,000 customers utilize Anthropic’s Claude models on Amazon Web Services (AWS), positioning Claude as a leading model family on Amazon Bedrock.
To build on this existing framework, the two companies have announced a significant deepening of their collaboration, marked by substantial financial commitments, new infrastructure deployments, and expanded platform integrations.
Unprecedented Financial and Infrastructure Commitments
The expanded partnership involves large-scale, long-term financial agreements from both organizations aimed at scaling AI research, development, and deployment.
- Anthropic’s Infrastructure Spend: Anthropic has committed to spending over $100 billion on AWS technologies over the next ten years. This investment will fund the use of tens of millions of Graviton CPU cores and multiple generations of Amazon’s custom AI silicon, including Trainium2, Trainium3, and Trainium4.
- Power and Capacity Allocation: To support the training and operation of advanced AI models, Anthropic will secure up to 5 gigawatts (GW) of capacity. This includes significant Trainium3 capacity slated to come online this year, alongside an expansion of international inference capabilities in Asia and Europe.
- Amazon’s Capital Investment: Amazon will immediately invest $5 billion into Anthropic. Furthermore, Amazon has committed up to an additional $20 billion in the future, contingent upon certain commercial milestones. These figures are in addition to the $8 billion Amazon has previously invested in the AI research firm.
Streamlined Access via AWS
To simplify enterprise adoption, the partnership introduces the Claude Platform on AWS. This integration allows AWS customers to access the full Anthropic-native Claude console directly through their existing AWS accounts.
By eliminating the need to manage separate credentials, contracts, or billing relationships, users can utilize their pre-existing AWS access controls and monitoring systems. This offers customers the flexibility to choose between the native Claude Platform on AWS or utilizing Claude through Amazon Bedrock based on their specific operational needs.
Key Pillars of the Ongoing Collaboration
The expanded relationship continues to center on infrastructure development, custom silicon engineering, and broad enterprise deployment.
- Custom Silicon Development: Anthropic utilizes AWS Trainium chips for the building, training, and deployment of its models. The company works directly with Annapurna Labs to provide feedback from Claude training workloads, which informs the architectural decisions and low-level optimization of future Trainium chip designs.
- Project Rainier: The two companies collaborated to launch Project Rainier, a massive AI compute cluster featuring nearly half a million Trainium2 chips. Anthropic is currently utilizing this cluster to train and deploy current Claude models globally, as well as to develop future iterations. The project serves as a foundational template for deploying the large-scale computational power required for advanced AI research in fields such as medicine and climate science.
- Enterprise Adoption and Performance: Over 100,000 organizations currently run Claude models (including Opus, Sonnet, and Haiku) on Amazon Bedrock. Organizations are utilizing the technology to streamline operations and reduce costs:
- Lyft: By integrating Claude via Amazon Bedrock for its customer care assistant, Lyft reduced average customer service resolution times by 87%, handling thousands of daily requests and routing complex issues to human specialists.
- Pfizer: Utilizing Bedrock and Claude, Pfizer developed a voice-command chatbot to search through roughly 20,000 documents per drug development project. This implementation saves scientists an estimated 16,000 search hours annually and has reduced infrastructure costs by 55%.
Looking Forward
Moving forward, Anthropic will maintain AWS as its primary cloud and training provider for mission-critical workloads. In turn, Amazon’s internal developers and engineers will continue utilizing Claude models to enhance customer experiences across Amazon’s broad portfolio of businesses.