Anthropic today announced a significant milestone in its growth trajectory, securing a $4 billion investment from Amazon, which brings Amazon’s total investment in the company to $8 billion. As part of this expanded collaboration, Anthropic has designated AWS as its primary cloud and training partner for developing its advanced AI systems.
The partnership centers on the joint development of AWS Trainium hardware and software. Anthropic is collaborating with Annapurna Labs at AWS to optimize future Trainium accelerators, writing low-level kernels for direct silicon interface and contributing to the AWS Neuron software stack. This integrated hardware-software approach aims to maximize computational efficiency for training advanced foundation models.
Claude, Anthropic’s AI model, has become an essential infrastructure for numerous organizations through Amazon Bedrock. Notable implementations include Pfizer’s use of Claude to accelerate medical research and reduce operational costs, Intuit’s deployment for explaining tax calculations, Perplexity’s enhanced search engine performance, and the European Parliament’s Archibot system, which has reduced research time by 80% while processing 2.1 million documents.
The partnership enables organizations to fine-tune Claude models, including Claude 3 Haiku, within Amazon Bedrock while maintaining data security within AWS. Government customers can access Claude through both Amazon Bedrock in AWS GovCloud (US) and Amazon SageMaker in Secret and Top Secret Cloud Regions, ensuring compliance with stringent security requirements.
Interested users can access Claude through Amazon Bedrock by visiting aws.amazon.com/bedrock/claude/.
