SAN FRANCISCO — In a significant development, Anthropic has expanded its collaboration with Amazon Web Services (AWS) to further advance AI systems and solutions. As part of the enhanced partnership, AWS has made a substantial $4 billion investment in Anthropic, bringing its total investment to $8 billion, cementing its role as a key strategic cloud and training partner.

Strengthening AI Hardware with AWS Trainium

A central focus of this expanded partnership is the collaborative work on AWS’s Trainium hardware. Anthropic is working closely with AWS’s Annapurna Labs to refine future generations of Trainium accelerators, aiming to optimize machine learning hardware. This collaboration involves the development of low-level kernels that enable direct interaction with Trainium’s silicon and contributions to the AWS Neuron software stack. With this combined effort, Anthropic’s engineering team, in collaboration with Annapurna’s chip design team, is striving to extract the highest computational efficiency from the hardware, which will be pivotal for training Anthropic’s advanced foundation models.

This deep integration of hardware and software, coupled with Trainium’s price-performance benefits and scalability, enables Anthropic to optimize its model training across every aspect—from the silicon to the full software stack.

Claude Models Becoming Core Infrastructure for Global Enterprises

Through the AWS-powered Amazon Bedrock, Anthropic’s Claude models have been adopted as core infrastructure for numerous companies looking for reliable, scalable AI solutions. For instance, pharmaceutical giant Pfizer is leveraging Claude to speed up the development of critical medicines, saving millions in operational costs. Financial tech company Intuit uses Claude to assist millions of users with complex tax calculations, especially during tax season. Additionally, AI-powered search engine Perplexity benefits from Claude’s capabilities, offering faster and more accurate search results. The European Parliament also uses Claude to power ‘Archibot,’ which has improved research efficiency by making millions of official documents easily searchable in multiple languages, reducing research time by 80%.

Secure and Customizable AI Solutions

Claude models, integrated into Amazon Bedrock, offer a powerful and secure solution for enterprises. Organizations can now fine-tune Claude models, including Claude 3 Haiku, to fit specific business needs and improve task accuracy. Importantly, customer data remains secure within AWS, upholding stringent privacy and safety standards. With AWS’s robust security features, organizations across various sectors can deploy AI solutions that meet regulatory and compliance requirements. For government clients, Claude is also available through Amazon Bedrock in AWS GovCloud (US), as well as highly controlled environments like AWS Secret and Top Secret Cloud Regions.

Paving the Way for Next-Generation AI Development

In partnership with AWS, Anthropic is setting the stage for the next wave of AI research and development. With Anthropic’s frontier AI expertise and AWS’s cutting-edge infrastructure, the collaboration aims to build a secure, enterprise-grade platform that gives organizations access to the latest advancements in AI technology.

If you’re interested in using Claude in Amazon Bedrock, you can get started at aws.amazon.com/bedrock/claude/.